Wikipedia talk:Manual of Style/Accessibility/Alternative text for images/Archive 5

Archive 1

←

Archive 3

Archive 4

Archive 5

Archive 6

Rewrite of lead

I've rewritten the lead as a placeholder until we get the issues resolved. This takes into account Jared Smith's advice but without going all the way down that road. I'll also be adding something to this effect to the MoS in place of the current text.

Alternative text (alt text) is text that replaces an image for text-only users. An example of an image with a caption and separate alt text is [[File:Tall red vase.jpg|thumb|The vase that was thrown at the president's head is now in the National Museum of American History.|alt=A tall red vase]]. Alt text is aimed at visually impaired readers who use screen readers, which translate the contents of pages into speech; at readers who use browsers with images turned off for whatever reason; and at indexing bots.

Non-decorative images should have an alt attribute in the |alt= parameter. Without an alt attribute, screen readers may read out the file name or repeat the caption. Filling in an alt attribute need not involve adding a separate description of the image: where the caption or nearby text is sufficiently descriptive or evocative, the alt attribute can be |alt=see caption or |alt=see adjacent text. Where additional alt text is deemed appropriate, it must comply with the content policies and should offer only descriptions that can be verified by any reader without specialist knowledge. Its length will depend on the context, but it should be as succinct as possible so that readers are not burdened by unnecessary words—a good rule of thumb is that seven words is often enough, while more detailed images may require 20 words or more. Images that are purely decorative, such as images in templates, should be given a null or empty alt attribute such as |alt=|link=.

Does anyone know what we need to write for regular (non-decorative) images to produce an empty alt attribute? Jared Smith said alt="", but someone above said that wouldn't work. What does work? SlimVirgin ^{TALK contribs} 09:19, 20 March 2010 (UTC)

I've made a pedantic change to the lead, since screen readers also output text to refreshable Braille displays. The code alt="" with two quotes produces an alt text with two quotes. The code "alt=" (without the quotes) is exactly the same as not specifying the alt text at all. The problem is that alt text is needed when an image contains a link, but some images don't really need alt text because the captions are informative enough. Graham87 11:16, 20 March 2010 (UTC)

Thanks, Graham. We need an alt attribute always (according to Jared Smith) to stop screen readers reading out file names etc., even if it's only an empty attribute. So can you tell me: is "alt=" an empty attribute or an absent one? SlimVirgin ^{TALK contribs} 11:34, 20 March 2010 (UTC)

"Alt=" is the same as an absent alt attribute. Graham87 11:37, 20 March 2010 (UTC)

Okay, got it. So do you know what we would have to write to create an empty or null alt attribute, similiar to link= for decorative images? What I'm getting at is if editors want to write nothing (not even "see caption"), but don't want the screen readers to be reading out file names, what should they fill in for the alt attribute? SlimVirgin ^{TALK contribs} 11:40, 20 March 2010 (UTC)

I don't think that's possible at the moment, unfortunately. Graham87 11:57, 20 March 2010 (UTC)

What does the screen reader do if we simply add alt=? Because I'm thinking that alt= is the same as alt=|link=, and we do use that. Sorry if I'm being technically dense. I have no technical knowledge at all, and can only grasp this conceptually, and conceptually it's currently making no sense to me. SlimVirgin ^{TALK contribs} 15:30, 20 March 2010 (UTC)

It should be possible to test this yourself in a screenreader (I assume there are free ones available). One point I think needs to be made is that all images-acting-as-links (which I think is all images displayed in the mediawiki framework in whatever form - certainly all images I've ever seen in Wikipedia articles I've been able to click on them) need to say "more information on this image can be found by activating this link". That would be a good start at a default. Technically, the file page displays either the highest resolution image available, or an image at a suitable resolution for the screen with the image then being a link to the highest resolution image available (with some exception for very large images, I think). The alt text would then be a page-specific (rather than image-specific) attribute that explains the function and appearance of the image in the context of the article the image has been placed in. In other words, the "it is a link to further information" (i.e. a link to the file information page) is something all images have in common, but the alt-text for an image is part of the page (not the image) and helps to integrate the image into the page it is being used on (i.e. this image is being used here for this reason, and [where it helps] looks like this, and don't forget to read the caption for more information". Does that all make sense? In other words, the same image used in different pages will have different alt-text, but it would be good to have a universal image attribute that tells screenreaders that the images are also a link to further information, and this would appear on all images where-ever they were used (possibly this already exists and there is just a confusion in terminology). You have to remember that most websites don't have this feature where you click on an image to get more information (though some increasingly do). Carcharoth (talk) 17:09, 20 March 2010 (UTC)

I did download a free screenreader but I couldn't see how to adjust its preferences and it was talking all the time so I had to uninstall it. It came with a default of skipping over links, including images, and didn't read out any alt text. Could you write out an example of what you mean by referring people to the image page? So for example if we had "alt=a red vase", where we would write "please click on the link for more information"? SlimVirgin ^{TALK contribs} 17:19, 20 March 2010 (UTC)

At the moment, in the "alt=" text. This is not ideal, however. I wish that there was a way to decouple the "alt=" text from the link to the image description page. Graham87 08:21, 21 March 2010 (UTC)

Also, if you add "alt=" to image markup, the screen reader acts as if you never specified the alt attribute. Graham87 08:24, 21 March 2010 (UTC)

For specifying no alt text the HTML syntax is alt="" or alt=" " (note the space), since MediaWiki will display the default alt text if it is empty or null we need to use |alt= (HTML escape code for a space). However this was rejected at the time since one screen reader (Orca) would read the link title attribute (the tooltip) instead. But since the default alt text update, this attribute is omitted entirely (and thereby breaking our linking convention). — Dispenser 22:30, 21 March 2010 (UTC)

Request for clarification

Myself and User Tuxedo junction are unsure as to the descriptive alt of the picture in this infobox of this article Hermaphrodite, issue is about sexuality, man, woman? Please comment, thanks. Off2riorob (talk) 19:37, 21 March 2010 (UTC)

You could write "alt=see caption" because the caption is quite descriptive, or if you want to add a separate description, you could talk in terms of a reclining figure without trying to specify the sex. SlimVirgin ^{TALK contribs} 19:55, 21 March 2010 (UTC)

thanks, alt=see caption is a good idea and I also support as it is unclear by viewing the picture that it is better in this case to avoid attempting to identify sexual definition.

• Marble statue of a person of unclear sexual definition, lying on a bed, back to the viewer, the persons legs are entangled in cloth. Off2riorob (talk) 21:24, 21 March 2010 (UTC)

I'm wondering if our "see caption" kludge for cases where MediaWiki prevents us saying nothing, is really a wise choice of words. For starters, the words will prefix the caption in the spoken stream of words. Secondly, perhaps "see" isn't the best choice. Thirdly, once this MediaWiki limitation is ever fixed, we're going to want a standard piece of text to search and destroy. I'll chuck "thumb" into the pot as a random suggestion for this situation. Colin°^Talk 22:54, 21 March 2010 (UTC)

Refer to caption would make more sense. I don't see it as a problem that it would read that out first. You'd hear "refer to caption," then the caption would be read out, which would seem to make sense. SlimVirgin ^{TALK contribs} 22:57, 21 March 2010 (UTC)

Violating POV

Vanilla Ice based the song's lyrics upon the South Florida area in which he lived.

I have gotten into an edit war as to whether an alt text should include the words "a neat goatee". To me, that violates NPOV. I need ideas. Woogee (talk) 21:00, 21 March 2010 (UTC)

The page in question is Ice Ice Baby. I welcome all take a look and provide an opinion All advice and suggestion is welcomed by me, as I am just learning how to do alt text. Tuxedo junction (talk) 21:09, 21 March 2010 (UTC)

I'd say it's original research, I suggest removing anything from the alt text that is challenged. Regards, SunCreator ^(talk) 22:45, 21 March 2010 (UTC)

It's a goatee. Whether it is neat is probably just opinion and I'm sure you guys don't want to edit war over that. Colin°^Talk 22:48, 21 March 2010 (UTC)

How can you keep the alt text message on?

Some of my articles have now been edited with alt text. In general, I think it is a good idea to have alt text messages for those who cannot see the images but I have difficulty in reading them myself because they disappear after a few seconds. Can anyone tell me whether there is a simple way of keeping the alt text message open long enough for it to be read in full? At the moment, the only way I can read the messages properly is to go into edit mode. -- Ipigott (talk) 11:20, 26 March 2010 (UTC)

The actual method is dependent on the specific browser being used, but all browsers have a means of turning off images. Find the options or preferences in your web browser, and you'll locate some sort of check box, or whatever, which can disable the display of images and other multimedia. If you mention the exact brand and version of browser that you normally use, people could provide specific instructions.
— V = IR (Talk • Contribs) 18:36, 26 March 2010 (UTC)

Consolidation?

Please note that this page has been nominated to be consolidated with the primary Manual of Style page. Please join the discussion at the MOS talk page in order to discus the possibility of merging this page with the MOS. Thank you.
— V = IR (Talk • Contribs) 14:39, 27 March 2010 (UTC)

The very first example seems bizarre

First let me say that I commend the good intentions behind all of this.

Though I've been aware for some time that FAs have alternative text, it's only today that I've first read the project page. It amazes me. Here's its very first example:

Pavel Bure with the Vancouver Canucks|alt=A hockey player on the ice before spectators. He wears a white jersey with a big "C".

I'm not blind and I have not made any serious study of web usability for the blind (or anybody else). So I may be very wrong when I speculate that neither half of this -- neither what's before the pipe nor what's after it -- would be of any interest to a blind person hearing this page. Indeed, if this were my page and I were writing the XHTML directly rather than going through the Mediawiki preprocessor, I might do away with the regular caption and instead have the tag read

<img src="whatever.jpeg" class="whatever" title="Pavel Bure with the Vancouver Canucks" alt="" />

thereby clearing the whole thing, caption and all, out of the way of any blind hearer.

I realize both that I haven't thought about the partially sighted and that I could be very, very wrong in my speculation about the blind.

(NB I am not expressing my disapproval of alternative text in general. The project page also has examples that I imagine would be very helpful.) -- Hoary (talk) 15:38, 26 March 2010 (UTC)

Well for me at least, as a totally blind person, I'm interested to know that the image is there and roughly what it represents. It's late here, so I can't really articulate *why* I'd want to know that the image is there, but I just do. Graham87 16:01, 26 March 2010 (UTC)

To the sighted, imagery speaks to the reader. That an image appears alongside certain text provides some context, and information. Imparting at least a general sense of what the image contains, to those who can't actually see it (due to either technical or physical reasons), is therefore important.
— V = IR (Talk • Contribs) 18:43, 26 March 2010 (UTC)

I'd quibble. That an image appears alongside certain text should make the combination more informative or easier to understand or something. But very often indeed in Wikipedia it does not. The picture, mentioned below, of Napoleon that illustrates some episode of a TV reality show is merely a particularly blatant example of this. A lot of WP editors seem to think of their readers as drowsy 12-year-olds in a class whose textbook needs to be "spiced up" to provide a minor impediment to falling asleep. Or perhaps they're just trying to anticipate what the FAC reviewers will demand: at least one illustration, almost any illustration, for each major section of the article, in order to keep the article "interesting". -- Hoary (talk) 00:24, 27 March 2010 (UTC)

Graham, from your own experience, how long would you suggest an alt text should be? The general consensus seem to be somewhere between 7 and 20 words. Would you agree with that? Malleus Fatuorum 20:36, 26 March 2010 (UTC)

I think a maximum of 7-20 words is what is suggested; according to some, a word or two might often suffice. But I too would be interested in getting Graham's opinions this.

And Graham, if you have time, I would also love to know your thoughts on a couple of other questions:

• Does context influence how long or how detailed the alt-text should be? It seems to me that if you are reading an article about Napoleon, someone might like details of what he looks like, while in another article where the picture is tangential, mostly decorative and not adding any specific content to the article (such as in this picture of Napoleon used to illustrate an Amazing Race article) it might be best just to say "portrait of Napoleon", without any detail at all. Any thoughts about this?
• As a kind of related question, do you think that as an education/information website, Wikipedia at times may need somewhat different alt text objectives/lengths than other websites, in that the pictures are may be an important part of the information provided? --Slp1 (talk) 21:17, 26 March 2010 (UTC)

• Yes, the portrait of Napoleon shouldn't be used in that way in anything that I'd call an encyclopedia article. On the other hand writing such excited prose, and decorating it with uninformative graphics, clearly keeps a number of contributors happy. Since the world of "reality TV shows" is as alien to me as is that of Pokemon or dog-breeding, I'd let it be. -- Hoary (talk) 00:24, 27 March 2010 (UTC)

There is an enormous amount of work to be done in making worthwhile graphics worthwhile for all. As an example, try this section of "Education in Finland". (If you're sighted, that is. If you're blind, please don't bother -- for now, at least.) -- Hoary (talk) 00:24, 27 March 2010 (UTC)

I would raise the word limit to 40, which is just over a line of text with JAWS, where a line takes up 150 characters. I would not make this a hard and fast rule, however. Twenty words seems inadequate to describe charts, maps and the like. However alt text should be concise. Yes, the detail in alt text should depend on the context of the image. "Portrait of Napoleon" would be a good alt text in the Amazing Race example. However, because Wikipedia is an educational website, and because of its unique situation in relation to image attribution, its requirements for alt text are somewhat different from most sites. Yes, the example at education in Finland is terrible; precisely what is the purpose of the chart? Is it just a visual aid, or does it convey information that cannot be found in the text? Graham87 05:50, 27 March 2010 (UTC)

Twenty words isn't a limit, Graham. It says "A good rule of thumb is that seven words is often enough, while more detailed images may require 20 words or more." So 40 words are fine, so long as 40 words are needed. But if we make 40 words the rule of thumb, people will add 40 words, needed or not. SlimVirgin ^{TALK contribs} 16:41, 27 March 2010 (UTC)

I'm beginning to feel that we're getting somewhere now. The distinction does indeed seem to be between "alternative to the image" and "description of the image", with the complication of the caption thrown in. Perhaps we need to think more carefully about how the caption and the alt text ought to fit together. Malleus Fatuorum 06:07, 27 March 2010 (UTC)

Graham87, I'm puzzled by your suggestion that the alt text "Portrait of Napoleon" would be good in that Amazing Race article. The portrait of Napoleon seems to me to add nothing whatever to the article. Arguably it adds visual appeal, but to me any such appeal is so obviously bogus as to make the image laughable -- rather like a doorman at a hotel whose uniform is adorned with epaulettes and medals. I really don't understand how blind people are helped in any way by knowing that there is a portrait of Napoleon there. I'd be inclined to delete the image, but I don't want to spend my time on articles about the murky world of reality TV shows. If it does stay I'd tend to think that it shouldn't pointlessly complicate the business of aural browsing, and I'd therefore give it a null alt text (nothing to separate the pair of double quotation marks). As for the chart in "education in Finland", the intention seems honorable. Elsewhere, the article explains (via plain text) each component (e.g. masters' course at university) of its education and something about the interrelationships among these components. What this chart tries to do is to show the interrelationships more understandably as a flow chart. So far so good. However, at least one pair of modules is connected by a two-way arrow and also by an arrow in one direction; and certain modules are very close to others but neither adjoin them nor are linked by arrows. So all in all it looks like the second draft of something that needed to go all the way to a fifth draft before being added. The audio browser is, I think, better advised to read the text of the article. -- Hoary (talk) 15:17, 27 March 2010 (UTC)

Hoary, have you considered that, as an editor on Wikipedia, Graham87 might also like to share your amusement that someone has gratuitously decorated that article with a "Portrait of Napoleon", and be able to contribute to any talk-page discussion that might ensue if someone suggested its deletion. The issue with that picture goes beyond alt-text and us discussing how best to write alt-text for it is akin to polishing a turd. -- Colin°^Talk 16:38, 27 March 2010 (UTC)

I think we do need to discuss it, because many articles are illustrated by fairly tangential pictures. Take a look at Learned Hand for example. (No alt-text to be found, sorry Graham). Do the picture of a NY skyline, Macarthy, James, Roosevelt, etc help understand the subject any more than good old Napoleon on Amazing Race? Personally, I don't think so. Once again, I would say that "portrait of x" or "refer to caption" is likely desirable. --Slp1 (talk) 16:48, 27 March 2010 (UTC)

I agree with Malleus about the importance of making sure the alt text and caption work together, and I would add the adjacent text to that, and to bear in mind that the alt text is often read out before the caption. That means, depending on context, that the reader may not have a clue what image is being described, in which case it might be helpful if the alt text were to tell them that it's a picture of Napoleon, rather than "man with his hand in his waistcoat standing next to 18th-century furniture." SlimVirgin ^{TALK contribs} 16:35, 27 March 2010 (UTC)

As far as I can tell, the alt text is always read out before the caption in a screen reader. Is there an editor that knows this to be false? Tuxedo junction (talk) 19:54, 30 March 2010 (UTC)

Right and left?

Some alt text describes that such and such is to the left, and such and such in the upper right, while such and such is further to the left, etc. Some of this is quite complicated to follow. I am wondering if such wording should be discouraged. Thoughts? Tuxedo junction (talk) 21:22, 30 March 2010 (UTC)

I have always argued that such descriptions should be avoided even in the article body itself, as the precise rendering of a page is not under the editor's control. In the case of alt text, I think that all such descriptions are a sure sign that the alt text has become too detailed, acting as a description of the image rather than an alternative to it. Malleus Fatuorum 21:35, 30 March 2010 (UTC)C)

I agree that it can be complicated to follow, and must be difficult to listen to. SlimVirgin ^{talk contribs} 21:38, 30 March 2010 (UTC)

We currently have a problem that the content of this page disagrees with the message given to anyone on the talk page. We need to settle this description=wrong issue as it permeates the page. There's little point maintaining the current text if every reply posted here is totally at odds with it.

Are there cases where a short description is required? For example, the design of a flag that is red with a star in the top left. If we decide that is not something we describe in alt text (but instead merely say "Flag of XXX") should we insist that the information conveyed by the image is instead contained in the body text. If so, does that belong here or some other accessibility guideline? Should we burden our sighted readers with information on a person's appearance in a bibliographic article, when that appearance (they are white, they have brown hair, they wear glasses and have a beard) is quite obvious from the photo?

We could do with some FAQ guidance on colour (nothing intrinsically wrong with it) or spacial references like left and right (again, nothing wrong with those). While I agree with Malleus that these might be "signs" that the wrong approach is taken, our guidance should take care not to blame the wrong thing. There's an ignorance that blind people have no concept of colour (or of the meanings society gives to colour) or of up, down, left and right. Colin°^Talk 22:44, 30 March 2010 (UTC)

There's no "ignorance"; everyone accepts that many people lose their sight in later years, and so will have an understanding of "turquoise", for instance. But it has never been clear who the audience is for alt text, or what its purpose is, and it still isn't. It hasn't even been agreed whether alt text is an alternative to the image, a description of the image, or something in between forced on us by the way the wikimedia software works. Some basic decisions need to be taken. Malleus Fatuorum 22:52, 30 March 2010 (UTC)

There is guidance somewhere about colour. (I forget where.) It says don't use colour if the description depends on it, like "The map shows the rich countries in blue and the even richer countries in red." The blind editor Graham, who has always been blind, said that, although he has never seen colours, he has learned to associate certain colours with certain things, like a "blue sky" means good weather. But it seems to me that mostly it is better to not use colours to describe. How do we "know" what various editors have learned to associate with various colours? Tuxedo junction (talk) 22:59, 30 March 2010 (UTC)

Clearly, if colour is being used to colour countries on a map, then describing which "colour" each country is is pointless. But I disagree with Malleus about there being no "ignorance" here. Time and again in this discussion, the issue of "should we mention colour" is raised. If (and I agree it is and if and a when decision yet to be firmed up) we describe something real (as opposed to "colours" of countries on a map) then don't assume the blind reader is any less able to comprehend than a sighted person. Some may be but the range of comprehension is likely to be wide and just the same as the range for other senses and experiences one gets through age or circumstances. On the other hand, there are visual aspects that aren't important to describe and identifying that is crucial. We should get that point overwith early in the guideline.

We do all agree it is an "alternative" to the image. Yes there are some compromises due to the way the software works and editorial choices over whether some things appear in alt, in captions or in body text. There are cases in an encyclopedia where the picture isn't decorative and where it isn't sufficient that the only "alternative" to the image is to simply say "Flag of Denmark" or "Portrait of Napoleon". Those pictures are supplying lots of information about what the flag looks like or what Napoleon looked like. The question for any accessibility guideline is whether, how and where to make up for the loss of the image. Sometimes the caption identifies the image, sometimes it describes it, and sometimes it contains a little soundbite from the article which has nothing to do with the image. We don't have the longdesc attribute and I don't think the WikiMedia image description page is a friendly place to direct a screen reader for a long description. We have all moved away from thinking that alt-text is the place to describe the image, should a description of the image or the subject of the image be required. However, moving it into body text isn't without problems. One can describe a particular apple's appearance based on the picture as a primary source. To say that's what all Cox's Pipins look like would require another source... Colin°^Talk 08:10, 31 March 2010 (UTC)

I agree that we have a problem because we can't decide whether to describe the image or in some other way replace it. But if we're going to describe, I see no problem using colour adjectives. Alt text is for anyone who can't see images, including people on cell phones with images switched off. They know what colour is, as do people who lost their sight later in life, and that audience is likely to account for the bulk of ALT text users. And it harms no one to have a word included that they can't understand conceptually, if it's informing others. We can't produce a one-size-fits-all solution here.

We also have to leave something down to editorial judgment. For the same reason I'd be very opposed to mandating that images must be described somewhere in the text. SlimVirgin ^{talk contribs} 17:58, 31 March 2010 (UTC)

... which I think underlines my comment above; the audience for alt text is ill-defined, and different audiences will have different requirements. Malleus Fatuorum 18:03, 31 March 2010 (UTC)

Which part of "Alt text is aimed at visually impaired readers who use screen readers, which translate the contents of pages into speech or Braille; at readers who use browsers with images turned off; and at indexing bots" do you think is ill-defined? Colin°^Talk 18:35, 31 March 2010 (UTC)

All of it. Malleus Fatuorum 22:39, 31 March 2010 (UTC)

That's not a helpful response. The sentence clearly categorises the audiences. Defining "who" the audience is is not the problem here. Perhaps you feel their needs or our commitment to each audience is ill-defined? Or some other aspect of the audience groups? Colin°^Talk 07:38, 1 April 2010 (UTC)

Let me be more specific then. Which of those three audiences are we writing alt text for, as they clearly have different needs? Your first audience, the visually impared, also needs a bit more thought; those who have never seen, those who have lost their sight, or both? Malleus Fatuorum 23:42, 2 April 2010 (UTC)

All of them and both. :-). All three will make use of our alt text. I've given my opinion on the importance of each of those three to Wikipedia below. Where the needs of those audience groups conflict in what we write, the priority should surely be to satisfy 1 before 2 before 3. And in terms of weight given by the guideline to editors spending effort meeting those audiences, it should also be 1/2/3. If other editors disagree on those priorities, then we can come to a different consensus. Wrt the last point, it is a bit like our guideline to "write for the general reader", who doesn't exist. This paper makes it clear different blind users want different things, which shouldn't come as a surprise as the same is true of body text and sighted readers. However, the authors make the point that in trying to satisfy the majority of the readers we should err on the side of too much detail than to little, while bearing in mind the need to keep it short. In reading that paper, I don't get the feeling that the difference between the two visually impaired groups you mention is the biggest cause for variation in opinion on what alt-text should contain. Colin°^Talk 08:09, 3 April 2010 (UTC)

Audiences: which matter?

We have three audiences for alt text. The visually impaired, those with images turned off and indexing bots. The first of these is our primary concern IMO and there is pressure to meet their needs both from regulatory agencies (government guidelines, W3C guidelines) and morally. The second is an editorial/community judement of no more concern than whether to write a Pokemon article, take and upload a picture, or support a client O/S or browser. Editor consensus to serve that audience better might appear and be enshrined in a guideline recommendation, but I don't see any groundswell yet. If editors choose to aid the second reader, we should offer guidance on how best to do so, but not yet request (to any degree) that they do. The latter (bots) is of zero concern to editors on Wikipedia IMO. If Google Image Search can't work out what our image is based on either the caption, the article title, or the image's own page here or at Commons, then Google need to improve their algorithms, not request we write for them. If they find our alt text useful then that's a bonus but we certainly shouldn't request our editors write text aimed at bots. I suggest this guideline be weighted according to the above, giving most space to the needs of visually impaired readers and a short amount of space to the issue of some readers having images turned off. Colin°^Talk 07:38, 1 April 2010 (UTC)

Is there a problem with alt text being served?

I've noticed during the last couple of days that the alt text no longer appears when the mouse hovers over an image. This is occurring both in IE 8.x and Firefox 3.6.x. The alt text in infoboxes still works, but not on images within articles. Just curious and am unsure where else to look for why this suddenly is happening. • Astynax ^talk 18:42, 3 April 2010 (UTC)

AFAIK, Firefox has never displayed alt text as a tooltip, per HTML standards. IE used to display alt text as a tooltip. In recent experiments I've conducted, it now appears to only display link "title" attributes as a tooltip, which is what the HTML standards say. Most of our images are links (to the image page). MediaWiki doesn't provide an explicit way (that I know) to set the link title attribute, but it does get set to the "caption text" (i.e, the unnamed parameter in a file or image wiki element) when the image format has no visible caption (a plain image, not a frame or a thumb). If you can say which pages/images the tooltip works on, then you/we can look at the HTML to see whether it is the link (the <a> element) title attribute or the image alt attribute.

I've been experimenting with the image alt/caption markup in a sandbox here. Colin°^Talk 18:58, 3 April 2010 (UTC)

It might be just an update to IE8, which has an option to expand the alt text for images. On pages with infobox images (see Christian Conventions infobox at the top as an example), the image from the infobox does show a tooltip with the alt text in IE8, while the other images in the article, which all have alt text, show nothing. A few days ago, IE8 with the option to expand alt text enabled, was showing text for all images. You are probably correct that it is a browser issue, just confusing as to why only the infobox images were showing the alt text in a tooltip. • Astynax ^talk 19:16, 3 April 2010 (UTC)

There has been a change for IE8 see here. In the example you give, looking at the source for the Infobox, the picture is a plain picture with no frame or thumb. There is therefore no caption (the "caption" is generated by another row in the Infobox table). This follows the second example in my experiment page, which has ([[File:Green check.png|Alt/Title text]]). The "Alt/Title text" is used for both the link title attribute and the img alt attribute. IE8 displays the title. Colin°^Talk 19:26, 3 April 2010 (UTC)

Thank you for the explanation! • Astynax ^talk 06:44, 4 April 2010 (UTC)

Overhaul

It is clear that the current guideline text no longer has consensus support; it isn't an official guideline any more. The guideline is far too long, with too many lengthy examples, and spends too long on rare situations. The examples don't need the mini-essay "flawed alt text". In addition, there are contentious aspects of writing good alt text that (a) haven't been fully resolved and (b) may ultimately reflect editor preference or the preferences of one reader.

I think we should split the guideline in two. Have a short core guideline that covers the essentials both conceptually and in practice, with relatively few examples. The intention is that this would later be re-adopted into the MOS. The second page should contain more examples with longer rationale and also cover the unusual image types that appear on few articles. That page would have no official status as a WP Guideline and its talk page could become a place where editors ask for and offer advice.

Would anyone mind if I took an axe to this page and worked on it directly. Or would it be better to work on a draft version first? Any thoughts on a name for the second example/how-to page?

Colin°^Talk 19:22, 2 April 2010 (UTC)

The problem is that we don't know what we're doing conceptually, and it can't be a guideline (alone or as part of MoS) until that's sorted out. Is alt text a replacement, a description, or both? I was hoping the expert who posted above would help out—he said he might when he had more time. Perhaps we should email him again. SlimVirgin ^{talk contribs} 23:37, 2 April 2010 (UTC)

I got a brief response from an expert yesterday. It is possible they may be able to find time to help next week. The problem is that we can't really use the current text as a starting point because none of us believe in it any more. It is also too long. I do believe there is enough basic detail and common ground to write something core. And a separate examples page would allow us to form our "what to write" guidance in a collaborative way and hopefully with input from an expert or blind editor. I'm not proposing this short core page become a guideline for quite some time -- until we've sorted ourselves out and got expert approval for it.

In response to your query: the important aspect from the W3C guidelines and others is that we provide a textual alternative to the (important) content and the function/purpose of an image. Sometimes, the image has no important content, has no function and has no purpose. Sometimes the only purpose of the image is as a stock photo of somebody/something and the identity of the image subject is the only important bit, not what the subject looked like or is doing. Sometimes the whole article is about the image. The degree to which one writes a "textual alternative/replacement" is an editorial judgement call. Secondary to this is where to put this textual alternative: the body text, the caption or the alt text attribute. The later is the main focus of this guideline but it always needs to be done in context and in relation to the other two.

Is any of that still an area of disagreement? If you guys still think we can't move on yet, then I'll work on a draft rather than edit this directly. But I don't think the current text is a good place to start discussion with an expert, as none of us believe in it. Colin°^Talk 07:55, 3 April 2010 (UTC)

I suppose I'm worried about removing all the work Eubulides put into it. I'm also worried about starting from scratch when we really don't know what we want to say. I do agree, though, that this page is problematic. Maybe you could start writing up something (on this page or a user page if you prefer), and we can see how it goes. SlimVirgin ^{talk contribs} 08:00, 3 April 2010 (UTC)

Oh I agree and there's a great deal of practical how-to help in here that is worth keeping. The question is: if we get an expert turn up here next week, do we want to start a discussion based on this text or something else? Should we create the "examples" page in order to have an expert discussion? I can create a user page sub-page. I'd rather not do it on a talk page as you just get bothered by signbot ever time you make an edit. Or is there another way? Colin°^Talk 08:19, 3 April 2010 (UTC)

It seems to me we need two separate pages. The first could simply be a revision of the current Wikipedia:Alternative text for images describing the general concept while the second should be very specifically directed at user guidance. It could, for example, be called something like Wikipedia:Alt-text user guidelines. I would further suggest that the user guidelines should be short and to the point with as little unnecessary commentary as possible. In order to get things moving, you could always start off with a sandbox page with a link from this talk page. -- Ipigott (talk) 12:19, 3 April 2010 (UTC)

Sandbox

I've made a start on an overhauled version of this guideline in my sandbox here. It has no equivalent of the "other examples" section but I think it is the better for it. A long guideline is off-putting. I think an extensive set of examples and solved-problems should go on another page, separate from the guideline which should focus on principles and core practicalities. The biggest difference from the current guideline, is the change of focus from the alt text attribute to the principle of providing a text alternative. When that text alternative is satisfied by the caption, then the alt text attribute becomes superfluous. As far as drawing advice from the many sources we found, I didn't get much beyond the W3C pages. So there's more advice we could include. I cover the problem with blank alt text not doing what we'd hope it would do, and provide a new solution I think is an improvement. It is a very rough draft so please don't slag off the prose. I've had to fix the related guideline Wikipedia:Extended image syntax#Alt text and caption because what it said differed substantially from reality. See my User:Colin/Alt sandbox page for where I experimented with different syntax.

Is this closer to something we can agree on? The lead paragraph is probably a little too purist and you should read the advice later before shooting it down. For example, "provides the same information as an image" needs to be qualified that we are talking about just the information the image provides that is essential for this article; images contain lots of information that we don't need. Thoughts? Feel free to edit it and use the talk page. Colin°^Talk 21:50, 5 April 2010 (UTC)

Compliance with content policies

I found "strange" alt text on Derek Jeter and was told the proper format for alt captions is to not use proper names. This seems absurd. Your example is this: Headshot of a young man with wavy hair, a high forehead, and deep-set eyes. That seems horrible. Why not, "Van Gogh, displayed as a young man with wavy hair..."

Just putting the text "a man" is pointless. Which man? Why is that man important? Why is that man on the article?? If you actually name the person in the picture, with a description, this is far more sensible than the current method. I can't imagine that people who actually use the "alt" captions find the current method acceptable. — Timneu22 · talk 18:17, 5 April 2010 (UTC)

Many have found the recommendations for alt text to be strange, as a cursory glance at this talk page will show. In the Van Gogh case you mention, the best alt text would probably now be considered to be "refer to caption", but the guideline is still being worked out. Malleus Fatuorum 18:32, 5 April 2010 (UTC)

Where is the discussion for these guidelines? I'd like to be part of that; this is the worst idea I've encountered on WP. — Timneu22 · talk 18:33, 5 April 2010 (UTC)

It's on this talk page, right above you. There's also been some dissension raised at WT:FAC. Malleus Fatuorum 18:36, 5 April 2010 (UTC)

I saw a little of that, but nothing too specific on "a man" (there's a little subject on that above, but it quickly jumps into other topics). I guess I'll just watch this page for a while. It seems the decision should be made here before being applied to MOS; I can't believe the captions were ever dumbed down to "a man"... just awful! — Timneu22 · talk 18:43, 5 April 2010 (UTC)

See above for a discussion on redrafting this guideline. Do you find the new draft helpful? Colin°^Talk 21:51, 5 April 2010 (UTC)

Where? Really, nothing specifically addresses this. Add a link to it. — Timneu22 · talk 22:16, 5 April 2010 (UTC)

I think your find the above comments in new light when you see currently WP:ALT is not a guideline. Regards, SunCreator ^(talk) 00:14, 6 April 2010 (UTC)

Proper names and WP:ALT#Proper_names seem incorrect to me also. Regards, SunCreator ^(talk) 00:14, 6 April 2010 (UTC)

It doesn't seem right to me either. I kept it in when I did the copy edit, but I didn't agree with it. SlimVirgin ^{talk contribs} 04:10, 6 April 2010 (UTC)

The diff for it's being added is here which derived from discussion here. Greta Garbo was discussed and it was thought many people wouldn't know what she looked like. I still think the name should be used. File:Greta Garbo in Meyers Blitz-Lexikon 1932.jpg You could have a head shoot of a 1000 brunettes ladies all very similar, some people could pick out and name who it is specifically, but to describe them all 'head shoot of a brunette' does not seem to inform and not what an Encyclopedia is about. Regards, SunCreator ^(talk) 00:16, 7 April 2010 (UTC)

I have removed the section what was WP:ALT#Proper names. I see no agreement for it but various disagreement with it. Regards, SunCreator ^(talk) 15:44, 9 April 2010 (UTC)

Proposal to replace current alt-text guidance

The guideline for alt-text on Wikipedia was recently demoted after consensus no longer supported the approach taken. Expert opinion suggested we should rethink. I have prepared a draft revision in a sandbox here that I believe is a better starting-point for us to begin afresh with a new approach to alt-text. It also corrects a number of technical errors. I propose the current text be replaced by this draft revision. Colin°^Talk 13:38, 9 April 2010 (UTC)

The only quibble that I have with the proposed text is that its advice about non-English text should match point 3 of Wikipedia:Accessibility#Text. Otherwise, it sounds good to me. Graham87 14:43, 9 April 2010 (UTC)

Re: this question on my talk page: the first point is correct, and the second point is not quite correctfor reasons that I don't understand. My version of JAWS, JAWS 8, just says "images/Magnify dash clip" without the ".png" extension. Newer versions of JAWS may be different, but I can't test that out because of problems with shared files. NVDA, the free Windows screen reader, says the "title=" attribute of the "magnify/clip" image and therefore it says "enlarge". Graham87 15:15, 9 April 2010 (UTC)

Thanks. So it looks like NVDA, in its attempt to say something about the link is using the link "title" when the alt is blank, but JAWS just gives up and uses the filename. It looks like JAWS is dropping the ".png" file extension, which I suppose is a reasonable thing to do as it isn't very interesting. In the case of normal thumbnail images, there is no title attribute for our link. What does NVDA do in that case? Does it say nothing when the alt text is blank or does it too speak out the filename? Colin°^Talk 15:28, 9 April 2010 (UTC)

What do you mean by a normal thumbnail image? I'll answer this tomorrow morning my time, as I'm off to bed now. :-) Graham87 15:38, 9 April 2010 (UTC)

Sorry, I just mean the actual image in a thumbnail, not the magnify icon. If it has blank alt text, does it read out the link filename in both screen readers? Colin°^Talk 21:10, 9 April 2010 (UTC)

Yes. Graham87 05:15, 10 April 2010 (UTC)

Thanks. I've tweaked the text to remove the file extension and also to mention that NVDA uses the title for the thumbnail-magnify icon. Colin°^Talk 10:24, 10 April 2010 (UTC)

A good starting point. Needs more description about the actual text that should be used. This is my chief complaint of alt text that I've seen: "a man with brown hair" or some nonsense is always used, instead of saying who the man is. — Timneu22 · talk 15:21, 9 April 2010 (UTC)

Colin's sandbox does NOT have the section on 'Proper names' which at the moment in our understanding is a step in the right direction. Regards, SunCreator ^(talk) 15:43, 9 April 2010 (UTC)

Agree. In fact, taking Timneu22 comment, unless the image was included simply to illustrate an article on male hair colouring, the wording "a man with brown hair" is unlikely to be a good choice. Alt text, if serving an identifying role, will probably include the person's proper name. But that role is usually performed by the caption, leaving us with blank alt text (or close to). An example should be added to the guidance. Colin°^Talk 16:47, 9 April 2010 (UTC)

Yes, please add that. My opinion comes from looking at the Derek Jeter article, and later others, where the alt text was just as I described. I agree that in these cases, alt text wasn't necessary as the caption seemed to do the job. — Timneu22 · talk 17:39, 9 April 2010 (UTC)

I've added a new example of "two men shaking hands". Colin°^Talk 10:24, 10 April 2010 (UTC)

• Support. I'm not sure how these RFCs are supposed to work, but this is just about the most sensible proposal for implementing alt text here in wikipedia that I've seen. We're somewhat hamstrung by the way that the wikimedia software works, but this proposal's emphasis on considering the alt text and the caption as complementary is spot-on. There's clearly no way that the fairly straightforward software change required to address this problem will be implemented in our lifetimes, or probably even that of our children's lifetimes, so we have to come to a compromise. Malleus Fatuorum 21:39, 9 April 2010 (UTC)

Revision ideas

I'm not sure about this: Alt text that contains just a nonsense filler such as "*" or "-" is annoying, and wordy text such as "refer to adjacent text" becomes tedious. The least-bad solution to this is to supply a word that is at least minimally useful to the reader. A suggestion is therefore to supply the single word "photograph", "painting", "sculpture", etc. When combined with a caption, the result "photograph Tony Blair meeting George Bush" or "painting Napoleon Bonaparte" should be acceptable.

For the most part I agree, but it seems tedious to add "photograph", or "painting", etc., throughout every page on WP. Isn't there a way that the text could be omitted altogether, and possibly some automatic text would be used? Is the automatic text even necessary? I guess I'm just thinking that there are hundreds of thousands of photos with self-explanatory captions; is it really necessary to add "alt=photograph"? My two cents. — Timneu22 · talk 17:52, 9 April 2010 (UTC)

Probably what is required is a Wikipedia skin designed for screen readers that omits the links to the graphics and the additional "enlarge" icon-link for thumbnails. The link is only needed for sighted users to see a larger version of the image, and to supply essential the attribute/licence details. But since the screen reader doesn't actually use the image, no licence conditions need to be upheld. If that was done, we could genuinely write "alt=|" (or miss it off in the certain cases that MediaWiki handles properly) and the screen reader would ignore the image, leaving just the caption. I should probably create a MediaWiki bug report to suggest this, but given the glacial speed such bugs are dealt with, it could be years before anyone does anything about it. As I said, I'm trying to find the least-bad solution. Nobody should have to suffer "slash toothbrush underscore x three underscore two zero zero five zero seven one six underscore zero zero two". Colin°^Talk 18:27, 9 April 2010 (UTC)

I think you're right, a specific skin for screen readers is probably what's needed, but you're also right in suggesting that we'll all probably be dead long before that happens. Malleus Fatuorum 18:32, 9 April 2010 (UTC)

I'd prefer to repeat the caption or even remove the caption(and put the caption in the alt) rather then go with 'alt=photograph' Regards, SunCreator ^(talk) 22:00, 9 April 2010 (UTC)

Right; can't alt=caption if alt=null? — Timneu22 · talk 22:17, 9 April 2010 (UTC)

Repeating the caption is not an option. Repeating the caption is not an option. See, that's annoying. That's what a screen reader would do if we repeated the caption. And many of our captions are quite lengthy. Plus, there are things a caption can contain (wiki markup, HTML, inline citations, additional information) that absolutely cannot appear in the alt text. Removing the caption is usually also not an option as sighted readers won't see the alt text. As for any software change to automatically achieve these effects, well we might as well ask for the no-links skin software change, so that blank alt text can be done just like other websites do. Colin°^Talk 22:36, 9 April 2010 (UTC)

I don't think you would repeat either. If the alt text is the same as the caption then the caption either becomes self-evident to the sighted person by glancing at it or it's not a replacement for the image. If the caption is not a suitable alt text then you'll require some alt text anyway. Regards, SunCreator ^(talk) 23:00, 9 April 2010 (UTC)

SunCreator, by "alt text" are you referring to "alternative text" in general or specifically to the "alt text" attribute. If the former, would you please stop abbreviating it as it is very confusing. Either way, I can't make head nor tail of your comment. Could you give an example or two. Colin°^Talk 09:10, 10 April 2010 (UTC)

Draft revision installed

The half-a-dozen comments I've received here and on the draft talk page suggest the draft text is an improvement we can work with. So, since this is a wiki, I have moved it out of my sandbox into Wikipedia so we can collaborate on it here. I'm keen to email our experts to ask if they would review version 2.0 to see if it meets their approval, and that is best done if they are viewing the real thing rather than my personal sandbox. Lastly, I want folk to start trying to apply this new approach to see if it works and then to incorporate what we discover back into the guidance.

Not only does the new alt-guidance take a different approach to the old one, but it is also shorter, which I think is important. It does cover gallery, timeline, maths, chemistry, maps, and diagrams, but it doesn't give lengthy examples of these. The Wikipedia help pages for those elements should cover the syntax for specifying the alt text; this page should worry about what to write. Colin°^Talk 21:03, 11 April 2010 (UTC)

Nutshell

Extended content

Timneu22 (talk · contribs) hid the Nutshell in this edit, saying "sorry, that nutshell was entirely confusing. I'm not sure what's trying to be said. please fix and uncomment". I've undone the HTML hiding as it means nobody can see the text to comment on it. What do other people think about the nutshell? The WCAG guideline is "Provide text alternatives for any non-text content". It later qualifies this "text alternative" as needing to "serve the same purpose and present the same information". The nutshell is currently Provide a text alternative for images that serves the same purpose and presents the same context‑relevant information. Which adapts this for our purposes (images) and restricts the "same information" to a more reasonable demand (context-relevant). The problem with just "same information" is, as other commentators have said about the WCAG guideline, that a picture is worth a thousand words: it contains more information than the context requires. Perhaps there's another way of saying "context-relevant"? Colin°Talk 07:52, 12 April 2010 (UTC) It's not even grammatically correct! I really don't get it. We can keep this discussion, but I'm removing the text. — Timneu22 · talk 10:28, 12 April 2010 (UTC) I don't know why it offends you so, and I refuse to get drawn into an edit war over it. Does its presence, even though you personally don't understand it, cause so much harm to the project that it must be removed prior to even discussing it. How is it not grammatically correct? Which bits do you have a problem with? I request you restore the text and discuss it constructively. Colin°Talk 11:04, 12 April 2010 (UTC) I'm not edit warring; I just want "nutshell" text to be clear. I have read and re-read that text multiple times, and I don't see how the information is clear to a user. I'm not being unconstructive here, but providing bogus text isn't helpful to readers. If anything, there should be a banner on the page (I'm not sure which template applies) that states "this information is in flux; we're working on it." Here are other nutshell texts: Article titles should be recognizable to readers, unambiguous, and consistent with usage in reliable English-language sources. Any material challenged or likely to be challenged, and all quotations, must be attributed to a reliable, published source using an inline citation. So shouldn't the nutshell text here be similar? Why aren't we explaining what ALT text is, as opposed to just imperatively telling the user what to do? Alternative text should be used for image captions when ... blah. That's the type of thing I'm looking for. — Timneu22 · talk 12:22, 12 April 2010 (UTC) Once you've reverted a revert of your previous edit, you're edit warring. This nutshell isn't that different to those above or the others used for policy and guidelines. There's no law saying which grammatical form they have to follow. Stick an "Editors must " in front of it if you want and it would be constructed identically to the WP:NPOV policy nutshell. This isn't a Wikipedia article so defining alternative text in the nutshell isn't appropriate. Saying the text is "bogus" doesn't help me understand how it is wrong. The nutshell is practically identical to a WCAG guideline sentence written by a committee who might know a thing or two about writing web guidelines as a professional occupation. You still haven't explained why it is not "grammatically correct" as opposed to just not being worded how you'd like it. I still don't see why it needs to be removed twice from the page rather than discussed. The suggested replacement text doesn't make sense: the image caption might be used to supply alternative text, but other other way round is nonsense. Colin°Talk 13:56, 12 April 2010 (UTC) I'm just an editor who is confused by this entire discussion of alt text. So I'd like to see it described, clearly, in a nutshell. Everything else is irrelevant. — Timneu22 · talk 14:17, 12 April 2010 (UTC) The current nutshell (before you deleted it) explained what alternative text is and what it is for. This one doesn't. It completely misses the point and doesn't actually have a point. It is more than just the "alt" tag. If you are still "confused by this entire discussion of alt text" then perhaps you shouldn't be trying to write the nutshell and shouldn't be deleting a nutshell you say you don't understand. Colin°Talk 15:25, 12 April 2010 (UTC) I don't appreciate your attacks on me or my knowledge. I'm trying to come up with something to help the WP community. Perhaps this is something that you don't understand. — Timneu22 · talk 15:40, 12 April 2010 (UTC) Oh for God's sake Timneu22, act your age, not your shoe size. Colin makes a perfectly point; don't mess with what you don't understand. Malleus Fatuorum 15:45, 12 April 2010 (UTC) Stop the attacks. I'm trying to help. This latest nutshell clearly explains the point of alt text, how it is added, how it is "rendered", and when it should be used. What does this explain: Provide a text alternative for images that serves the same purpose and presents the same context‑relevant information. The answer is that it explains nothing. Is it even a full sentence? — Timneu22 · talk 15:48, 12 April 2010 (UTC) Might I suggest this is no longer a nutshell. And still doesn't get across the point, which is what a nutshell is for. The "serves the same purpose and presents the same context‑relevant information" is the absolutely vital bit. The rest is just detail, which belongs in the body. Timneu22, would you please restore the original text that you rudely removed twice when at the same time admitting you didn't understand. I'm not saying anything about you that you haven't already told us. The nutshell above makes some basic errors like "caption is adequate to describe the image" or vast oversimplifications (like alternative text is added to images using the "alt" tag -- if you read the guideline you'll realise most alternative text is to be found in the body text and caption, not the "alt" tag). Colin°Talk 16:26, 12 April 2010 (UTC) Alternative text is complex. We've all been through enough grief on this page to know that. The purpose of the nutshell is not to provide a little orange box that tells an editor everything they need to know. Alternative text is more than just what you write in the alt attribute. If I was to review a FAC to work out if this guideline was met, I'd ask myself, for every image, does the article "provide a text alternative ... that serves the same purpose and presents the same context‑relevant information." If that is the key test for whether this guideline has been met, then surely that is a fitting nutshell for the guideline. Colin°Talk 16:42, 12 April 2010 (UTC) You're just hung up on some syntax issues now; you could clarify them (in this nutshell proposal) and it would be even better. How about working together? I believe this nutshell proposal is a better starting point than the original text: Providewho/what? is this telling the user to do something? a text alternative for images that serves the same purposewhat purpose? and presents the same context‑relevant informationto whom about what?. In all my time in the Wikipedia namespace, I've never had to read an entire page to understand what the heck it means, except for this alt stuff. Usually, I'll see the nutshell and understand completely. There is a precedent for a nutshell that is this long (WP:LIST, for one), and this starting point really explains what this means. I found "alt" because I saw a page that didn't make sense... I went to "alt", and its hatnote says to come here. But coming here, seeing that nutshell, well I'd be totally confused as an editor. Look, I maintain four wikis, I write code and customer-facing documentation; I am simply telling you that the previous "nutshell" was not customer-friendly (where users are our customer), and it needs to be fixed. It is a wiki; one shouldn't be offended that edits are being undone... the goal here is having a description that makes things easy to understand. — Timneu22 · talk 16:44, 12 April 2010 (UTC) And may I suggest that we could split this page into the help namespace, where, perhaps, my nutshell would be better suited. — Timneu22 · talk 16:46, 12 April 2010 (UTC) In answer to your blue questions. The first one is obvious. This is a guideline for Wikipedians. Yes it is telling them to do something. As I said, stick an "Editors must" in front of it if you want to, but it is just redundant text. The "to whom" question is more complex and varied than is worth trying to summarise in the nutshell: we have a whole section called "audience" for a reason. The other questions must be answered by the editor. Here's the difference between the nutshells: the previous nutshell made the editor think "How do I achieve this?" and summed up the "Have I achieved this?" test. The proposed nutshell is trying to ensure the editor doesn't have to think at all, just follow some instructions. By all means write an alt-text help page with step-by-step instructions for editors who don't read guidelines, just the orange boxes, but do it somewhere else. This is a guideline page, for editors who appreciate the benefits of thinking a bit before they write. Will you please restore the previous nutshell. Colin°Talk 17:07, 12 April 2010 (UTC) The policy is WP:BRD, so I don't know why you're asking me to change the page. You could. That being said, there does need to be a help page, but additionally, I still strongly disagree with the old nutshell. Show me another nutshell that is imperative and as vague as this one. I have never encountered one written so poorly. — Timneu22 · talk 17:10, 12 April 2010 (UTC) I'm aware of BRD essay and there's only one R in the middle. I'm simply asking you to undo your second revert, which is against BRD. The reason I won't revert you a second time is the One revert rule. Many of our guideline and policy nutshells are imperative. Really, your statement "Usually, I'll see the nutshell and understand completely." tells me you are expecting the wrong thing from the nutshell. It isn't a "just read this and that's all you need to know" replacement for the guideline. It sums up the essence of the the guideline. Your proposal doesn't do that. You're entitled to your opinion on whether it is poor, but not entitled to edit war over something what appears to be a matter of opinion. Colin°Talk 17:29, 12 April 2010 (UTC) Again, I never edit-warred. Please move past this. Geez! — Timneu22 · talk 17:30, 12 April 2010 (UTC)

Whoever made the new nutshell: well done. — Timneu22 · talk 21:34, 12 April 2010 (UTC)

Consolidating and rearranging guidelines and information

After some discussion above, it seems that there should be a help page on the alt text topic. Before getting started, I've looked around to see where current information is located. Here's what we have.

• Wikipedia:Alternative text for images
• Wikipedia:Extended image syntax, specifically, the alt text and caption section.
• Information on the WP:CAPTION page, especially the intro.
• More?

Proposal:

1. Add a hatnote to this page:
2. Create the Help page.
3. Move much of the following sections to the help page:
   • Wikipedia:Alternative text for images#Caption, alt, link and title
   • Wikipedia:Alternative text for images#The alt text attribute

That being said, I'm not sure if there is enough text to add to a Help page. As an example, WP:CAPTION has no corresponding help page. — Timneu22 · talk 17:29, 12 April 2010 (UTC)

There is merit in having an examples/how-to page for alt text. Some of the examples from the earlier version of this guideline, showing how to do table-based timelines or template-based galleries or chemical formulae could go there. The extended image syntax section on alt text could be expanded in such a page. But, that doesn't mean that the sections in this page need to move there. The two sections you highlight specifically don't give much syntax help or examples. They explain the purposes of the caption, alt and title elements and explain the rules about the alt text attribute. A help page would concentrate more on the syntax and examples. So, start the help page if you like but please don't remove any more from this guideline. It isn't a help page.

However, and it is a big one, this page has just recently undergone a rewrite and it still awaiting community and expert feedback. It might be best to wait a week or two before basing any help-manual on it. Colin°^Talk 17:40, 12 April 2010 (UTC)

Archive problem

There was a section 'External review', which was most important in the change of the status of the guideline. This section I cannot find on this page or any of the archives. I suggest it's recovered somehow. Regards, SunCreator ^(talk) 20:02, 12 April 2010 (UTC)

It is at the bottom of archive 4. Colin°^Talk 21:04, 12 April 2010 (UTC)

Bot needed?

I noticed that the the "alt text and caption" section of the image syntax page says that the caption will be used if no alt text exists. Doesn't that mean that, currently, every image with a caption (and no alt text) will have the caption repeated by screen readers? If so, isn't this a huge problem? Do we need a bot? (And if it's not a huge problem, isn't the section on this page about "blank alt" problem contradictory?) — Timneu22 · talk 17:50, 12 April 2010 (UTC)

Why wouldn't we want a screen reader to read the caption? –xeno^talk 17:56, 12 April 2010 (UTC)

My question is, I guess, if the screen reader would read the caption AND the alt text? The referenced link implies that a screen reader will read the alt text, and if it doesn't exist, will read the caption in its place. Does that mean the reader will read the caption AND the caption again (as the alt text)? It is unclear. — Timneu22 · talk 18:04, 12 April 2010 (UTC)

Maybe the question is again one of clarity: The explicitly requested Caption, if the image type has no visible caption; how is there a caption if there is no caption? — Timneu22 · talk 18:10, 12 April 2010 (UTC)

The problem with the Extended image syntax page is that for legacy reasons, it refers to the "anything not already recognised" parameter as the Caption (it is bold-italic for a reason -- it is just a parameter name, and not a very good one). The wiki syntax should have had separate alt, title and caption attributes and that syntax page wouldn't be half as confusing as it is. See User:Colin/Alt for a test page with all the permutations. There are no situations where the alt text is repeated as the visible caption text, so the text never gets repeated by screen readers. Within the combinations, there are cases where MediaWiki spits out the filename as alt text (yuk) and cases where it is blank yet is also a link so the screen reader spits out the file name (yuk). Colin°^Talk 18:42, 12 April 2010 (UTC)

So I guess that's why you always need alt text? Because various combinations produce various results? — Timneu22 · talk 18:48, 12 April 2010 (UTC)

Don't understand your second sentence but yes, currently some alt text is needed except for a very limited scenario (the last example on the page). If there was a Skin for WP, designed for screen readers, that didn't produce links to image description pages, then we could recommend blank alt text per most Web guidelines. Colin°^Talk 19:17, 12 April 2010 (UTC)

You said: There are no situations where the alt text is repeated as the visible caption text, so the text never gets repeated by screen readers. Within the combinations, there are cases where MediaWiki spits out the filename as alt text (yuk) and cases where it is blank yet is also a link so the screen reader spits out the file name (yuk). I guess I read into this, thinking you are saying alt text is always needed because of the different combinations. I'm confused; the guidelines here say that alt text is always needed, however the Extended image syntax contradicts this (alt text isn't needed if there is a caption). So... which one is right? (You mentioned Extended image syntax as legacy, but this isn't clear on any page.) — Timneu22 · talk 21:38, 12 April 2010 (UTC)

The Extended image syntax isn't a guideline on whether you should specify alt text or what to write, more of a manual on how to get the syntax right should you want to. There are circumstances where you can specify alt text using what they call the caption parameter: where there is no visible caption and the effect is to set both the alt and the title to the supplied text. The reason it does this is backward compatibility, rather than because it is a good idea to reuse the caption parameter for several purposes. The image syntax guidance page isn't itself legacy, my point is that the syntax is the way it is because it has to support old-style wikimarkup that pre-dates the ability to separately specify alt text. Colin°^Talk 08:09, 13 April 2010 (UTC)

I understand your point, Colin. But the screen readers will use the default skin. We should not provide a link to a special version of the website, or tell them to use another skin. The website itself should be accessible, the default skin should be accessible. That is easier said than done, of course.

But good solutions can be found to fix this issue, as an accessibility expert told me. But the Wikimedia Foundation has to employ an expert and to enforce accessibility to make it possible. The choices of the developers are very random concerning the accessibility of the images. For example, the last changes 7 month ago or so were devastating. Before, decorative icons were made accessible by removing the link (link=). Efforts were made to spread the use of the link= parameter for decorative icons. The some developer made a change, and suddenly it wasn't working anymore. Now we have to use both link= and alt=. For example: [[Image:Imbox notice.png|28x28px|alt=|link=]]. That's why we need to have an accessibility expert working with the developers. Does someone knows how to contact the Wikimedia Foundation and how to convince them to employ an accessibility expert?

In a way, Timneu22 is right: various combinations produces various results. The way Mediawiki handles images lack coherency and is unstable.

@Timneu22: the Extended image syntax is most probably outdated. And it is really unclear, I didn't understand it as well. You should refer to Wikipedia:Alternative text for images, it is the most detailed and up-to-date. Dodoïste (talk) 22:19, 12 April 2010 (UTC)

I disagree that we should rule out providing an screen-reader-accessible different version of the website. We are hamstrung by the fact that Wikipedia has chosen to provide attribution/licensing details via a link on every image. Other websites do this via a link at the bottom of the page, and there's no reason why Wikipedia couldn't do this for all the images on a page but it would probably be a sub-optimal experience for sighted readers, and be a disruptive change for those familiar with the site. Because each image has a link, it is now an active object rather than just some passive eye candy. The screen reader can't ignore it and must say something about the link. To compound things, we have two links for every thumbnail image. I don't know who thought that was a good idea. The developers made a mistake by changing the "alt=enlarge" on the thumbnail magnify icon to "alt=". I keep meaning to register with Bugzilla and ask them to revert that "fix". The current "alt=" version causes JAWS to read out the magnify icon's filename. I agree that Wikimedia should employ an accessibility expert. I would have thought they could get a US government grant for this since Wikipedia is an important learning resource. I plan to re-contact some academic experts to ask them to review our alt text guideline. Colin°^Talk 08:09, 13 April 2010 (UTC)

The solution could be pretty simple: "alt=enlarge" by default on every image, and better caption guidelines to make the caption sufficient for non-sighted users. That is the solution a accessibility expert (who is admin on fr.wikipedia) came up with. But he also stated that currently Wikipedia is unable to implement such a thing. Any modification of the software would be nothing more than another change in a list of random changes. Some developer is likely to break it in a few month or so. That is why we need an accessibility expert to work with the developers. Yours, Dodoïste (talk) 23:04, 15 April 2010 (UTC)

Using "Alt=enlarge" by default? That's a great idea! I should have thought of it when I was asked about this topic a couple of years ago. The devs aren't quite as random as they'd lead you to believe. I'll contact one if we come to a definite consensus here. What should be done with images which use the "thumb" syntax which make them contain two links? Graham87 03:37, 16 April 2010 (UTC)

Technically, the pic on the image description page isn't always larger than the one in the article. If one wanted to truly describe the "purpose" of the link, it would be "image details for file Tony_Blair19970606.jpg". The ideal would still be to remove it (and the second thumb link) as neither are required for screen reader users. For thumbnails, both links would need some alt text. They both serve the same purpose though one is a picture and the other an icon. We could have "alt=enlarge" for both but "enlarge enlarge" spoken for every picture in an article could be wearing. What about giving the icon "alt=thumb" for thumbnail images? Alternatively, "alt=thumbnail" if the abbreviation isn't best. That would read "enlarge thumb". For frame or plain images (such as tend to appear in infoboxes or tables), the single picture link would cause just "enlarge" with no secondary "thumb" link. Colin°^Talk 08:05, 16 April 2010 (UTC)

Hmmm, in that case, the best alt text that I can think of would be "show file details". IMO that alt text would succinctly describe the function of the link without either showing the filename or assuming its an image, since certain ways of adding video and audio also generate bare links to the file description page. Also "enlarge" or even "enlarge and/or show file details" would work for me, but I doubt that a person who uses alt text would often need to enlarge the image. "Show file details" implies that the image will be shown with more detail, in a way. I would prefer that users of alt text don't encounter the original filename, since sighted users don't normally see it. As for thumbnail images, "thumbnail icon" would be my preferred alt text, since it is, after all, an icon. As for the links being shown to screen readers, I would prefer to encounter them because I sometimes want to find out more about the attribution of a certan image/video/sound; there's no reason to deny this information to people just because they can't or don't wish to see images. The thumbnail links are a different matter, however. Graham87 10:03, 18 April 2010 (UTC)

See my proposal at Wikipedia:Village pump (technical)#Javascript help needed with experiment on image links and alt text. I think we could experiment on different options and then ultimately request a bug fix/change for MediaWiki. I think we all agree the thumbnail icon links aren't useful (in fact, I question their use for sighted readers too). I wondered if moving the picture links down to the bottom of the page would be a compromise: they would still be present for those who need them, but they wouldn't interrupt the article which really should be readable without interruptions concerning the technical aspects of the content or its licensing and attribution. There's also scope for user-preferences: some screen reader users may not wish to see the image links in place, some may want them at the bottom, and some may want the stripped out altogether. The concern I have with the above proposal to default some standard alt-text is that it could make it harder to post-process the HTML to identify which images genuinely had no editor-specified alt text. Colin°^Talk 10:48, 18 April 2010 (UTC)

Would it be possible to add a CSS class to images with no editor-specified alt text? Something like <span class="noalttext"> .... This would automatically be removed by MediaWiki if an editor added alt text. That's the easiest way I can think of to solve the problem. Unfortunately I don't know any JavaScript. Graham87 12:53, 18 April 2010 (UTC)

Well in that case it looks like there is much cleanup to be done. — Timneu22 · talk 00:11, 13 April 2010 (UTC)

I recently revised the Extended image syntax section on alt/caption/title but was afraid to make too radical a change (such as dropping the word caption for the overloaded parameter). I think it is currently correct but still difficult to follow. The real problem is that the syntax itself is badly designed but there is really nothing we can do about that. Colin°^Talk 08:09, 13 April 2010 (UTC)

Image unlinking script

Responding to the Village pump request, I've created a script which will move image attribution links into the footer. To install this script add {{subst:JS|User:Dispenser/imageunlink.js}} to your skin's script page. I have only tested in Firefox 3.6 with the Monobook, Vector, and Modern skins. — Dispenser 19:53, 18 April 2010 (UTC)

It works in IE7 with JAWS as well. The only major problem I can find is that with the script, when an image has no alt text, the screen reader reads the caption but doesn't indicate that there is an image there. This turns the image captions that I wrote at white cane to nonsense; those images have no alt text and probably don't need it. The advantage of the script is minimal when the image has alt text, since images with and without links read the same way, except that images with links have the text "link graphic" appended to them while images without links just have the text "graphic" appended to them. However the script is good for removing those pesky thumbnail links, and it's good to experiment with different approaches. Graham87 00:52, 19 April 2010 (UTC)

Thanks for doing this, Dispenser. Really appreciated.

Presumably before the imageunlink.js, the photos at white cane read out the filename, because the alt text is blank yet the images had links. That isn't a nice option. The WebAIM guidelines only recommend blank alt text where the image is purely decorative. The white cane article goes further and has blank alt text for images that aren't decorative yet the alternative text requirement is fully met by the caption. I guess these are situations where some very basic alt text is required (such as "photo"). In the guideline, the Water Fluoridation image is purely decorative and the caption text makes sense even when the graphic completely disappears as far as screen reader users are concerned. This would be an appropriate use of blank alt text. But the Toothbrushes example is similar to your white cane examples: the picture isn't decorative and the caption only makes sense when you know it is describing a picture. So the toothbrushes require some alt text such as "photo".

I prefer the flexibility this script gives user: we can now truly decide as editors whether the alt text should be blank.

We should to amend this guideline to no longer give the impression that blank alt text is a consequence of a descriptive caption. It should distinguish that case from the case of purely decorative images.

Where do we go next? It isn't satisfactory to expect blind users to get an account and edit some obscure setting to run some Javascript. Can users choose different skins without having to logon? Can JAWS be made to run Javascript on Wikipedia pages? Colin°^Talk 10:53, 19 April 2010 (UTC)

As Graham87 stated, it's good to remove the thumbnail links. They are useless for sighted users, never used on user-friendly Websites nowadays, and bad accessibility-wise. But this change should be make in MediaWiki, through bugzilla. Rather than generate such links and remove them with Javascript, it's much better to not generate them at all. Plus, anyone without Javascript would still have those annoying links.

Now, removing the link to a larger image or its details is a bad idea. It won't help blind users at all. And accessibility is not only about blindness, there are a huge lot of other things to take into account. It is way too complicated for non-experts to think about all those cases and find good solutions, that's why the experts created standards. Standards exist about this common issue, so we should follow these standards. That's all there is to say.

"Photo" as an alt text wouldn't help: it doesn't add any useful information. Instead of "image /wiki/File:Long_cane.jpg", a screen reader would read "image photo". The difference is not significant. "Enlarge" would have been a good choice. Colin, I know you want to help, and that's nice. But unfortunately, you are going in the wrong direction. I am at your disposal to answer your questions about accessibility. Together we can work on good solutions. But if you try to find solutions without the right approach and knowledge you'll and up making mistakes.

Oh, and please read my reply about the idea of accessible skin. Yours, Dodoïste (talk) 11:41, 19 April 2010 (UTC)

Dodoïste, I'm not sure what your problem is. This was an experiment. I used the word "experiment" in the village pump request. I don't think any Javascript-user-logged-in solution is viable. But it is a way of testing ideas. I see you have opinions on this. Please don't patronise me with your "that's nice" and "It is way too complicated for non-experts to think about" comments. I respect experts and I respect standards. Until I experimented with various caption/title/alt options at User:Colin/Alt, our Wikipedia:Extended image syntax page was total bollocks. It is still over-complex and confusing but at least it isn't bollocks. Experiments are useful. It is only through experiments that we discover things like the blank alt text problem.

Both "photo" and "enlarge" are reasonable compromise words. Arguments can be made for either, and I don't much care which one wins. That the thumbnail icon has title "enlarge" and should have had alt text "enlarge" is an argument against repeating the word for the photo itself. Ultimately, the thumbnail icon should go, we all agree on that. I strongly disagree that there isn't much difference between a short word and the filename, which is typically much more tedious than the "long cane" example.

I have not proposed anything yet, just kicked some ideas about. My experiment involved removing links to the detail page. My preference would have been to move the links to a box at the bottom of the article, but that kind of work wasn't the purpose of the experiment: which was to see what the effect on screen readers was. Graham has confirmed this.

If your responses contained information rather than just criticism and calling things a "bad idea" without saying why, we might make progress. Please tell us (a) what might be wrong with the current guideline assuming the MediaWiki software is as it currently is and (b) what you think could be changed to improve things. But let's keep it pragmatic. It would help if you cited some published standards to back up some of your opinions. Colin°^Talk 12:44, 19 April 2010 (UTC)

I understand that you were making experiments with Javascripts. I'm fine with that; as long as it stays in a few personal accounts it's fine. I could be useful for some. But you were also saying that this could lead to the creation of an accessible skin. And that is precisely what I'm opposed to.

Concerning "photo" and "enlarge". There is no alt text describing the image. We still have a link pointing to the file's page that doesn't contain any useful information. See how to meet 2.4.4 : identifying the purpose of a link, W3C. So the solution is to replace the empty alt by "enlarge" to describe the purpose of the link. This solution was recommended by an accessibility expert who is admin on fr.wiki, and wrote a list of best practices : fr:Wikipédia:Atelier accessibilité/Bonnes pratiques. This best practices are made especially for Wikipedia and it's context. Yours, Dodoïste (talk) 12:36, 22 April 2010 (UTC)

I've just quickly read your French guideline. I'll read it more carefully later, but on first glance it looks remarkably similar to our guidance. It correctly identifies that on Wikipedia the usual advice for blank alt text is undesirable and provides similar advice to our Water Fluoridation example. It stresses that the fact that our images link to image-description pages shouldn't affect what we write for alt text, though it does affect the fact that we then have to write some alt text. I don't know where you find any advice that "the solution is to use the empty alt text to describe the purpose of the link." The WCAG guidelines only recommend blank alt text for links if the link also contains link text (which ours don't, the text doesn't form part of the link) and the link text adequately describes the purpose of the link. —Preceding unsigned comment added by Colin (talk • contribs) 13:41, 22 April 2010 (UTC)

How does empty alt text describe the link? Isn't it the same as the status quo for most images now? What's wrong with changing the MediaWiki software to spit out some unique default alt text like "show file details"? Graham87 14:04, 22 April 2010 (UTC)

Oh sorry, I wrote that answer in a hurry. What I meant was to replace the empty alt by the text "enlarge". In this context, "enlarge" describes the purpose of the link, whereas "photo" would not help at all. Dodoïste (talk) 20:11, 22 April 2010 (UTC)

For most images on WP, we have two images and links: the one the editor put there for editorial purposes, and the thumbnail icon the software adds. The WCAG guidelines recommend the latter should have alt text that describes its purpose, not its appearance: it should say "enlarge" or "show file details" or similar. This is what it used to say and what the link title says but a misguided MediaWiki bug "fix" replaced it with blank alt text. For the image the editor put there, I don't think the primary focus of the alt text should be on the incidental (to the reader) purpose that clicking on the image links to the image description. That isn't why the editor put it there. Look at the WCAG questions in our guidance and consider that the link is only there because the image is: if this image was removed, the link would be too. Therefore the link is the least important priority for the image alt text.

I think we should follow WCAG guidelines, the French WP guidelines and our own example at Water Fluoridation to include the most basic descriptive alt text in the case where we might otherwise have blank text. At its extreme, in the Toothbrushes example, there really isn't much to say beyond what the caption ("toothbrushes") already says, and I proposed "photo" or "painting", etc would be a reasonable option in the circumstances: this is a photo of some toothbrushes.

There seems little point in having two alt texts saying "enlarge" or "image details", especially when that isn't the primary reason why an editor included the image there at all. On this particular issue, perhaps what is best is if we create a new section and invite suggestions on how best to resolve the "blank alt text" problem with the minimal changes to MediaWiki, knowing that radical changes are unlikely. Colin°^Talk 21:30, 22 April 2010 (UTC)

While I was making a quick listing of english accessibility guidelines, I found the IITAA wichs explains that we should use separate accessible versions (of a Website) only as a last resort. This confirms the french explanations and guidelines that I previously found.

Concerning the blank alt text issue, it's too early to focus on that. We need involvement from the developers and the Foundation, see section below. Dodoïste (talk) 14:28, 1 May 2010 (UTC)

Screen reader audiences and uses

Not all screen reader users are totally blind. Some have partial sight, and either use screen readers in addition to screen magnifiers, or only use screen readers because reading text is too slow for them. I also know of a fully sighted computer user who uses screen readers because he has severe dyslexia. Screen reader users can interrupt their speech synthesizers at any time; personally, when I know that a filename is coming up, I just press down arrow to interrupt the text and move to the next line.

I originally wrote the above text at Wikipedia:Village pump (technical)#Javascript_help_needed_with_experiment_on_image_links_and_alt_text. However it was suggested that I copy it here, so that's what I've done. I thought it would be a good idea to note how people use screen readers. Graham87 14:01, 19 April 2010 (UTC)

Wikimania 2010 and accessibility

I would like to talk about accessibility at Wikimania 2010, and convince the Foundation and the developers to care about accessibility. That is a necessary step. I'd like to give a conference about accessibility there, inspired from the conference about Wikipedia two french accessibility experts held 7 months ago. Is someone willing to come with me ? User:Sj is interested in accessibility and will come to Wikimania 2010. Dodoïste (talk) 14:28, 1 May 2010 (UTC)

I'm not in any position to go to Wikimania 2010, but that sounds like a great idea! Graham87 14:52, 1 May 2010 (UTC)

I wouldn't be able to attend. I agree it would be good to encourage WP to do something about accessibility. Colin°^Talk 17:35, 1 May 2010 (UTC)

Separate editions of WP

Some potential solutions to the "blank alt text problem" could involve editions (skins) of WP that work better with assistive technology. Dodoïste seems dogmatically opposed to such an idea, which I think is a mistake. The guideline cited above discourages "text-only versions" on the grounds that they tend not to be kept up-to-date and that proper application of other guidelines make them unnecessary. The former doesn't apply to WP: the HTML we see is generated from Wiki markup which is ignorant of the final appearance. The latter is a matter of opinion that ultimately can only be tested through experiment. Currently, WP's HTML is suboptimal for users of screen readers.

The typical web user is changing. People are increasingly surfing on Netbooks, PDAs, mobile phones and even their television. Some may use Wikipedia through a different client than a web browser. Editions of WP for these different technologies are being developed; assistive technology variants could benefit from variants too.

WP also has two kinds of user: reader and editor. The former greatly outnumber the latter yet the UI is designed round the latter. Many people read Wikipedia via mirror or other reference sites, which provide their own HTML variants and don't allow editing.

This is somewhat off-topic for this guideline, but I just want to encourage any technology-change discussion to be open-minded about the solutions. If we do "convince the Foundation and the developers to care about accessibility" we should allow them freedom to develop a solution in consultation with experts, rather than impose a solution or restrict the options. Colin°^Talk 17:35, 1 May 2010 (UTC)

Having trouble accessing alt text on mouseovers

I have tried restarting my IE 8.0, I also rebooted my computer, I even switched over to Firefox, and I still cannot see the alt text when I mouseover an image that contains the alt command. The mouseover works when I hover the mouse over a link, but not when I try to view the accessibility alt codes for images. What am I doing wrong? Is the alt code mouseover function blacked out? Is this adversely affecting blind users of Wikipedia?
 —  Paine (Ellsworth's Climax)  07:41, 10 May 2010 (UTC)

• PS. Also asked about this at the Village pump. —Preceding unsigned comment added by Paine Ellsworth (talk • contribs) 10:32, 10 May 2010 (UTC)
  • Nothing has changed on my end regarding alt text and how it is spoken. Graham87 13:04, 10 May 2010 (UTC)
    • See Wikipedia talk:Alternative text for images/Archive 5#Is there a problem with alt text being served?. Colin°^Talk 14:03, 10 May 2010 (UTC)

head shot of distinguished elderly man wearing eyeglasses and a smile

Napoleon Hill

Thank you, Graham and Colin! Colin, I caught your note at the Pump and responded there.

 —  Paine (Ellsworth's Climax)  09:15, 11 May 2010 (UTC)

And now I'm seeing the tooltip again when I mouseover. Go figure.
 —  Paine (Ellsworth's Climax)  03:27, 13 May 2010 (UTC)

Have you got the "Compatibility View" option ticked (under the Tools menu) on IE8? If that is ticked, then you see alt text as a tooltip. Colin°^Talk 12:36, 13 May 2010 (UTC)

No, it's a real puzzle, Colin. I was trying different things, but then I set everything back to normal and checked, but still didn't see the alt text as a tooltip. Then I started doing some other stuff on the Hill article, and viola! I happened to run the mouse over the image and there was the alt text as a tooltip. Still working, so I'm keeping my fingers crossed <g>

 —  Paine (Ellsworth's Climax)  22:30, 13 May 2010 (UTC)

Complaint

This guidance is badly written and self-contradicting. It extols us "where [an image] has been chosen to illustrate what [a person] looked like—the alternative text should describe his appearance" and at the same time says "[an image showing the appearance of Elizabeth II] should not be described as "an elderly lady wearing a yellow dress". So, now I am unclear as to whether the alt text[1] I removed from Elizabeth II, which is solely used to describe her appearance, should have been left in after all. Guidelines this long and confusing are unlikely to be followed. DrKiernan (talk) 18:27, 2 June 2010 (UTC)

Is this the only aspect that is "badly written and self-contradicting" or are there other examples? Looking at your edit to that article, I see you've simply removed all the alt text, which is almost never the right action. Where does this guideline recommend that be done wholesale?

The two quotations from the guideline above are not contradictory. The "alternative text" can include the caption and body text, not just the alt-text attribute. Aspects of Elizabeth II appearance should probably be covered in the body text of the article on her. The statement "an elderly lady wearing a yellow dress" does not tell us what Elizabeth II looks like. It tells us what the subject of a photograph is. But unless the photo could be replaced with one of my gran in a yellow dress, that text is not an "alternative" to the purpose that photo serves in this article.

One of the aims of the rewrite of this guideline was to shorten in. Most of the guideline consists of examples. Alt text is a fairly complex issue, not just one of whether to put punctuation inside quote marks or not. Colin°^Talk 07:43, 3 June 2010 (UTC)

I wrote that alt text: [2]. By my reading of the current guidance, the alt text I wrote is all wrong, so I removed it. I have not replaced it because I cannot tell from the current guidance what the alt text for any of the images should be: indeed, for most the guidance seems to say that alt text is not required, since the caption describes the image anyway. DrKiernan (talk) 09:03, 3 June 2010 (UTC)

That may be so, but see the "Blank alt text problem" section for why having no "alt" attribute isn't optimal. I'd offer to look at the article and have a go myself but I just don't have any free time at present. Colin°^Talk 13:36, 3 June 2010 (UTC)

Template:Infobox NBA Player

Template:Infobox NBA Player needs to be made WP:ALT compliant.--TonyTheTiger (T/C/BIO/WP:CHICAGO/WP:FOUR) 23:45, 3 June 2010 (UTC)

Another complaint

The guidelines seem to have changed dramatically since I last looked yet I can find do record of any discussion to this end. Can someone please tell me what's going on? As far as I was aware the purpose of alt text was to aid blind and partially sighted readers by providing them with a description of images in articles. I really don't see the point of using alt-text to effectively just duplciate a caption. Cavie78 (talk) 22:09, 21 June 2010 (UTC)

• I should also point out that the FA and GA communities seem to be largely unaware of any changes and are passing articles which feature 'old style' alt-text.Cavie78 (talk) 22:13, 21 June 2010 (UTC)
  • Alt text is not a requirement for either GA or FA, so there's no reason for those "communities" to be concerned about it. Malleus Fatuorum 21:35, 22 June 2010 (UTC)

The situation is that we have no guideline on alt text at the moment, because there's no clear consensus on how it should be used, so if people want to use the old-style long alt text that's fine, but there's no requirement for anyone to do so. SlimVirgin ^{talk|contribs} 20:55, 22 June 2010 (UTC)

Edits 3rd June

There are problems with the edits made 3rd June. Sorry this is so long, but I've mostly reverted them so feel I have to justify why.

Firstly, they heavily promote the "Refer to caption" or "Refer to adjacent text" alt-text suggestion, which is actually one of the options the guideline was changed to discourage. The lead text now suggests this along with a reference to two sources. However, neither of those sources make that suggestion, and I haven't seen any alt text guideline suggesting that approach. I ask that we consider whether a screen-reader user wants to hear those deeply unhelpful words several times every time she reads a Wikipedia article. The caption is spoken next anyway and the reader is reading the adjacent text anyway. If we don't need to write alt-text then we should ensure we write as next-to-nothing as possible, and what we do write shouldn't be annoying.

Secondly, many of the edits indicate a misunderstanding of the what this guideline is about, and what all "alternative text" guidelines are about. The W3C and Wikipedia guidelines are about providing alternative text for images. They are not about writing alt text attributes in HTML or Wiki markup, though that is one solution. This "alternative text" may be in the alt text attribute, or the caption or the body text.

Thirdly, the edits fail to maintain the natural progression of the guideline and the purposes of each section. Instead, suggestions for alt-text wording are scattered throughout. The guideline was written with a planned order. We explain the who and why of alternative text. We explain some of the syntax involved and the other elements such as captions and link titles. We give some general advice. Once the reader has a grounding in the issues and technicalities, we then give detailed advice on the alt text attribute. Lastly we discuss the problem of blank alt text. Then, as an appendix, we give a number of examples.

• This edit promotes the "Refer to caption" or "Refer to adjacent text" approach in the lead above all others. Indeed it suggests it before the alt attribute is even explained to the reader (third paragraph of the lead). Let's keep the lead paragraph dealing with the what and why of alternative text, without getting bogged down in one technical approach, and one that isn't particularly good. The two sources given here are already in the references.
• This edit also promotes the "refer to approach". Regardless of the merits or otherwise of that approach, this is not the place to offer alt-text suggestions. This section is a purely factual description of the caption, atl, link and title aspects of an image, offering no guidelines. It should remain so. Every section/paragraph should have a purpose and this section is informational, not guideline.
• The above edit also removed the following:

For example, the caption may simply be a sound bite from the article where the combination of image and caption is a technique used to grab the attention of a reader who is merely skimming. The caption often includes information that is not present in the image and goes beyond just identifying the image. It can do this because a caption is able to support inline citations should this information be challenged.

This is important information. People find the alternative text concept confusing (we all have, and see above) so let's not start removing helpful information. During earlier discussions, it was commonly suggested that the caption sufficed as alt text. In fact, if you look at thumbnail images on Wikipedia, you find that many have absolutely nothing to do with the chosen image. You might be familiar with the sound-bite-below-an-image concept but not all of our editors will. The second half of the text here underlines a very important difference between what one can say in a caption and what one can write in alt text. Why would we remove that?

• This edit moved the information about what the alt attribute is and some tooltip details to the start of the Audience section. Why? This is nothing to do with audience the audience section deals with "alternative text", not just "alt text". It also shortened the section header "Caption, alt, link and title" to "Caption" but the content still referred to links and titles. How about "Image syntax elements" as a section heading.
• This edit once again promotes the "refer to caption" approach to the exclusion of all others and removes the link to the "Blank alt text problem" section which deals with the issue in more detail.
• This edit has a distinctly misleading title as it removed a whole paragraph rather than "moving something about captions into that section". The removed paragraph offered two solutions to the "blank alt text problem". Now it just leaves the reader with a problem. !
• This edit tries to restore a little of the blank-alt-text-solution but puts it in the newly titled "Caption" section. As said earlier, this section provides information, not guidance, so isn't the place to start making alt-text suggestions. Having alt-text guidance in a section called "Caption" is just confusing. This section appears before the "General guidance", "Context" and "Alt text attribute" sections which lead the reader towards an understanding of "alternative text" before getting into the technical guidance of what to write where.
• This edit removes the sentence "If the image has a caption, this should be considered in combination with the image alt attribute when composing alternative text." That sentence seems to be good advice in a section on Context. Why remove it?

Please can we discuss "blank alt text" solutions such as "refer to caption" before re-introducing them. If there are issues with the order that information is presented, let's discuss how to improve that. But there is a plan to the current order and littering the guideline with suggestions out-of-order will make it a confusing mess. Colin°^Talk 09:06, 4 June 2010 (UTC)

Blank alt text is not the same as "refer to caption," which had consensus, and I think was removed without discussion, so I'd like to restore it. One of the sources did indeed recommend it, as I recall, which is one of the reasons it was added. We don't have to recommend it, but we also shouldn't discourage it—just add it as an option for editors who don't want to be repetitive where the caption or nearby text suffices. SlimVirgin ^{talk contribs} 03:01, 5 June 2010 (UTC)

Let's not kid ourselves that the previous text, when this was a guideline "had consensus". It didn't. You made it quite clear during the discussions that the community had not been sufficiently involved in establishing that guideline. And a swelling of opinion rose up against it. When a damning verdict was cast by an external expert, it was demoted. The approach taken by the previous text was wrong. If there was anything of merit in the previous text that hasn't been incorporated into the current text, it needs to be argued for again, rather than making any assumptions about previous consensus.

I like your "I think was removed without discussion" slur. The draft replacement text was proposed and widely advertised. I left a message on your talk page, along with several others who had shown an interest. The most vocal critic of the previous text (Malleus) heaped praise on the new text and helped copyedit it. Graham87 also reviewed it and was happy with it. SunCreator also participated in the discussion. You chose not to participate. Later, another reader came along who pointed out the French WP had a similar guideline,[3] apparently written with an expert. It too discusses the difficulties of not being able to use blank alt text (as it causes the filename to be read out). Their solution, like one of ours, is to provide the concisest possible description of the image -- "Dans ce cas, l'alternative, aussi concise que possible, devrait décrire factuellement l'image. [In this case, the alternative, as concise as possible, should factually describe the image.]"[4]

We already have better "option[s] for editors who don't want to be repetitive where the caption or nearby text suffices." A wordy alternative that has no information and in one case ("Refer to adjacent caption") is positively ridiculous (because 10 milliseconds later the caption is spoken out) doesn't seem worth suggesting, let alone promoting in three places including the lead.

If you can get consensus for the "Refer to caption" or "Refer to adjacent text" solutions to the blank-alt-text problem, then by all means insert it in the relevant section. I would appreciate a link to the source that "did indeed recommend it". I'd be very surprised to find an external source recommend that solution when having just blank alt text would be infinitely preferable and achievable to anyone with better control over the HTML than we have. Colin°^Talk 18:29, 5 June 2010 (UTC)

I'm not talking about the previous version when it was a guideline, but about the version after guideline status was removed. We added that "refer to caption" was an option when the caption adequately described or evoked the image, or described its function. Ditto with nearby text. There were no objections to that.

I find the current version of the page confusing, so it's important to offer some clear options. One of those (just one) is "refer to caption" to prevent repetitive and pointless ALT text being added where the caption suffices. SlimVirgin ^{talk contribs} 18:37, 5 June 2010 (UTC)

No, the "see adjacent text" / "see caption" option was part of the guideline before. You may have moved some text around and copied it to the lead since but it was in there along with the Mona Lisa example. No discussion took place on that particular suggestion post-guideline. I would appreciate if you'd get your facts right or better still avoided history and discussed your proposed changes on their merits. The "refer to caption" solution is "repetitive and pointless ALT text". So how does that "prevent" such a thing?

Perhaps people wiser than us will come up with better solutions. Perhaps the MediaWiki developers will take accessibility seriously and offer some alternatives to a seriously unfriendly UI for screen readers. The vital thing about this page is to offer guidance on "alternative text" (which as WebAIM also state, is not just about alt text attributes). Once editors understand the purpose of the alternative text, and the limitations HTML and MediaWiki impose on us, they may find their own solutions. I very much doubt "refer to ..." will be one of them. Put yourself in the screen-reader user's shoes. Imagine that most times you look at a thumbnail image on Wikipedia, you have to listen to the words "refer to caption" before the caption text appears. At what point would you go insane and start writing angry letters to the WP:ALT authors? Colin°^Talk 22:07, 5 June 2010 (UTC)

Colin, I'd appreciate a less aggressive response. What do you see as the problem with allowing "refer to captIon" as an option so long as the caption is sufficiently descriptive? It is surely better to hear those three words repeated than to hear a larger number of words repeating what the caption says. SlimVirgin ^{talk contribs} 22:28, 5 June 2010 (UTC)

You get the response you deserve SV. Let's stick to discussing proposed changes on their merits. We all agree that we don't want to repeat the caption or adjacent text when those satisfy the "alternative text". We also agree that MediaWiki prevents us writing "alt=" since, for linked pictures, that causes the screen reader to read out the filename, which is pretty awful. We, mostly, can't remove the link. Whatever solution we have to this tricky problem is an artefact of current MediaWiki limitations and choices. I do believe that when the Foundation takes accessibility seriously, someone will design a screen-reader-friendly skin. It may, for example, remove the picture link but keep the thumb icon link, and give the thumb icon the proper alt text it deserves (e.g., "magnify", or "image details").

So, for situations where any web designer would have chosen blank alt text, the suggested text options are kludges. Let's keep those suggestions low profile. They aren't the most important aspect of understanding what alternative text is, or how to deal with it. That's one reason I don't want any one suggestion repeated three times.

Please read the final body paragraph in this guideline. It suggests two solutions, one where the caption is descriptive of the image and one where it is not. Those two solutions also have examples: the Water Fluoridation photo and the Toothbrushes photo. I'm very open to suggestions of better solutions but don't think "refer to ..." is one of them. Perhaps others will disagree and consensus will go against me. Why don't you find an example where you'd use that solution and where you think it is preferable to one of the solutions already proposed. We can then mock-up an example and judge it. Colin°^Talk 07:23, 6 June 2010 (UTC)

Colin, you reverted all the changes I made on June 3, including the simple copy-editing that I felt tightened it and made the structure flow a little better. [5] I haven't restored those changes, but I did restore "refer to caption" or "refer to adjacent text" as options, but only as options. [6] I hope that's a sufficient compromise.

I'm willing to look around for examples, but I'm not entirely clear what kinds of examples you'd like. SlimVirgin ^{talk contribs} 01:45, 9 June 2010 (UTC)

I carefully explained in at the start of this section why your edits weren't "copy edits" and actually worsened the flow rather than improve it. Your "tightening" removed important information and inserted new material. I appreciate my comments were probably TLDR but at least I've explained myself.

You ask "I'm not entirely clear what kinds of examples you'd like"? I don't see what is hard. This is a guideline (or wants to be). You are proposing we adopt a form of words for the alt text. Find an example where that works. Find an example where it works better than the two other options you've demoted in favour of yours, which is also now singled-out in the lead. If you can't find such an example, the proposed "refer to" option fails and should be removed.

Why couldn't you find examples first and discuss this on the talk page rather than edit war to get your preferred option back on the page? I'm sure you are familiar with WP:BRD. I hoped we could resolve this by following such a practice. Clearly you'd rather take the alternative approach, and have restored your favoured text without discussion or agreement. I follow the WP:1RR and so won't revert your text again without us coming to some agreement on the talk page. I shall propose some changes I would like to see here, later when I've got some time. You've gutted the "Blank alt text problem" section and you've moved specific solutions to informational sections that appear so early in the guideline that the reader hasn't read enough general advice to appreciate the problem those solutions fix. Colin°^Talk 08:39, 9 June 2010 (UTC)

Policies and guidelines can't be imposed on the project by one or two editors; they have to be descriptive as well as prescriptive. This means they have to evolve to the point where it's clear to any active editor that they have rough consensus. Therefore the best thing for now is to have a page laying out some reasonable options for people. Over time, best practice will emerge as people choose from those options, then a guideline can be constructed that reflects that best practice.

You asked me to find a source supporting that the caption or nearby text was sometimes sufficient, so I added a footnote quoting what the source says. The source offers it as advice; we offer it only as one option among several. SlimVirgin ^{talk contribs} 08:46, 9 June 2010 (UTC)

I didn't ask you to "find a source supporting that the caption or nearby text was sometimes sufficient". I have no problem with that statement. It is quite clear from many of our sources that this occurs and that the best solution is blank alt text, which unfortunately we can't do. That is the essence of the blank text problem. I asked you to find a source that recommended writing "refer to caption" or "refer to adjacent text" as a solution to that problem. The sources you cite do not make that recommendation. Nor would they. It is not a good solution.

How is the "refer to adjacent text" option "descriptive"? It does not describe existing WP practice or Web practice. It is prescriptive. But it isn't an evidence-based rule. Rather it is just an idea that occurred to someone. Ideas are good but they belong on the talk page until they can be shown to have any value.

Let me be really clear so we don't continue to misunderstand. I want a source that recommends the words "refer to caption" or "refer to adjacent text" or similar. The source citation in the guideline is misleading as it doesn't back up the advice given by the guideline. I also want an example, in the form of the examples at the bottom of the guideline, where "refer to caption" or "refer to adjacent text" is presented. We can then compare that approach to the other options which are to use the word "photograph" or "painting" or similar when the caption is descriptive or identifying, or to use the shortest possible description when the caption is not. We have two examples currently, where the existing two solutions work quite well. So those solutions are evidence-based. If you can make a third solution work, fine. Colin°^Talk 09:29, 9 June 2010 (UTC)

Colin, blank alt text is when there are no words.

I have restored "refer to caption" or "refer to nearby text" as an option (only as an option), as the sources do. I added a source and I quoted from that source. "WebAim writes: "[T]he alt attribute (sometimes called the alt tag, though technically this is incorrect) is not the only mechanism for providing the content and function of the image. This information can also be provided in text adjacent to the image or within the page containing the image. ... The term alternative text, as used in this article, refers to the text equivalent for an image, regardless of where that text resides. It does not refer solely to the alt attribute of the image tag." See WebAIM. Appropriate Use of Alternative Text, accessed June 8, 2010 (my bold).

Now that I know what kind of example you've asked for, I will look for one. SlimVirgin ^{talk contribs} 09:37, 9 June 2010 (UTC)

No, blank alt text is not "where there are no words". That would be "missing alternative text", which is not acceptable under any circumstances. We've been through this before. The word "alt" is used solely wrt to the attribute. Please stop confusing this issue. If you can't grasp this I despair.

The sources do not recommend including those "refer to..." words anywhere in the article. I completely agree that alternative text can be supplied through a combination of alt text tag, caption and body text. It is not a contentious issue that alternative text can be the adjacent text. The contentious issue is writing the words "refer to caption" or "refer to adjacent text" in the alt text attribute. How much clearer do I have to make it? Colin°^Talk 09:56, 9 June 2010 (UTC)

Colin, if you keep responding to me in this way, I'm going to stop replying. We all may despair of each other from time to time, but there's no need to spell it out.

Can you show me a source showing that "blank alt text" means what you say it means, whatever that is? SlimVirgin ^{talk contribs} 10:07, 9 June 2010 (UTC)

This guideline uses the exact phrase "blank alt text". Both the WebAIM and the W3C guideline pages exclusively use the term "alt text" for the attribute's value and "alternative text" for the text-alternative-to-the-image wherever that may be. The WebAIM document says "empty or null alt text" or "empty alt text". Not really seeing a difference between "blank" and "empty" but we could change the word if you think it is better. Colin°^Talk 10:20, 9 June 2010 (UTC)

Example for "refer to caption"

Here is an example I've referred to before from Death of Ian Tomlinson, about a British man who died during a G-20 protest after being pushed to the ground by a police officer.

Ian Tomlinson remonstrates with police after being pushed to the ground, minutes before he died.

I'm offering this as an example of when alt=refer to caption would be appropriate. The actual alt text I added to this image was a tediously long description of it—this was before the guideline changed:

A small crowd scene. On the right, four people dressed in uniform, their heads and face mostly not visible, wearing yellow and blue jackets, black trousers and black shoes. They are carrying long thin sticks, and round transparent shields. On the left, there are three men. One is on the ground, sitting with his legs straight out, and his arms raised, looking at the people dressed in uniform. He is wearing a grey and blue top and black trousers with a white stripe. Two men are leaning over him; one is holding his arms. The latter is wearing a dark hooded top, grey trousers with white stripes, and has a grey and blue bag over his shoulders.

SlimVirgin ^{talk contribs} 09:49, 9 June 2010 (UTC)

Normal viewing	Screen reader	Rationale
[[File:Ian Tomlinson remonstrates with police.jpg|200px|thumb|left|alt=refer to caption|Ian Tomlinson remonstrates with police after being pushed to the ground, minutes before he died.]]
Ian Tomlinson remonstrates with police after being pushed to the ground, minutes before he died.	link graphic refer to caption. Ian Tomlinson remonstrates with police after being pushed to the ground, minutes before he died.	Article: Ian Tomlinson The purpose of the image is to show a scene from the events where Ian Tomlinson was interacting with the police prior to his death. The caption describes the image and also contains additional information ("minutes before he died") that is not present in the image.
[[File:Ian Tomlinson remonstrates with police.jpg|200px|thumb|left|alt=Photograph|Ian Tomlinson remonstrates with police after being pushed to the ground, minutes before he died.]]
Ian Tomlinson remonstrates with police after being pushed to the ground, minutes before he died.	link graphic Photograph. Ian Tomlinson remonstrates with police after being pushed to the ground, minutes before he died.	Article: Ian Tomlinson The purpose of the image is to show a scene from the events where Ian Tomlinson was interacting with the police prior to his death. The caption describes the image and also contains additional information ("minutes before he died") that is not present in the image.

How is the "refer to caption" option an improvement on the "photograph" option? It advises the reader to do something he's about to do anyway (read/hear the caption). It implies the reader has come to the wrong place for information and needs to be referred elsewhere. It is tediously wordy and will just be annoying in an FA with a dozen similar photographs. I dread to think how tedious a FL with thirty photographs might be. Colin°^Talk 10:10, 9 June 2010 (UTC)

You dread to think, regarding a difference of two words? It was you added to the page that "All readers, when presented with the alt text, will be aware this element is an image. Therefore it is not necessary to contain words such as 'picture of'."

This page needs to be clear to people who've never heard of alt text and alt attribute and alt tag and blank alt text and missing alt text, and all the other terms you're throwing around and despairing because I'm not understanding. If I don't understand the page, there are lots of others who won't get it either. SlimVirgin ^{talk contribs} 10:19, 9 June 2010 (UTC)

No, it is a difference of three words. The "refer to caption" is three words that tell me nothing, indicate that I'm in the wrong place, and point me somewhere I'm about to go anyway. At least "Photograph" tells me something and doesn't imply there's something wrong with the article or the reader. There is a difference between "it is not necessary to contain the words such as 'picture of'" and actually outlawing such words. The point is that we have to fill the blank alt text attribute with something and it might as well be a minimally useful word than an annoying phrase. Colin°^Talk 12:46, 9 June 2010 (UTC)

Alt text vs alternative text

(responding to comment above) Actually, the article was quite clear on the difference between "alternative text" and "alt text" until this edit which conflated the two in the lead sentence. The word "alt" when referring to the attribute, was always formatted in code style (at least that was the plan, perhaps some escaped formatting). The guideline used that word exclusively for a wikimarkup attribute, as do our best sources. Yes this page needs to be clear on the meaning of terms. Perhaps before you make the guideline any more confusing, we might agree on them on the talk page, and you might consider reverting that particular terminology mistake. I'm not "throwing around" terms. I'm using the official W3C terms. What terms are you using? Colin°^Talk 10:32, 9 June 2010 (UTC)

Further review of the above edit (with summary "rewrote first few lines and nutshell to reflect source"). The existing text actually matched the source (the W3C definition of "alternative text") and the edit replaced it with the WebAIM source and text based on that source, using their terminology. I resent the implication that the existing text didn't reflect the source.

IMO the WebAIM terminology ("content and function") is weaker than the W3C terminology ("same information" / "equivalent purpose"). If you are interested in why, I'll explain, but I suspect I'm wasting my breath here.

The lead now completely confuses the distinction between "alt text" the attribute value and "alternative text" the concept. I'm seeing carefully chosen words and correctly used terms being ripped up and replaced by a confused mess. The article used to have a flow where each section had a purpose and introduced concepts before using them, described problems before offering solutions. My attempts to explain what was wrong with certain edits just gets ignored; they are reinserted without discussion.

Is there any point in me watching or contributing to this page? Colin°^Talk 12:46, 9 June 2010 (UTC)

Colin, you're not writing for people with technical knowledge, but for everyone. There are two problems with it: (a) it's unclear to anyone who doesn't already understand the issues; and (2) it doesn't seem to follow what the sources are saying, or at least not obviously so.

This version is better than the previous version, because it's shorter, and because it's not recommending the long screeds of alt text we were previously expected to write. But it has still has the essential problem of large chunks of it being inaccessible, and it's not clear what the structure of it is. There's no point in voicing your frustration at my lack of technical knowledge. I could similarly voice frustration that people with technical knowledge are sometimes unable to explain it in a way others will understand. What I suggest is, instead of us tearing each other's hair out, we try to combine our strengths. SlimVirgin ^{talk contribs} 15:00, 9 June 2010 (UTC)

If you are prepared to work with me to discuss how to improve this text then yes we can start again. If you are just going to edit the text where you completely ignore my comments then I might as well not bother. Nobody wants our articles and guidelines to be accessible to non-technical people more than me. I'm prepared to accept the text may have confusing bits (it certainly does now) and welcome comments about which bits. The problem with your last edit is that it is now wrong. Confusing is bad; wrong is worse. How can I fix this wrongness if you don't accept you've got "alt text" and "alternative text" completely muddled in the edit you made. How can we properly explain these terms if we merge them together when all other sources keep them apart? We might both be misunderstanding each other and we might both have some concepts wrong in our heads. I'd rather be wrong on the talk page and get that corrected, than be wrong on the guideline page and screw it up. I'd like you to stop editing and slow down. The words I used and the format of the guideline were carefully planned. I'd like to see the same care taken over any changes to the word use or the guideline structure. So far, I'm just seeing ad-hoc undiscussed edits that don't seem to appreciate or respect the existing text or structure. You could ask me what I meant by X, or query if X matches the source, rather than just chuck X out and write something else instead. I'd like to be able to suggest a change or disagree with a change and for you to show me some respect by responding to it honestly. You're welcome to do the same, of course, and expect the same from me.

Is this possible? If it is, can we try to get some agreement on this alt/alternative issue. Colin°^Talk 15:37, 9 June 2010 (UTC)

First paragraph

Let's go through it bit by bit, starting with the first paragraph. Can you say exactly what is wrong with it? It seems very clear to me, and it's sourced to WebAim.

Alternative text (alt text) is a textual alternative to images. The point of alt text is to allow the content and function of an image to be understood by text-only readers. It is displayed by browers when images are not displayed, is read out loud by screen readers for those with visual impairment, and is used by search engines to determine the content of an image.

• Source. SlimVirgin ^{talk contribs} 15:45, 9 June 2010 (UTC)

And if your objection is that alternative text and alt text are not the same thing, please link to a source that says that, so I can read it for myself. SlimVirgin ^{talk contribs} 15:49, 9 June 2010 (UTC)

There are two concepts we discuss. The first is the requirement to provide a text alternative for images. We agree that requirement can be met in various ways. IMO this is the guideline topic and what I want the lead paragraph to concentrate on. The second concept is the text one actually writes for the "alt=" parameter/attribute/thingy in an image element in our wiki markup. That text is only one of the means we have to satisfy the requirement to provide alternative text for images. We can also use the caption and the body text. Indeed, in many HTML situations, the "alt=" attribute text would be empty yet the alternative text requirement is still met by text elsewhere, or less commonly, no text was required because the image was just decoration.

The WebAIM guideline consistently uses the terms "alternative text" for the wider requirement-concept; "alt attribute" for the HTML markup element and "alt text" for the text within that markup element (though it does conflate the latter two sometimes). It also consistently uses a code font for the alt attribute term, to distinguish it from the alternative text concept.

The W3C guideline generally uses the phrase "text alternative" and less commonly "alternative text" to mean the concept/requirement. When discussing how to meet this requirement here it uses the terms "short text alternative technique" and "long text alternative technique". It presents a number of ways implementing those techniques. In particular, the G94: Providing short text alternative for non-text content that serves the same purpose and presents the same information as the non-text content is of most interest to us. Ultimately, when you burrow down through all the links, it describes how to use the HTML alt attribute for the short text alternative, as one means of doing so. The W3C requirements pages aren't particularly easy to navigate. However, I believe they have been very careful and precise with their words. I believe the above title: "that serves the same purpose and presents the same information as the non-text content" is the best definition of alternative text you will find on the web. It is better than WebAIM's definition. But I'll deal with that later.

I'm sure there are some carelessly written sources that conflate alt text with alternative text but the best ones don't. We should be similarly careful and precise. Do you think saying "text alternative" rather than "alternative text" would help resolve some confusion? Do you agree with me that the abbreviation alt should be reserved for markup attributes and presented in a code font like WebAIM do? Colin°^Talk 18:44, 9 June 2010 (UTC)

I can't see anything in those sources that distinguishes "alt text" from "alternative text." WebAim seems to use it to mean the same thing: " ... it is worth noting that additional information is often injected into alt text to provide additional information or to provide additional keywords to search engines. Such practices are not an appropriate use of alternative text." Can you quote here from a source that indicates the terms are being used to mean different things? SlimVirgin ^{talk contribs} 00:13, 10 June 2010 (UTC)

We learn what words mean from their usage. I've explained how both texts use the words and that they use them differently. If you want to think mathematically, the text within the alt attribute is a subset of the text that constitutes "alternative text". Therefore the WebAim quote you give makes sense. The alt attribute text is part of the alternative text for the image and must obey the requirements for a text alternative. However, the alternative text is not equal to the alt attribute text. The caption on the other hand, might satisfy the need for a text alternative but isn't so constrained by the rules.

Let's try a different approach. Open your dictionary and look up "alt". It is either a musical term or the angular distance above the horizon. The usage we see here derives solely from the HTML authors deciding to name their attribute with the abbreviation "alt". When you write your FAs, you don't use abbreviations; it isn't a professional thing to do. Similarly, when careful sources talk about providing a text alternative for an image, they don't use the abbreviation alt. Both WebAIM and W3C only use the word alt when referring to the HTML attribute.

Can we agree there are two concepts here? There's an HTML attribute (or Wikimarkup attribute) and there's a need to provide a text alternative for images. Can we agree that it is important to distinguish them. For example, if the text alternative is satisfied by our caption (which it often is) then it makes no sense to say "It is displayed by browers when images are not displayed, is read out loud by screen readers for those with visual impairment,..." That's just wrong and confusing. Can we agree that one way of distinguish them, the way used by WebAIM and others, is to never use the word "alt" as an abbreviation for "alternative text" and only use that word in a code font to mean the Wiki attribute and any text you write for it. Colin°^Talk 08:15, 10 June 2010 (UTC)

I understand that alt text and alt attribute are not the same thing, but I can't see a third concept that distinguishes alt text from alternative text. I think you need to produce a source that either says, or shows clearly, that alt text is not the same as alternative text. SlimVirgin ^{talk|contribs} 04:56, 11 June 2010 (UTC)

I've tweaked the writing in a way that I hope addresses your concern:

Alternative text (alt text) is a textual alternative to images. The point of alt text is to allow the content and function of an image to be understood by text-only readers. When alt text is added after an image, such as |alt=A painting of Napoleon Bonaparte, that text is displayed by browers when images are not displayed, is read out loud by screen readers for those with visual impairment, and is used by search engines to determine the content of the image.

SlimVirgin ^{talk|contribs} 05:30, 11 June 2010 (UTC)

Our best sources do not use the abbreviation "alt" when discussing alternative text. They only use that abbreviation when discussing the HTML alt attribute or the text supplied for that attribute. If you continue to use the "alt" abbreviation when discussing alternative text, you will confuse our readers/editors. That I'm unable to get this simple concept across to you is a matter of despair to me. This used to be a guideline on "alternative text", with the issue of what one writes for the "alt attribute text" just an aspect of that guideline to be considered alongside the caption and the body text. Now it conflates the two and does so from the first few bolded words. I'm going to ask a Wikifriend to review what I've written and see where I'm causing confusion. Currently, this guideline is significantly worse than the version here for all the reasons I've given above. That version can be improved, but the edits made since then have not done so. Colin°^Talk 07:37, 11 June 2010 (UTC)

Comments

Colin has asked me to comment on this discussion. It is sad to see two of Wikipedia's finest contributors at loggerheads, but having spent most of yesterday and this morning trying to get heart of the matter I think I can understand why. This is the situation as I see it: Editors will come to this page because they need help with an article they are editing. They will want to find quickly solutions to problems with alternative text for images and then return to their work on "their" article. Other editors, those reviewing FACs for example, will need fuller guidance that includes explanations of the limitations of the alt attribute in wiki mark-up. Wiki software developers may read it with a view to improving accessibility.

I think the version Colin last edited while well-structured, comprehensive and accurately sourced was not helpful to editors looking for a quick solution and is difficult to understand in parts on first reading, particularly those lacking expertise, such as Colin's, in mark-up functionality. This is not helped by the attribute's being called "alt", which I think still is a source of confusion for readers. And I think SV's motivation to offer clear practical advice to readers seeking a quick solution is admirable.

I installed Fangs yesterday and read a few articles to gain a deeper understanding of the experience of those of us who use screen readers. I have to agree with Colin in that expression such as "refer to" or even worse "see" would be tedious in the extreme and could give the impression of laziness. (But, to try to get this into perspective, the most tedious of all was the references, which are read out for example as "This page link left bracket eighty-two right bracket"!)

I agree with SV in that you should "combine your strengths" but would add that you should acknowledge your weaknesses. The current article is not an improvement on the version last edited by Colin and errors have been introduced with regard to the function of the alt attribute and the expression "alternative text", which are different things. On the other hand, the earlier version is confusing to those seeking practical help. I think it has to be made clear:

1. That "alt" is not short for "alternative text"
2. Using the alt attribute is just one way of producing the required text.
3. If it is not "used", how it must be tweaked to stop the filename being read out.
4. When using the alt attribute that a form of words that just point the reader elsewhere is not acceptable.

Please don't fall out with me for saying but I think the emphasis should be on practical help and clear explanation, rather than a well-sourced article. Your strengths are expertise and clear writing. Your weaknesses are impatience, and an outsider reading the above might perceive a little arrogance too. Graham. Graham Colm (talk) 13:25, 13 June 2010 (UTC)

Graham, thanks for your feedback. First, there has been no conflation of "alt attribute" and "alt text" that I can see. Can you point out where the two have been confused?

I think perhaps here: "Blank alt text is rarely an acceptable choice on Wikipedia". This presumably means leaving the alt attribute blank? Graham Colm (talk)

It was Colin's version that discussed blank alt text, e.g. here. What is meant is a blank alt attribute i.e. an absence of alt text. SlimVirgin ^{talk|contribs} 15:27, 13 June 2010 (UTC)

The issue is whether "alternative text" is the same thing as "alt text," and I can find no evidence that they're not. I take your point about needing to be clear rather than well-sourced, but if we introduce distinctions no one else has mentioned, and that may not exist, we're not going to end up with a clearly written page of advice. If in fact "alternative text" and "alt text" are used by others to refer to different things, I'll be happy to help explain the distinction if we can find a source who does.

Yes, this confuses me. I think they mean exactly the same when the alt attribute is used to render the text, but not when the text can be seen as well as the image and not instead of it. Graham Colm (talk)

I think they always mean the same thing. If you look at Alt attribute, it discusses "alternative text (alt text)". I feel we're introducing unnecessary confusion. I can't see what benefit is derived from having three concepts: (a) alt attribute, (b) alt text, and (c) alternative text, especially when we only need two (alt attribute and alt text/alternative text).

I disagree, and think alt should only be used when referring to the attribute and the text used with it. Graham Colm (talk) 16:16, 13 June 2010 (UTC)

Can you say why? SlimVirgin ^{talk|contribs} 16:44, 13 June 2010 (UTC)

Yes, a footnote in the article makes a clear distinction The term alternative text, as used in this article, refers to the text equivalent for an image, regardless of where that text resides. It does not refer solely to the alt attribute of the image tag. Graham Colm (talk) 17:07, 13 June 2010 (UTC)

I understand that. It was me who added that footnote, and I've been trying to stress it throughout this. That footnote does not say or in any way imply that alt text is not an abbreviation of alternative text.

• Alt text = alternative text
• Alt text/alternative text does not equal alt attribute.

Two concepts, not three. If you think we need three concepts, what are the benefits? SlimVirgin ^{talk|contribs} 18:09, 13 June 2010 (UTC)

I'm just catching up and it will take me a while. I'm greatly indebted to GrahamColm for helping and for the considerable time he has spent getting up to speed with the issue. I agree with SV there are only two concepts. My point is that our best sources are careful never to use the word "alt" wrt the first (alternative text) and furthermore some use a code font for "alt" when referring to the attribute to further emphasise this is an HTML markup keyword, not an abbreviation of an English word. WebAIM use the term "alt attribute" for the HTML attribute and any text it contains (hence a missing alt attribute (no 'alt="..."') is different from empty alt text (i.e., 'alt=""') and they use "alt text" to mean the text supplied for the alt attribute. They use the full "alternative text" for the overall text requirement which may be met by a caption, say. W3C are similar, though they tend to say "text alternative" and only a very minor part of their work even mentions the "alt" HTML attribute. It is clear that writing "alt text" is ambiguous here on WP as various contributors to this page have used it as shorthand to mean the "alternative text" concept and others to mean the HTML (or Wiki) attribute text. Repeatedly asking for a source that explicitly says a ≠ b is as pointless to me as someone asking for a source that says London ≠ England. You won't find one but you can tell from the usage by careful writers that they are different things and you can tell from their own individual definitions that they are different things.

Can we please agree that some people find the use of "alt text" as an abbreviation for "alternative text" to be confusing and contrary to the usage in our best sources. There is no need, in professional writing, to abbreviate. Please can we be strict in only using "alternative text" for the guideline/concept. And agree that for the text within the "" of the Wiki image markup's alt attribute, we can use "alt attribute text". Colin°^Talk 08:19, 14 June 2010 (UTC)

Refer to

The other issue we're trying to decide is what to write when (as you put it) the alt attribute is not "used." We could say "photograph," or "illustration," or where appropriate "refer to caption." What is your suggestion? SlimVirgin ^{talk|contribs} 13:42, 13 June 2010 (UTC)

Just a word such as "photograph" or "drawing" would be better than "refer to caption", because the caption which will be read out straight away in any case. They don't have to refer to it. Graham Colm (talk) 14:19, 13 June 2010 (UTC)

I think we do suggest that. We can certainly make the suggestion stronger. SlimVirgin ^{talk|contribs} 15:27, 13 June 2010 (UTC)

The Lead gives "refer to" as an option when I think it should be discouraged. Graham Colm (talk) 16:16, 13 June 2010 (UTC)

I've removed it from the lead, and added to the caption section:

Where the caption is sufficiently descriptive or evocative of the image, or where it makes clear what the function of the image is, one option is to write |alt=photograph, |alt=drawing, or |alt=refer to caption. Where nearby text in the article performs the same function, it can be |alt=refer to adjacent text. Editors should use common sense and keep the alt text as tight as possible when the caption suffices, to prevent screen readers reading out unnecessary words.

Is that a compromise we can agree on?

SlimVirgin ^{talk|contribs} 18:41, 13 June 2010 (UTC)

Side by side

Placing the versions of the lead side by side for clarification:

Previous

Current

Alternative text is text that serves an equivalent purpose and provides the same essential information as an image in an article, and is intended for text-only users. It should preferably be short, and co-located with the image. If this is not practical, for example because the image presents a lot of information, then a very brief description of the image or overview of the information should be given and a longer text alternative offered elsewhere. Absent or unhelpful alternative text is a source of frustration for blind users of the Web. An image that is purely decorative (provides no information and serves only an aesthetic purpose) requires no alternative text. Ideally, such images would be ignored by screen readers and text-only browsers but in practice this is not generally possible on Wikipedia. A short text alternative is typically implemented on Wikipedia through a combination of the image caption and the image alt attribute. MediaWiki does not support HTML's longdesc attribute, so any lengthy text alternative must be offered within the article body text. Image wiki-markup should specify the alt attribute and supply it with carefully chosen text as the MediaWiki-generated alt text is unlikely to be optimal. Blank alt text is rarely an acceptable choice on Wikipedia, because almost all images are also links, and screen readers will read out a link filename if the HTML alt attribute is blank or missing. In this regard, Wikipedia must diverge from most alt-text web guidelines, where blank alt text is often recommended.

alt=Painting of Napoleon Bonaparte|The Emperor Napoleon in His Study at the Tuileries by Jacques-Louis David With the alt parameter empty, the screenreader would read out the file name: "link graphic slash Jacques hyphen Louis underscore David underscore zero one seven jay pee gee." Alternative text (alt text) is a textual alternative to images. The point of alt text is to allow the content and function of an image to be understood by text-only readers. Alt text can be specified in images with the "alt=" parameter: File:Jacques-Louis David 017.jpg |alt=Painting of Napoleon Bonaparte| The Emperor Napoleon in His Study at the Tuileries by Jacques-Louis David The words after "alt=" ("Painting of Napoleon Bonaparte") are displayed by browsers when images are switched off, read out loud by screen readers for those with visual impairment, and used by search engines to determine the content of the image. Absent or unhelpful alt text is a source of frustration for blind users of the Web. Alt text should be brief. Where an image presents a lot of information, a short description in the alt parameter can accompany a longer description in the caption or body of the article, where necessary. The alt parameter should not be left empty, because almost all images are also links, and if the parameter is empty screen readers will read out the link filename, such as (for the Napoleon example) "link graphic slash Jacques hyphen Louis underscore David underscore zero one seven jay pee gee." An image that is purely decorative—for example in a template where the image is not a link—requires no alt text. Where the caption or nearby text are sufficiently descriptive or evocative, or where they make clear what the function of the image is, one option for the alt parameter is to write |alt=photograph or |alt=drawing.

I think I've more or less finished the copy edit and tightening. It's down from 1875 words to 1059 (the readable prose size doesn't count text in tables or blockquotes). There are a few things I'd like to tighten still, but I'm just about finished for now. SlimVirgin ^{talk|contribs} 09:07, 15 June 2010 (UTC)

Readers who have difficulty understanding visual images

We give as an example of the audience for alt text "readers who have difficulty understanding visual images." This is a separate category from readers with visual impairment. The source is given as this, but I can't find a reference to it there. What do we mean by "readers who have difficulty understanding visual images"? SlimVirgin ^{talk|contribs} 18:33, 13 June 2010 (UTC)

I'll remove this for now until we can think of an example or source. SlimVirgin ^{talk|contribs} 08:50, 15 June 2010 (UTC)

The queen

It is worth watchlisting this page just to see what the Queen is wearing each day. :-) Seriously, though... I wanted a simple picture of someone recognisable yet also not blatantly (it could be just an elderly lady in a yellow dress) -- yesterday's picture was too distractingly a queen with a crown. In today's picture, the crowd form a distracting aspect that one might want to include in the alternative text. The yellow dress picture was, I thought, perfect for the task. Locating the image on the left is often not ideal as it breaks up the layout of the text. I wonder if the fact she is not facing into the text, when positioned on the right, could be overlooked. After all, this is a guideline on alternative text, not image composition and layout. Colin°^Talk 08:45, 15 June 2010 (UTC)

True, but it's always jarring. Not that that matters much here, but my thinking was that an image advice page ought at least to follow the image guidelines itself. We don't have that many photographs of the Queen, which surprises me. SlimVirgin ^{talk|contribs} 08:47, 15 June 2010 (UTC)

Dannebrog, definition and rationales

I'm very glad to see the Dannebrog example back as an example with the original alt attribute text and rationale. This was one of the most discussed pictures on the talk page and is also a good example of basic alternative text rather than one of the "what to do in this situation" examples. The rationale highlights the "essential information" that the alternative text needs to get across.

Which brings us to one reason why I think the W3C's "serves an equivalent purpose and provides the same essential information" definition of alternative text is considerably better than WebAIM's "content and function". The "content" word all too easily lead the reader into thinking he has to describe the picture ("a flag fluttering on a flag pole..."). Not all content is information and even less is essential to the article. I think "purpose" is better than "function". The words can be interchangeable to some degree but the former implies an active choice by the editor rather than a passive ability/feature of the graphic. It is that difference that is absolutely key to helping the editor write good alternative text. Their thoughts should not be: This is an image that happens to be in the article that I happen to have to write alternative text for. It should be: This is an image that I (or someone else) chose to serve a purpose and provide information, and for which I should ensure is still possible even if the image is not available to the reader.

The rationales were written to show what the editor had actively considered when selecting the picture: "The purpose of the image is...", "a stock photograph chosen to decorate a sound bite ...", "The picture is used to show ...". Considering the "purpose" and then the "information" are two of the four questions that the W3C ask editors to consider for alternative text. We should ensure our example rationales make that thought process explicit. Currently, the Water Fluoridation rationale has been weakened by unnecessary trimming. Colin°^Talk 08:45, 15 June 2010 (UTC)

I've restored "This is a stock photograph chosen to decorate a sound bite from the article regarding tap water," but I don't really know what it means. SlimVirgin ^{talk|contribs} 08:56, 15 June 2010 (UTC)

Terms

Colin, could you discuss your edits here, please, rather than restoring some of the problematic material? I've tried to keep the terms we use as simple as possible, because it's aimed at readers with and without technical knowledge, but your version restored terms that introduce confusion in my view. For example, you use alternative text, alt parameter text, alt text, alt parameter, and alt attribute in ways that are not clear. First of all, please provide a source that shows alt text is not used to mean the same thing as alternative text. SlimVirgin ^{talk|contribs} 20:53, 22 June 2010 (UTC)

Don't you dare lecture me on collaborative editing or the need to discuss edits prior to making them. What cheek! Let me remind you that the text prior to your edits starting 26 May was the result of a careful written sandbox draft, which was then proposed and widely advertised to interested editors including yourself, which was reviewed by several editors, which was copyedited by Malleus and which had remained stable for over two months. Your edits since then have been almost entirely done without prior talk page comment and none of them reflect the fruits of consensual discussion.

Contrary to the above comment, the issue of alt/alternative terms has been discussed at great length. The discussion reached a stalemate so a third opinion was sought. Let me quote GrahamColm: ""alt" is not short for "alternative text"" and "I disagree, and think alt should only be used when referring to the attribute and the text used with it." So the issue of what terms to use has been discussed and you've lost the argument. The alt/alternative terminology confusion is one WebAIM make pains to clarify in a big bold "Important" text block in their guideline. And I've already repeatedly answered your question about a source: all our best sources keep "alt text" and "alternative text" to mean different things; we should do so too. Could you perhaps consider that your failure to get the point is not down to a failure on my part but possibly one on yours. As I have said before, you are welcome to invite another third party to help explain this issue. Colin°^Talk 07:59, 23 June 2010 (UTC)

The facts are that the terms "alternative text, alt parameter text, alt text, alt parameter, and alt attribute" need to be used because they are different things (though the second is just a longer way of saying the third). The WebAIM source makes it clear they reserve the term "alternative text" for the text alternative regardless of where it appears and distinguish their use of the abbreviation "alt" for the HTML markup attribute. The "alt=..." bit is called the "alt parameter" in Wiki markup and is called the "alt attribute" in HTML. The words "attribute" and "parameter" are generally synonymous in computer science but it is best to use the terminology used by others than to conflate them here. The web world (and WebAIM) are happy to use the term "alt text" for the text between the quotes in the alt attribute (the Wikimarkup doesn't require quotes). They are not happy to use the term "alt text" to mean alternative text regardless of where it appears. These things are all facts. They are verifiable from the sources given. Any attempt to combine them and just use the term "alt text" for all of them just confuses. It is an oversimplification. There is a difference, for example, between not having an alt parameter at all and having blank alt text (i.e., "alt=|"). There is a difference between having no alternative text and having no alt text. Technical writing requires precision and an appreciation of using the right words. Colin°^Talk 09:48, 25 June 2010 (UTC)

From WebAIM:

Examples of the use of "alt text" for the text for the HTML attribute. Note the use of brown code font for "alt".

• "Because the image is the only object within a link, null alt text is not appropriate."
• "Perhaps a better solution would be to have the text "Next Page" or similar adjacent to the image and within the link, in which case, the image could be given null alt text. "
• "If it were not within the link, then the alt text might be different."
• "In another context, it may be important that the user know that this image is indeed an icon - in such a case, using the word "icon" in the alt text may be appropriate."
• "Option 4 (null or empty alt text) would not provide the important information that the image presents. "
• "In almost all cases, spacer and decorative images should have null alt text (alt="")."
• "though it is worth noting that additional information is often injected into alt text to provide additional information or to provide additional keywords to search engines."

In many other cases they refer to the "alt attribute", because they are concerned with the "alt=""" bit, not just the text.

To distinguish that "alternative text" means something else, here's the big warning box on their guideline:

"Important

The term alternative text, as used in this article, refers to the text equivalent for an image, regardless of where that text resides. It does not refer solely to the alt attribute of the image tag. alt attribute will be used when referring to the attribute itself, which often will, but does not exclusively, contain the alternative text.

And they conclude with

• "Alternative text may be provided in the alt attribute or in the surrounding context of the image. "

They are saying the text may be provided "in the" attribute, in other words, it is between the quotes, or it may appear in a caption.

Colin°^Talk 16:46, 25 June 2010 (UTC)

Improved lead

Last night's revert by SlimVirgin removed two changes. One was an attempt to clarify the alt/alternative terms issue and is being discussed above. The second was a significantly improved lead. The baby has been rather thrown out with the bathwater here and I implore SlimVirgin to allow me to reinstall those improvements, minus the contentious issue of alt/alternative naming. Here are the reasons why the lead was improved in this edit:

• The new lead paragraph defines "alternative text" using the superior W3C definition. The reasons why the W3C definition is superior to the WebAIM one that SV replaced it with (without discussion) is given earlier on this talk page.
• The new lead paragraph does not fall into the trap of oversimplifying "alternative text" to just being the text in the alt attribute. This is an aspect WebAIM are careful to emphasise early in their guideline: "Alternative text can be presented in two ways: * Within the alt attribute of the img element. * Within the context or surroundings of the image itself." Indeed, I would say that (once we have established clearly what alternative text is), the sentence "On Wikipedia, alternative text is typically supplied through a combination of the image caption and the text supplied for the image alt parameter in the MediaWiki markup." is the most important sentence in this entire guideline. The lead prior to the edit last night (which was the result of undiscussed edits) did not make this point.
• The example image and the example Wiki markup now correspond and the image now actually contains the alt text. I find it remarkable that our lead image lacked alt text yet recent edits have fussed over choosing a picture of the Queen that faces the "right" way--an issue that is of far less importance than choosing a good picture for the point being made.
• The issue of what is read out by screen readers in the absence of alt text is mentioned just once in the lead rather than also in the caption. The confusing mix of markup and commentary in the caption is also eliminated. Note: the screen reader does not read out the file extension.
• The new lead makes it much clearer why screen readers read out the filename. The previous text did not make that clear, only offering the confusing "because almost all images are links" which only contains half the information the reader needs to understand the issue.
• The final paragraph in the new lead makes it much clearer why one might end up wanting blank alt text (because the image is decorative or because the caption supplies all the alternative text). It makes it clear which limited circumstances blank alt text is allowed (without a confusing comment about templates). It better explains the solutions.
• The prior lead text made the mistake "a short description in the alt parameter" when in fact a short summary of the information is what is required, not a description of the image. Subsequent changes make the same mistake: "The text should be brief. Where a longer description is needed" Again, we should not emphasise "description" as that is only rarely what is required of alternative text.

Colin°^Talk 07:59, 23 June 2010 (UTC)

Two leads side by side

SlimVirgin

Colin

Alternative text (alt text) is a textual alternative to images. The point of alt text is to allow the content and function of an image to be understood by text-only readers. Alt text can be specified in images with the "alt=" parameter: File:Jacques-Louis David 017.jpg |alt=Painting of Napoleon Bonaparte| The Emperor Napoleon in His Study at the Tuileries by Jacques-Louis David The words after "alt=" ("Painting of Napoleon Bonaparte") are displayed by browsers when images are switched off, read out loud by screen readers for those with visual impairment, and used by search engines to determine the content of the image. The text should be brief. Where a longer description is needed, a few words in the alt parameter can accompany a longer description in the caption or body of the article, where necessary. Absent or unhelpful alt text is a source of frustration for blind users of the Web. The alt parameter should not be left empty, because almost all images are links, and if the parameter is empty screen readers will read out the link filename, such as (for the Napoleon example) "link graphic slash Jacques hyphen Louis underscore David underscore zero one seven jay pee gee." An image that is purely decorative—for example in a template where the image is not a link—requires no alt text. Where the caption or nearby text are sufficiently descriptive or evocative, or where they make clear what the function of the image is, one option for the alt parameter is to write |alt=photograph or |alt=drawing.

Alternative text is text associated with an image that serves the same purpose and conveys the same essential information as the image. In situations where the image is not available to the reader (perhaps because they have turned off images in their web browser, or are using a screen reader due to a visual impairment) the alternative text ensures no information or functionality is lost. Absent or unhelpful alternative text is a source of frustration for blind users of the Web. On Wikipedia, alternative text is typically supplied through a combination of the image caption and the text supplied for the image alt parameter in the MediaWiki markup. The following example produces the image on the right: [[File:Jacques-Louis David 017.jpg |thumb |160px |alt=Painting of Napoleon Bonaparte |[[The Emperor Napoleon in His Study at the Tuileries]] by [[Jacques-Louis David]].]] The alt parameter text ("Painting of Napoleon Bonaparte") is not normally visible to readers but may be displayed by web browsers when images are switched off, is read out loud by screen readers for those with visual impairment, and can be used by search engines to determine the content of the image. In keeping with other Internet guidelines, the term "alt text" (in a code font) is used here to refer to the text supplied for the image alt parameter and which generates text for the HTML alt attribute; the term "alternative text" refers to the text equivalent for an image, regardless of where that text resides. For images that link to their image description page (which is nearly all images on Wikipedia), the alt text cannot be blank nor should the alt parameter be absent. This is because a screen reader, in order to describe the purpose of the link, will default to reading out the image filename when no alt text is available. This is usually not helpful: in the above Napoleon example the screen reader would have read out "link graphic slash Jacques hyphen Louis underscore David underscore zero one seven" had we not supplied the alt parameter. An image that is purely decorative (provides no information and serves only an aesthetic purpose) requires no alternative text. Often the caption fully meets the requirements for alternative text. However, the only situation where blank alt text is acceptable is where such images are unlinked, which is rarely possible. One solution is to provide something at least minimally useful such as |alt=photograph , |alt=painting, or |alt=sculpture. Another solution, if the caption doesn't already describe or identify the image, is for the alt text to do so as briefly as possible.

SlimVirgin ^{talk|contribs} 08:05, 23 June 2010 (UTC)

• Given the absence of comments from other editors and any further discussion from SlimVirgin, I intend to restore Colin's version, which is better IMHO. Graham Colm (talk) 22:27, 24 June 2010 (UTC)

Graham, could you explain why, please? It seems self-evident to me that my lead is clearer, as I believe the rest is too. I could be wrong, of course! But could you explain your reasoning, starting with the lead? SlimVirgin ^{talk|contribs} 00:13, 25 June 2010 (UTC)

Sorry for the delay in replying SV. I think Colin's version is written better, not massively of course, but noticeably. I particularly don't like "alt text" used as shorthand, it looks lazy and is inaccurate for the reason I give below and have given before. Having said that, there is some unhelpful use of technical words in Colin's version, which could be simplified, MediWiki markup for example. But it is clearer in Colin's version that the "alternative text" is not just what is written after alt, it can be elsewhere, say in the caption. Colin's version is more in keeping with the recognised guidelines and more clearly explains why alt text cannot be blank and/or the alt parameter absent and why this is particularly the case on Wikipedia. It is difficult to be precise but SV's version reads like a quick "How to do it", whereas Colin's is more comprehensive, better structured and professionally written IMHO. I know that this is not an adopted guideline yet, but I would like to see it as well-written as our best articles—and Colin's version is closer. Graham Colm (talk) 17:32, 27 June 2010 (UTC)

I find Colin's version wordy and confusing. I'll work on my version further to clarify the point about alternative and alt, but it's important to avoid jargon as far as possible. SlimVirgin ^{talk|contribs} 19:14, 27 June 2010 (UTC)

Would you be happy to discuss any proposed changes here first? Graham Colm (talk) 19:58, 27 June 2010 (UTC)

Graham, normally yes, of course I would. What concerns me here is that there's an attempt to lock the page against further editing. I've been liberally insulted, asked how dare I do x, y, and z, accused of dishonesty, people are being emailed about me, there's been a request for pre-emptive page protection, a request for admin action on AN/I, and various people asked to come here to comment. I'm finding it all somewhat disturbing!

My aims are: (a) to have the writing as tight and clear as possible; (b) to have the page as jargon-free as possible because it's aimed at people with no technical knowledge for the most part; and (c) to keep the advice light, in the sense of realistic and manageable. My feeling is that the current version doesn't achieve the first two. It does achieve (c), but the way it's written is obscuring that. So I would find it a lot easier to work with the tighter version, and build things up from there. SlimVirgin ^{talk|contribs} 20:22, 27 June 2010 (UTC)

Given the problems we are having, I think it would be for the best to work with the current version and work towards your goals following discussion here. Reverting will not help; was it you or Colin who reminded us of keeping the baby in the bath? Graham Colm (talk) 20:42, 27 June 2010 (UTC)

It was you and Colin who reverted the copy-editing, and you did it because he asked you for support. [7] It isn't helpful to support this kind of situation, Graham. The way editing progresses, especially with guideline development, is that people try to build on the previous editor's work, as I did when I took Colin's version and tightened it, instead of starting from scratch as I'd prefer to. Two other editors have said the current version is wordy with too much jargon. I agree with them, and there's no reason to halt editing just because Colin is having the heebie-jeebies for some reason. Sorry, but this is getting silly.

I'm normally able to get policies and guidelines in good shape in terms of the writing. Therefore, as I asked before, please join me in doing that to make sure I make no technical mistakes as I'm copy-editing. Let's combine skills. SlimVirgin ^{talk|contribs} 21:01, 27 June 2010 (UTC)

Colin asked for my opinion, which I gave and it was critical of "his" version too. He did not ask for my support and never has. OK, you may have superior skills than me at guideline writing, but this requires Colin's expertise, not mine. I am not sure what you mean by the "kind of situation" I am supporting. And, please let's not use expressions like "heebie-jeebies"; they don't help. I suggest you try to work out a way of collaborating constructively with Colin. Graham Colm (talk) 21:24, 27 June 2010 (UTC)

Graham is welcome to explain why he thinks it is better, and to contribute his own changes too. But SlimVirgin, your comments expose the wrong mindset you've taken. I've written thousands of words about why my version is better and what is wrong with all the changes you've made. You haven't responded to any of that to any meaningful degree. Why should we think you'd value Graham's thoughts either? It takes two to have a "discussion" and you request that things be "discussed" yet you contribute nothing to the discussion other than your own brief opinion that your version is better and that you don't understand the terminology. Your attitude is that we should accept your changes as given and refuse to defend them when challenged other than to reinstate them with a revert. You ask us to explain ourselves yet give no response when we do. Rather, it is you that should be explaining all your recent edits and justifying them.

You've taken text that was reviewed by several people, and copyedited by a fine writer, and chucked it in the blender. Your writing isn't just poor, it is wrong. The previous text did have problems, I can see some of them now and would like to fix those. But under the guise of "tightening" you've hacked this guideline so much that it no longer makes sense. Only someone who already understood the points could possibly fathom what it is trying to say. The alt/alternative terminology aspect is a clear example of where your limited understanding is contributing to the damage and I ask you to consider whether editing a guideline from a position of confusion is actually helping Wikipedia. It is one thing to read about a subject to a degree that one can understand and apply it. It is quite another to understand a subject sufficiently well that you can then write about it and explain it to others. I'm not claiming I'm completely able to do that but the wonderful thing about Wikipedia is that editors can collaborate and learn from each other and together build something better than any one person could. Your editing and talk page approach on this guideline just exposes all that can go wrong with Wikipedia.

It is clear you don't wish to follow a BRD style (more of a BBBRBBBR style). Eubulides used to follow a propose...discuss...revise-proposal...discuss...install style and this worked well for the highly contentious articles he worked on. I'd be happy to work that way. I'd be very happy to see more editors here to mediate such a style and contribute. I want this wannabe-guideline to be understood by everyone, including yourself, and to be shown to an expert again for approval. But this needs a collaborative approach not one where one editor dominates and who bullies others. Colin°^Talk 09:06, 25 June 2010 (UTC)

Colin, I've asked you several times to provide a source that says or shows that alternative text and alt text are not the same thing, and you've responded with insults. There's no point in continuing to discuss it until you provide that source.

I would like to know what Graham's reasoning is, or I'll be restoring the text that I feel is clearer and tighter. And I don't appreciate your POV description, followed by an apparent apology followed by more insults. I would just as soon not swap any more personal opinions with you, but stick to sources instead; keep the page's terminology consistent; keep the writing tight; and keep the page as jargon-free as possible. SlimVirgin ^{talk|contribs}

I've repeatedly told you that that source request is as pointless as requesting a source that explicitly says London and England are not the same thing. They each have different definitions and they are each used for different things by good writing. You have not provided any reason why my definitions and terminology usage are incorrect or inconsistent with correct practice. Your merging of those terms into one abbreviation is unexplained and indefensible. This is an example of where your writing is just wrong. If you want, I can provide examples of where it no longer makes sense and of where the writing is flabby (it certainly isn't what a copyeditor would regard as tight). The lead sentence is particularly weak. How you can regard your lead as self-evidently superior is beyond me. That it is shorter is the only self-evident quality it has, other than that it has "Slim Virgin" above it. Oh, I get it now....

Please do not enter an edit war with Graham. I've given you plenty reasons why the version he restored is superior and you have not engaged with me on any of them. If you want to have a grown up discussion on text please start by responding to the comments I've made about your lead. If you just want to be a bully and revert without providing any defence of your text then I shall have to seek measures. Page protection and formal mediation seem one course. Sanctions against you another. Colin°^Talk 16:16, 25 June 2010 (UTC)

I have tried yet again to explain, with examples, how our best sources use the terms. See the section above. Colin°^Talk 16:48, 25 June 2010 (UTC)

(from SV's talk) I'm giving you notice that if you carry out your threat to revert Graham's edit, continue to make such threats, or perform similar edits or reverts, then I shall report you for edit warring. I already regard your edits made 9th June as edit warring. Your revert of my edit 22nd June was unjustified and not at all in keeping with our guidelines on reverting. You have twice made accusations against me of making edits without discussion. Those accusations are entirely without foundation and I find them deeply offensive. They are all the more galling because that is precisely the characteristic of your edits that is causing all the trouble. Colin°^Talk 17:55, 25 June 2010 (UTC)

Just to make it clear: the above paragraph was not posted or copied here by me. SlimVirgin removed this notice from her talk page and inserted it here. I am required to post a notice on her talk page prior to making a formal complaint. Colin°^Talk 18:35, 25 June 2010 (UTC)

Forgive me interjecting myself, I came here after seeing Colin's request at WP:RFPP. I wanted to jump into the debate because when I read the first the few paragraphs of this version I cringed at the depressingly familiar sight of an editing guideline that should be for everyone but is actually written as if for a small subset of our editors. The language in the version I linked is far too technical and requires then non-tech-savvy reader to follow a bunch of links to understand it. If a techy section is needed, one could be added later on but the first few paragraphs need to be written for a general audience. Here's a list of totally unsuitable vocabulary in the current lead paragraphs:

• alt parameter
• Mediawiki
• markup
• code font
• HTML
• alt attribute

All those need to go from the lead section. The version in question is also excessively wordy (reads as if written by committee, which it probably effectively has been) and could stand some trimming. It would also be a good idea to juxtapose the image of a Napoleon with a demonstration of what users that do cannot display the image will see and an audio clip of what a screen-reader would produce. CIreland (talk) 16:34, 26 June 2010 (UTC)

Thank you for the feed back. I'm afraid "alt parameter" is needed (it is in Slim's version too) as there's no getting away from the fact that you've got to write "alt=" somewhere. We can get rid of the others. Here's what I propose:

On Wikipedia, alternative text is typically supplied through a combination of the image caption and the text supplied for the image alt parameter in the MediaWiki markup.

be replaced with

On Wikipedia, alternative text is typically supplied through a combination of the image caption and the text supplied for the image alt parameter.

and

the term "alt text" (in a code font) is used here to refer to the text supplied for the image alt parameter and which generates text for the HTML alt attribute

be replaced with

the term "alt text" is used here to refer to the text in the image alt parameter

Does that help? The example of showing what people see without the image was present in this version (the Appearance section). Would you like something like that here? We can worry about getting a copyeditor to refine the text once we agree on what it needs to say. Colin°^Talk 16:52, 26 June 2010 (UTC)

That seems along the right lines, but it's tricky judge without seeing it in context. I'm not sure why "alt parameter is needed" - SlimVirgin's version above has:

Alt text can be specified in images with the "alt=" parameter:

How about if that read:

Alt text can be specified in images by adding "alt=" followed by the alternative text. Example:

CIreland (talk) 17:00, 26 June 2010 (UTC)

Well SlimVirgin's confuses "alt text" and "alternative text" so it isn't usable as a base for that reason (and see other sections if you want to comment on that issue). I agree we could avoid "alt parameter" in this instance, though in the body it becomes hard to discuss something if you can't give it a name. That's its name. Colin°^Talk 17:37, 26 June 2010 (UTC)

CIreland, thank you for this input. The point I've been trying to make is that if the page is ever to become a guideline it needs to be clear to non-technical people—and that aside it needs to be much less wordy. This is the version I'd like to develop. If you have any comments about it (and in particular whether there are technical errors in it), that would be much appreciated. SlimVirgin ^{talk|contribs} 16:59, 26 June 2010 (UTC)

I agree it needs to be clear to non-technical people but there are ways of achieving that that don't involve edit warring. It also needs to be technically correct. SlimVirgin, I'm sure you'd like to work on your version but all the things that are wrong or inferior about it are listed at the top of this section. You could start by responding to those. Yes there are technical errors in it. There's the wrong terminology. And it contains oversimplifications. It contains a poor definition and misses out the key aspect that the caption and the alt text are both useful ways of supplying alternative text. The wordiness is the least of our problems at present. But if you wish to suggest (in the manner I did above) how to rewrite certain sentences for brevity, then we can all review it and judge it on its merits. What I want to stop now is any edits made without prior discussion and attempts to establish consensus. That's Wikipedia policy for goodness sake. And it is absolutely required where editors are in disagreement. Further reverts will be regarded as edit warring and reported as such. Colin°^Talk 17:37, 26 June 2010 (UTC)

Colin, you're the one who reverted the recent edits, [8] but now you want to stop everyone else engaging in what you call edit warring when it's not you doing it.

I find your request for pre-emptive page protection, and your complaint on AN/I because no admin would go along with it somewhat bizarre, as was your failure to let me know you were discussing me on AN/I. As it comes on the heels of a fair number of insults from you that I've tried not to respond to, I feel it's best if we keep interaction to a bare minimum. For now, I would prefer to see whether others comment on the content. SlimVirgin ^{talk|contribs} 18:39, 26 June 2010 (UTC)

I have made one edit to this page that can be regarded as a revert. this one. It was utterly in keeping with WP:BRD and was followed up by an extensive attempt to engage you in discussion. Your subsequent revert of that revert (9th June edits) is edit warring. When your edits have been contested you are required to engage in discussion and attempt to seek consensus before re-applying them or anything similar. You did not. That's edit warring. My only subsequent edit to this page was 22nd june which was the result of extensive discussions on terminology and which ultimately involved a third party who decided against your use of terminology and in favour of mine. That edit was also accompanied by a detailed rationale of why it improved on the lead. That rationale is the basis for discussing any issues you might have with the edit. You have not engaged in that discussion but instead reverted it with the rationale that yours was "self evidently better". You are now threatening to re-revert once more. I'd like to use 4th June's version as a basis on which to continue but I'm not daft enough to think you'd let me. This guideline will not move forward by folk reverting all the time.

I did not discuss you at ANI so had no reason to inform you. I didn't go to ANI "because no admin would go along with it" but because I was advised to do that by the the newbie admin who declined the RFPP. I would really appreciate if you wouldn't spread disinformation about me. His reason for declining RFPP was shown to be incorrect and the discussion closed. It is that admin I informed as a necessary courtesy. If you want to stick to discussing the content of this page, you will find the insults disappear along with the comments on your outrageous behaviour. But it is you that is prolonging the agony by brining up issues like ANI that don't belong on this page. Are you going to respond to the comments at the top of "Improved lead" or not. If you refuse to I shall open an RFC on your refusal to discuss content in a content dispute. Colin°^Talk 18:55, 26 June 2010 (UTC)

Here are your reverts. [9] [10] I mind less about the reverting, but I do mind about the insults and your argument that when you revert, it's part of the normal editing process, but when someone else does, they're edit-warring. I see the threats continue, so as I said above I'm going to disengage, and wait for others to comment. SlimVirgin ^{talk|contribs} 19:31, 26 June 2010 (UTC)

Only the first is a revert. The R in WP:BRD. And I made effort to salvage what I could in keeping with our revert policy. The second was the result of extensive discussion yet your rationale for reverting that second edit was that I hadn't discussed it. You can insult me all you like if you like. It is the disinformation you spread that offends me. Colin°^Talk 19:55, 26 June 2010 (UTC)

Proposal 1

That the changes proposed (16:52, 26 June 2010) by Colin be applied to the lead in an effort to remove unnecessary jargon. I think we can re-word to remove the first two and the fifth use of "alt parameter". Removing the other two will be harder. I'm running out of time just now. Back later. Colin°^Talk 17:48, 26 June 2010 (UTC)

I object to that. I find your version unnecessarily wordy and confusing, and I'd prefer to use this version, which you reverted, as a base. I would like to hear from Graham before proceeding. SlimVirgin ^{talk|contribs} 18:29, 26 June 2010 (UTC)

That is not a valid objection. The way we move forward is we discuss whether the proposed changes improve the text or not. If you want to make a separate proposal that the text be reverted to some prior version then you can do that. I can see we will need a professional mediator if this sort of behaviour continues. The first thing the mediator will remind you is that when you discuss proposals you discuss the text, not the person. The only textual aspect of your comment did not refer to the proposal but a general comment on the lead. Please do not describe my edit 22 June as a revert. It bears no resemblance to any previous version. Spreading disinformation is not helping. Colin°^Talk 18:57, 26 June 2010 (UTC)

FWIW, and despite my remarks below about the standards, this is Wikipedia. I don't think it's necessary to go into as much detail about alt tags etc as you would for web developers. At the same time, I do think the current Wikipedia standard is a bit minimal. The alternative text (wherever placed - it can be in the caption, the alt text or the longtext) for Napoleon, if one were following the AA (replaced by) Priority 1 (A) standard, should be "painting by Carlsberg of Napoleon Bonepart in 1800-frozen to death, at age 106, showing him in the uniform of a French chasseur du cheval and wearing the Blue Star of the Republic. The subject is holding a small dog under his left arm, symbolic of his desire to subdue the nations, and is shown standing in front of a pair of navy blue velvet curtains, which hide a nasty stain on the wall." --Elen of the Roads (talk) 20:32, 26 June 2010 (UTC)

Please note the use of "alternative text" and "alt text" in the above comment, and the vital importance of saying it can also be in the caption (missing from the previous lead). Ultimately this is a guideline that when followed will involve editors having to place "alt=..." in the correct place, so they need to learn some markup. Perhaps I need to repeat the proposal diffs and to say this is just one step. It is a step that is removing some of the detail about HTML tags and other jargon. I'm not saying that after this proposed change it is perfect.

The way this works is that any editor can propose a small change to the text. That change, and only that change, is discussed on its merits. Editors are barred from discussing other editors. Only the text and the changes matter. The change or some revision of it is then applied and the process repeats for another proposal. This can work if people stick to the rules. I'm interested in your "AA standard" which I'm not aware of. This seems closer to the old alt text guideline. Perhaps we should continue that aspect in another section. Colin°^Talk 20:54, 26 June 2010 (UTC)

The AA standard is Priority 2 (see here) However, my bad, the requirements for text are in Priority 1 - I hit the A key twice.

Provide a text equivalent for every non-text element (e.g., via "alt", "longdesc", or in element content). This includes: images, graphical representations of text (including symbols), image map regions, animations (e.g., animated GIFs), applets and programmatic objects, ascii art, frames, scripts, images used as list bullets, spacers, graphical buttons, sounds (played with or without user interaction), stand-alone audio files, audio tracks of video, and video. --Elen of the Roads (talk) 21:20, 26 June 2010 (UTC)

The long and the short

As someone who manages a UK local authority website - which is required to conform to the AA standard (and frequently doesn't! - we are ranked continually [11]), I'd add the following notes.

There is a difference between alt and the more general notion of alternative text. this is the W3 guidance on the subject. The alt tag needs to be used on all images: - icons, buttons, spacers (we lost 50 places in the ranking once when someone put up some pages with a load of white rectangles with no alt tags used as spacers). The standard requires that an image has an alt text and an accompanying text which allows the users to access the information conveyed by the image. In some cases this can be extensive - audio and video files require transcripts. Graphs and charts require a paragraph explaining what the chart shows. If one were using an image to - say - illustrate the shape of a Danish axe, it would require a paragraph describing the shape.

Now of course Wikipedia isn't held to any standard in the way that UK public sector websites are, so it can set its own standards. I think what's happening here isn't primarily a matter of clarity. It's that the two editors concerned are trying to write text that explains two different standards. --Elen of the Roads (talk) 14:55, 26 June 2010 (UTC)

I don't quite follow your "two editors concerned are trying to write text that explains two different standards". However, I'm pleased to see a definition on the W3C page you linked:

• When using IMG, specify a short text equivalent with the "alt" attribute. Note. The value of this attribute is referred to as "alt-text".

The linked page is for the old 1.0 version of the guideline. It would be nice to use the 2.0 version cited by our page. I'll look later to see if it has an equivalent definition. When one reads this definition of "alt-text" and reads the definition of "alternative text" cited by our page, one can't help but notice they are different things. *sigh*. Colin°^Talk 15:57, 26 June 2010 (UTC)

See if you can find the 2.0 version - I just pulled out what I was familar with. Something more up to date may be more help. What I mean by two different standards is just that. Going back in time, Wikipedia had to be rather bullied into making any concessions to accessibility (or to using vulgar html), and it came up with an initially fairly minimalist guideline which of course applied only to itself. You (and I) are referring to an international standard which is very widely used, but which wasn't developed by Wikipedians. This is an essential thing to understand - Wikipedia has its own rules for everything, and they are only coincidentally the same as outside rules.--Elen of the Roads (talk) 18:41, 26 June 2010 (UTC)

The WCAG 2.0 guidelines on images are here.--Elen of the Roads (talk) 21:35, 26 June 2010 (UTC)

I think the text from here could be usefully modified to incorporate into the guideline, as (given the examples above with the picture of Thompson) I don't think we've quite grasped it yet.

a text alternative serves the same purpose and presents the same information as the original non-text content. As a result, it is possible to remove the non-text content and replace it with the text alternative and no functionality or information would be lost. This text alternative should not necessarily describe the non-text content. It should serve the same purpose and convey the same information. This may sometimes result in a text alternative that looks like a description of the non-text content. But this would only be true if that was the best way to serve the same purpose.

If possible, the short text alternative should completely convey the purpose and information. If it is not possible to do this in a short phrase or sentence, then the short text alternative should provide a brief overview of the information. A long text alternative would be used in addition to convey the full information.

The text alternative should be able to substitute for the non-text content. If the non-text content were removed from the page and substituted with the text, the page would still provide the same function and information. The text alternative would be brief but as informative as possible.

In deciding what text to include in the alternative, it is a often a good idea to consider the following questions:

• Why is this non-text content here?
• What information is it presenting?
• What purpose does it fulfill?
• If I could not use the non-text content, what words would I use to convey the same function and/or information?

When non-text content contains words that are important to understanding the content, the alt text should include those words. If the text in the image is more than can fit in a short text alternative then it should be described in the short text alternative and a long text alternative should be provided as well with the complete text.'' Elen of the Roads (talk) 21:39, 26 June 2010 (UTC)

I am pleased to note that much of the above is remarkably similar to my lead paragraph and the "following questions" is in the body text. The 4th June version used some of the "short text alternative" / "long text alternative" language used by W3C. I now think that introducing those concepts is probably not useful here as it is just more jargon to learn. In particular, some of the "long alternative" options (using longdesc or using a page linked to from the image) aren't suitable for Wikipedia. Colin°^Talk 22:40, 26 June 2010 (UTC)

I agree about the short/long text. Ideally, the aim is to give enough text to fulfil the goal of providing alternative text at all. That's where the philosophical discussion lies. The other should be very simple, and more instructional - use this template to add images, always add an alt= component, always add a caption, if neither of those convey enough information for someone who cannot see the image (eg with a map or chart), refer to the section of text that explains what is in the image--Elen of the Roads (talk) 15:39, 27 June 2010 (UTC)

What has surprised some of us is the remarkable number of images on Wikipedia that are really just decorative. Or where the caption has very little to do with the image. Take the Water Fluoridation example in this page. The picture of Napoleon on a list of great world leaders is there because the author of the list wanted to decorate it, not because we need to know what Napoleon looked like or the style of artistry used in the portrait. I agree that diagrams and charts are difficult cases and one thing reading this guideline might remind editors is when they are relying on the image too much. The caption does often fulfil the alternative text needs but editors can't always add a caption if the image is used in a table or some other layout scheme, which are common in our featured lists but less common in articles. Colin°^Talk 16:25, 27 June 2010 (UTC)

Yes, there do seem to be a lot of galleries, and pictures that are just there for decoration. At the same time, the piccy of Syd Barratt from 1975 [[:File:Syd Barrett Abbey Road 1975.jpg|thumb|left|alt=Photograph shows the subject as overweight, with a shaven head and eyebrows, pale complexion and red-rimmed eyes. This is a dramatic contrast to his appearance when he left the band six years earlier, when he was slightly built and had a full head of curly hair|Syd Barrett, visiting Abbey Road Studios on 5 June 1975]] really needs better alt text than "An overweight white male with shaved head and eyebrows, wearing a white short-sleeved shirt and black trousers, looks at the camera with a neutral expression. The room behind him appears dark, and several unidentifiable pieces of equipment are visible." Whoever wrote it has accurately described the photograph, but has conveyed nothing of its shock value.--Elen of the Roads (talk) 16:45, 27 June 2010 (UTC)

Any suggestions for how to write alt text for an image like that—in a way that wouldn't overburden the writers, but which would nevertheless be helpful—would be much appreciated. It's something we're all struggling with. SlimVirgin ^{talk|contribs} 21:31, 27 June 2010 (UTC)

Completely understand that. I've just had to calm the shattered nerves of a bunch of folks who uploaded vids onto our website, and were not expecting to be asked for transcripts. It was easy for the people who had the vid showing alternative traffic flows around the town centre following a new development, as the page text described the flows and the vid was just to make a visual of the text; but I had to explain to the ones whose vid showed how to complete some government return that people without sound or vision had to have some other way to find out how to fill the damn thing in. I'm happy to write up a section on the best advice I've come across. --Elen of the Roads (talk) 21:45, 27 June 2010 (UTC)

If you have the time to do that, it would be very helpful. SlimVirgin ^{talk|contribs} 21:48, 27 June 2010 (UTC)

Something is amiss here

When I see a polite well-worded post dismissed like this, and try to read through this mess, I have to say that somehow something has gone amiss here, and one of Wiki's truly finest editors is being mistreated, misunderstood, disrespected, ignored ... or something. I have known and worked with Colin since my earliest days on Wiki, and I consider him hands down one of Wiki's top ten editors-- in terms of knowledge, collaborative spirit, temperament, dedication, experience (not only as an FA writer, reviewer, and developer of policy and guidelines pages, but also as a content contributor), and every other good quality that benefits Wiki. I don't have time to keep up here, but I'm going to watchlist this page now and at least give it a try. Wiki will be ill served by alienating one of Wiki's finest, politest, and most collaborative and experienced editors in a protracted dispute that seems to be verging into rudeness. This page needs more experienced eyes to get a handle on this dispute. I encourage participants here to work collaboratively, as I can think of few editors who are as excellent at doing just that as Colin is; do not waste the opportunity to work with one of Wiki's finest. Please. SandyGeorgia (Talk) 11:46, 27 June 2010 (UTC)

Sandy, please read Colin's posts about me and to me here and elsewhere. There has been some very unhelpful behaviour, attempts to manipulate what's going on, and contacting people on and offwiki in a way I find quite bizarre over something as relatively trivial and easy to sort out as this. This is not how any editor should behave. SlimVirgin ^{talk|contribs} 19:21, 27 June 2010 (UTC)

Colin is never unhelpful, manipulative, or anything else. What I find bizarre is losing one of Wiki's finest editors over something you consider trivial. Keeping up with medical articles without Colin and Eubulides will be ... just about impossible. Was it worth it? SandyGeorgia (Talk) 00:10, 28 June 2010 (UTC)

There's no reason for any editor to be lost over a dispute on this page (or any other). But before judging I urge you to look at what Colin has been posting on this talk page, and in various other places; calling it manipulative is putting it mildly. I'm sorry to say that, but I've been pretty stunned by it. He accused me of dishonesty; he asked for the page to be protected pre-emptively on his version days after the last edit (something I've never seen an experienced editor request on any page, never mind an essay); and he complained at AN/I when it wasn't done, without even informing me; indeed, he seemed to be asking that I not be informed. All this because I disagreed with him.

Back to the point: this is currently an essay that we hope to develop into a guideline. As it stands, it has no consensus (nor did previous versions), because it's a new thing for Wikipedia and for websites generally—plus our needs are a bit different—so there lots of ideas about how best to handle it. Therefore, lots of people have to be encouraged to dip in, so that a consensus can slowly emerge from different perspectives. For that reason it has to be open for editing without any one person assuming they know best. SlimVirgin ^{talk|contribs} 01:50, 28 June 2010 (UTC)

I will ignore your misstatements about Colin and his edits. Re: "this is currently an essay that we hope to develop into a guideline ... ", who is "we"? I don't see collaboration here, and Colin doesn't seem to be the problem. SandyGeorgia (Talk) 06:59, 28 June 2010 (UTC)

The beauty of WP is that it's all here in black and white, so please read it before commenting. In particular, please read the post you opened this thread by calling polite and well-worded. [12] The only people who've written the text from scratch are Eubulides and Colin; everyone else has just copy-edited, tightened, or re-arranged, but that apparently is not allowed. SlimVirgin ^{talk|contribs} 07:09, 28 June 2010 (UTC)

Please don't misrepresent (perhaps a misunderstanding): I opened this thread with this post (and I wouldn't characterize Colin's post as you have, either). A consensus guideline does not result from both of the editors most involved in alt text leaving Wiki because editing here became heavy handed. Nor does Wiki, FAC, medical articles or any other area of Wiki benefit from the loss of two of Wiki's finest, most collaborative and experienced editors. The mischaracterization of Colin's behavior is noted, with concern. SandyGeorgia (Talk) 12:44, 28 June 2010 (UTC)

Final request

It is clear that never was any point in me wasting my time here. But, in vain I leave you with one request and a few comments (the following text refers to this version). Perhaps some of the editors here will agree with me on at least some things.

• I implore everyone here that you keep the lead paragraph and that the first paragraph of the second sentence^{This appears to be a typo-- apparent meaning is "first" sentence of the "second" paragraph. SandyGeorgia (Talk) 07:59, 28 June 2010 (UTC)} be kept but reworded (as proposed above or similar) to remove the references to MediaWiki markup. The first two sentences contain the finest definition of alternative text that has ever been written. I can say that because I didn't write them (other than to include the word "essential", which I think helps in an important way). They come from the W3C standard that both I and Ellen respect. Those guys know a thing or two about alternative text. I've already given my reasons why the particular phrasings they use are excellent (see above) and as someone who truly values using precisely the right words, I deeply admire their definition. The last sentence of the lead paragraph is why we should care. It deserves prominence does it not? The first sentence of the second paragraph, which states that alternative text is achieved through a combination of caption and alt text is (after the definition) the most important sentence in this guideline. As I said, it can be reworded to avoid jargon but it is of vital importance.

Of secondary importance:

• The blank alt text problem is a tricky one and one that took us a long time to realise and understand. So it will be doubly surprising and confusing to our readers. The MediaWiki implementers have cocked-up big time on this one, which really doesn't help. Please make it clear why this is a problem. The two solutions currently offered deal with different situations. The "refer to caption" solution is probably the worst idea I have seen in a very long while and I hope those present here will reject it with gusto. The "refer to body text" is not really any better.
• I sincerely hope the alt text / alternative text issue is dead in the water. This talk page is now littered with examples of why the current usage is correct and helpful and the previous conflation of those terms was deeply confusing and just plain wrong. Ellen's recent posts have only added to the source-based proof that we are currently using the right words the right way.
• If you are wondering why a particular sentence in the body text makes as much sense as marmite, you might want to look at the version 4th June which contains some of the text now randomly scattered about this one.

Colin°^Talk 21:26, 27 June 2010 (UTC)

I noted (above) an apparent typo in Colin's post. SandyGeorgia (Talk) 07:59, 28 June 2010 (UTC)

Key issues

The problem is that no one seems to know what the industry regards as standard, so the page has lurched back and forth. Eubulides version (5,538 words) required long and detailed ALT text even when the caption or text made clear what the image was; that this became a requirement at FAC and FAR didn't help, and the lack of consensus led to the page's demotion. After that it was rewritten by Colin here (1,872 words), which is shorter, but it's not entirely clear what it's advising.

Ideally, guidelines should be succinct and well-structured, so that editors can quickly get an overview by glancing through them. From that perspective, both versions of the text fail in my view, because they're wordy, repetitive, confusingly structured, and contain too much jargon. Some jargon is inevitable, but editors with technical knowledge are asked to bear in mind that the page has to be accessible to people with none. It seems to me that this version (1057 words), after I tightened and rearranged Colin's version, is clearer in that regard (though it isn't what I would write if I were starting from scratch).

My only interest in the page is twofold: (a) to have the page clearly written, and (b) to make sure it lays out sensible options, but doesn't compel editors to follow them. Other than that, I have no interest in recommending any particular form of alt text. SlimVirgin ^{talk|contribs} 07:40, 28 June 2010 (UTC)

The industry has two standards - one is about 10 years old. I've linked to both higher up. I'll take more of a look at the article over the next couple of days - it should not be hard to incorporate the WCAG 2.0 standard as regards alternative text. It should be presented as 'this is the best advice available at the moment", bearing in mind the limits of the MediaWiki software - which prevents users from actually following some of the standards guidance.--Elen of the Roads (talk) 13:14, 28 June 2010 (UTC)

1) We're never "compelled" to follow a guideline (unless it's an issue at FAC, but even then, reasonable exceptions are made, and alt-text was dropped from WP:WIAFA). 2) Colin writes clearly; the treatment of him here was unfortunate. 3) I have no interest in recommending any specific form either, but I am very concerned when discussion on an essay page is unnecessarily personalized, and we lose extremely valuable editors to what looks like page ownership. I encourage y'all to modify the tone here to one of collaboration and respect, since the two most knowledgeable and experienced editors who could contribute to turning this page into a useful guideline are now gone. SandyGeorgia (Talk) 14:42, 28 June 2010 (UTC)

Hey! I've got no beef with Colin as regards the text on this page!--Elen of the Roads (talk) 15:01, 28 June 2010 (UTC)

I am also committed to trying to get this article right, though I have been very busy with real life in the last little while. I welcome Elen's knowledgeable input, and will try to get myself up to speed once again in the next couple of days.

In the meantime, yesterday I threw caution to the wind and went to a Wikipedia Meetup, where I met Brion Vibber. I took the opportunity to explain that Media Wiki limitations were seriously affecting our ability to provide the kind of recommended alt-text support to blind and partially sighted users, and he kindly said he would look into it (again!). I suggest we take him up on this. Can we try to develop a summary of the changes required to bring the software up to standard? I'd like to strike while the iron is hot on this matter, but am not tech-savvy enough to formulate this clearly and completely myself. Perhaps this can be a short, non-controversial side project which could serve as a useful first project before looking at changes to the article itself. --Slp1 (talk) 16:44, 28 June 2010 (UTC)

Sounds like a great idea, Slp. SlimVirgin ^{talk|contribs} 18:51, 1 July 2010 (UTC)

There are only two issues I'm aware of the alt text on the enlarge icon and the issue of setting blank alt text. The latter can be workaround by using |alt= (which is valid and better than "refer to body") tool. — Dispenser 19:40, 1 July 2010 (UTC)

That's worth adding. At the moment we're recommending writing "photograph" or "drawing." SlimVirgin ^{talk|contribs} 19:44, 1 July 2010 (UTC)

The implementation we need to have MediaWiki handle images correctly is complex (we all know that already). Yet another small or random change to the software will confuse users even more. And guidelines of projects in several languages will have to be updated. So, we have to get it completely right this time. We need to be thorough.

There are various way to produce images: link/no link, thumb/no thumb, caption/no caption, alt/emtpy alt/no alt. In every case, MediaWiki should produce the best example possible. Documentation and examples for every cases would be way to long for a spoken discussion with a non-expert. Too long for bugzilla too. We should write a specicication in a subpage here, and ask experts to review it according to ATAG 2.0 guideline. Then strive to make it done by developers, as it would be a major and future-proof improvement. Dodoïste (talk) 04:09, 4 August 2010 (UTC)

No longer a requirement?

It's been some months since I've looked at WP:ALT, but it used to mention something about all articles needing alt text, now there's nothing that I could see. I do GAN reviews occasionally, so knowing whether it is required and at what article class it is required would be helpful. --Teancum (talk) 18:29, 26 July 2010 (UTC)

Thanks for your kind offer.

Basically, the MediaWiki software should be improved to handle images in a better way. Once it's done, it will be much easier to produce accessible images. This guideline will be shortened.

So I'd say we should wait for an optimized and stable way to produce accessible images before making huge efforts in teaching users. I do not want to waste your time. ;-) Your offer will be very helpful once MediaWiki's fixed. Dodoïste (talk) 04:26, 4 August 2010 (UTC)