User talk:Franamax/Ucontribs-0.3b

Questions, problems and requested enhancements to the current uContribs program should be left here. Franamax (talk) 05:30, 19 November 2008 (UTC)

Request

Since you offered—if it's not too much trouble, I'd like to see what a listing for me would look like. Thanks. Maralia (talk) 03:04, 22 November 2008 (UTC)

Done! From a quick look: you're a vandal-fighter (since you overflowed the page table >8000 articles); you're interested in article quality, but mostly as a gnome (low counts to individual articles, good counts to FAC and wikiprojects); and you've been losing interest lately (hope not, but monthly edits are down!). Also more interested in articles than political goings-on. Am I close? Franamax (talk) 04:23, 22 November 2008 (UTC)

Thanks much. Hm...hot and cold, really. Of course, it's possible my own perception could be terribly skewed, but in the interest of science, let's go over it:

1. you're a vandal-fighter (since you overflowed the page table >8000 articles)

   A reasonable interpretation, but there's a clue that it's not the whole explanation: "Total [mainspace] Contrib's: 12894 (6958 page, 5936 talk)".

   Yes, I missed that. High counts to article talk pages, but low counts to any single page. I'll switch over then - either you're an image/copyvio-patroller, or an article classifier. You have a low imagespace count, so the conclusion now is doing classify/assess work.

2. you're interested in article quality, but mostly as a gnome (low counts to individual articles, good counts to FAC and wikiprojects)

   The most off-base conclusion. I definitely have a tendency toward gnoming, but the statistics are misleading; I have a habit of making substantial edits in a manner that's apparently invisible to existing contribs tools: I often make numerous changes in one edit. (Incidentally, it is because I tend to edit articles that are not popular/controversial—as well as articles I'm requested to copyedit or that are otherwise explicitly up for review such as at FAC/FAR/WPR/PR—that I am able to incorporate numerous changes in very few edits.)

   Voila, the dangers of edit-counting! I personally try to always edit even the most placid of articles section-by-section (so the section name shows in the edit summary, which I feel is informative to all), but I do try to get the whole section in one shot. However I sometimes try to segregate my reference-adding from my text-editing, since my reference work often results in spectacular explosions. Note that I'm not comparing my own work to that of any serious editor, running these scans has certainly confirmed the reality of my own limitations...

3. you've been losing interest lately (hope not, but monthly edits are down!)

   A reasonable interpretation (apropos of nothing, it's a reflection of real-life changes more than anything else).

   Well, here's hoping it all works out. If there's a preference to opt for, choose RL! :)

4. Also more interested in articles than political goings-on.

   As accurate a guess as anything based on actual edits (versus, say, my page hits) could be. Actually, I tend to follow such things pretty carefully, but rarely comment; when I do, it's typically a single, laboriously-crafted statement. I almost never restate my opinion or get drawn into lengthy debates.

   I'd thought for awhile of creating an "index", so that you could actually follow down to see what the actual contribution was. In many project areas, the contribution can often be segregated as either sober or hysterical. Simple counts can never reflect that. I myself try to limit my contributions to any particular thread, just say my piece and get on with life. That's definitely a limitation of any "most-edited" tool. It may be possible to create a "most-verbose" tool, but for now, a "most-intelligent" tool remains out of reach.

I wonder why I've never seen a contributions analysis tool that looks at bytes changed per edit. It would help complete the picture when looking at an editor's edit counts to articles—and might lead to other interesting comparisons (signal to noise, anyone?). Maralia (talk) 18:43, 22 November 2008 (UTC)

Thx for fdbk, cmnts above at dbl-indents. I've yet to look at any single one of your edits, so it's very interesting to refine the concept of what a macro view actually conveys.

As far as bytes-changed, well, watch this space or better yet, probably a similar name but with a version number of 0.7. It's a little tricky, your bytes-changed could be a revert of page-blanking; could be a revert of a vandalism as the second edit after a bot stepped in meanwhile; your text addition may have been reverted seven times by an edit warrior (with a possible bot edit in between each); and of course, byte-counts also tell only a small part of the story, as you say, it's "bytes-changed" that matter, a la copy-edits. Now you need a lexical analysis system, which quickly approaches intractability.

What I am working on is a simple net-plus/minus-bytes analysis (possibly with "# spots" if I can manage it, but moving a sentence will likely break that completely :( ), summary of links-added (imo adding references points to a quality editor), and tag-counting (adding problem tags to articles is soo easy, fixing them to get the tag off is often quite difficult). Anyway, I'm on it, and thanx for the comments! Franamax (talk) 04:36, 23 November 2008 (UTC)

Another request

Could you run me through your program, as well? I'm curious.--Aervanath lives in the Orphanage 11:33, 23 November 2008 (UTC)

Done! I used 120 days as "recent", shooting for about half your edits. Franamax (talk) 19:29, 23 November 2008 (UTC)

Yet another request

I'm interested as well. -MBK₀₀₄ 19:18, 23 November 2008 (UTC)

Will do, but AFK 'til Tuesday. Franamax (talk) 07:39, 24 November 2008 (UTC)

Just a heads up, I have a huge edit count. Wannabe Kate says that I have edited 20,442 unique pages for a total edit count of 32,909 as I type this. -MBK₀₀₄ 02:06, 25 November 2008 (UTC)

Done! I used 200 days s your recent-cutoff. The mainspace listing overflowed the page table as you predicted. I'll shortly add an indicator for overflows (they are indicated in the detail). Once I have conquered the Charles Matthews overflow, you'll just be roadkill :) Franamax (talk) 23:39, 25 November 2008 (UTC)

Great tool

I hope you get it set up on the toolserver so that people can run it whenever they want; my first non-trivial userbox lets people see my editcounts as parsed by the tools from Interiot and SQL, and it would be great to add yours. Can do you a run for me? - Dan Dank55 (send/receive) 20:58, 23 November 2008 (UTC)

Done! I ran a listing for you using 120 as the recent-days cutoff. The software is rather bound up with Visual C++, so toolserver residence is a little ways off, but not necessarily impossible. I'm always willing to re-run, and I will give a copy of the software to any reasonably established editor who uses Windows. Franamax (talk) 23:43, 25 November 2008 (UTC)

I think this is a fantastic tool. - Dan Dank55 (send/receive) 02:52, 26 November 2008 (UTC)

Let's see how good your tool really is..

<evil laugh> ;-) J.delanoy^gabs_adds 04:31, 24 November 2008 (UTC)

Oh, wait. You already did Charles Matthews. I wanted to break it on you. Oh, well.... J.delanoy^gabs_adds 04:33, 24 November 2008 (UTC)

Yeah, heh, I sat watching that run and run, thinking "it must be going to end soon now", but noo... The report does have some minor breakage, so I'm going to build a mega-memory program version (which I'll call uCharles :) and let it run for a day or two, just to see exactly how many more than 8000 pages he's edited. Enquiring minds need this kind of information :) Franamax (talk) 07:51, 24 November 2008 (UTC)

Question

My prior account was User:Jaranda, is there anyways you can combine the edits from Jaranda and this account. Thanks Secret ^account 18:35, 24 November 2008 (UTC)

Yes, I saw your other note. I'm AFK for another 24 hours but will do it then. Franamax (talk) 23:14, 24 November 2008 (UTC)

Toss me in there, will ya?

Also, can you run it on Commons or on any particular wiki? Or all of them? ++Lar: t/c 02:08, 26 November 2008 (UTC)

Done! (for en:wiki) In principle, it should work on any MediaWiki-based wiki that supports the API. In practice, it would need to acquire the various proper space names. I'm tooling around with it right now, so watch the bottom of the page I just made for you. Franamax (talk) 04:43, 26 November 2008 (UTC)

Cool. Commons and Meta are probably the big two. One thing about Commons, the namespace is skewed... you can almost treat Image as if it were mainspace, much of the content creation work goes on in that space. Mainspace really is just for galleries and the like and is not used much at all (as your stats found, there were only what, 4 seriously active pages for me in mainspace). Another space to watch that is content related (proportionally more so than here) is Category. You picked up that most of the adminish action goes on in Commons and Commons_talk.... Meta also is a bit skewed but in a different way. A lot of the admin (and steward) tasks actually happen IN mainspace. For example, all the steward request pages are there. (m:SRP, m:SRCU, m:SRUC, m:SRB etc...) a fair bit of my recent edits are to those pages since I'm a steward. If you're really into translation, that skews even differently but I don't know the details there as I'm not much of a translator. Anyway... very nice. Thanks for doing this. ++Lar: t/c 06:27, 26 November 2008 (UTC)

As I suspected, and just confirmed, the API returns meta-info about the namespaces, including whether they can have sub-pages (page families), so I can build a namespace menu appropriate to each language wiki also. It wouldn't be internationalised fully, but in general it looks like I could run this for any WMF wiki, for sure. As you note, you need to know where the action is in each wiki, but Interiot and SQL's tools will give you the clue for that. I also learned the hard way that I should put a ":" before page names in spaces like Image: :) Thanks for the exercise! Franamax (talk) 06:36, 26 November 2008 (UTC)

Forgot to follow up here, what I meant by "into translation" is that one of the things that people go to Meta FOR is to work on translating things. (less now than originally as there now is a wiki just to house translations of the interface)... translators will have a lot of edits in the MediaWiki space if they're doing interface translation. Most regular admins hardly touch it at all, (I've done a few edits there from time to time but hardly any), and non admins never, as you have to be an admin (by default) to even edit there, the entire space is, by default, protected. Wasn't talking about you translating your results or anything like that. ++Lar: t/c 05:31, 17 December 2008 (UTC)

Run date

You might think of including a "run date" column under the "Other listings" header. While all the ArbCom candidates are counted from the same day, those of us "others" were not, and it might be helpful to include that information.--Aervanath lives in the Orphanage 16:20, 26 November 2008 (UTC)

Done! Franamax (talk) 00:37, 4 December 2008 (UTC)

Jumps on bandwagon

It'd be great if this could be done for me, if only out of curiosity. Thanks! Cheers, Daniel (talk) 13:51, 28 November 2008 (UTC)

Done! Franamax (talk) 00:37, 4 December 2008 (UTC)

ArbCom?

Would it be possible to get these statistics for the sitting ArbCom members as well? It'd be very interesting to see how we stand relative to the current candidates. Kirill ^(prof) 19:27, 28 November 2008 (UTC)

Done! Franamax (talk) 00:37, 4 December 2008 (UTC)

Me, me, me!

Please? Pretty please? I'd be very grateful if you could run this for my edits. My only reason for asking is sheer vanity really. But it's interesting stuff. Very well done, Angus McLellan (Talk) 00:45, 2 December 2008 (UTC)

Ditto. :) لennavecia 04:46, 2 December 2008 (UTC)

Done and done! Franamax (talk) 00:37, 4 December 2008 (UTC)

Thank you! لennavecia 19:47, 4 December 2008 (UTC)

Likewise, thank you very much. Angus McLellan (Talk) 23:30, 4 December 2008 (UTC)

Update revision 0.3b2

Repeated examination has turned up two "opportunities for improvement". Some might call them bugs, but I'll have no truck with the anti-lepidopteran crowd here :)

One problem was my only getting the API default 10 categories from the talk page to evaluate article ratings. Fixed (maximum is now 500). I estimate a maximum 10% of articles have more than 10 categories, almost none have more than 20. Of the crowded categories, at least half are asssessment categories. This implies a 4% error rate in article assessments, which is now fixed. The only place I'm aware this is of significance is with Jehochman, an ArbCandidate, where I managed to miss one of his good-article ratings. Sorry Jeh, I did say somewhere "experimental", not an excuse, thanks for the tip, fixed now - did I say sorry yet?

The other problem was defining "recent days" - I didn't adjust between "start day" and "today" in calculating "oldest day" - so recent figures for arb candidates will be overstated by 1-2%.

So there's an rms error of ~4.5% maximum in all my previous data. Everything I can find is now fixed - but please, anyone who has any doubts, tell me - as long as I'm counting to less than a thousand or so, I'll check it out by hand.

I'll redo a random sampling of Arb candidates to verify my error estimates, then do the current requests. Franamax (talk) 10:40, 2 December 2008 (UTC)

Re-running career listings for Casliber, Jayvdb and Rlevse (544 top articles, total) showed 69 discrepancies, all in importance ratings, CD inclusion and "former" status. The weighted error rate corrected from the 0.3b version is 3.4%. Franamax (talk) 00:46, 4 December 2008 (UTC)

Ooh. I enjoy stats.

Thanks for doing these. I would be interested to see how my own stats come out. I'd guess that renames that I do will have artificially inflated my user/user talk edits in recent months, but still fascinating. Cheers, --Dweller (talk) 11:59, 11 December 2008 (UTC)

Done! Franamax (talk) 23:53, 12 December 2008 (UTC)

Mind running mine? II | (t - c) 01:06, 14 December 2008 (UTC)

In order to know where I'm going, it may be helpful to know where I've been. Thank you!--Buster7 (talk) 14:02, 14 December 2008 (UTC)

Both done! Franamax (talk) 22:55, 15 December 2008 (UTC)

Me fourth (if that's a word). ~the editorofthewiki ^{(talk/contribs/editor review)}~ 23:41, 15 December 2008 (UTC)

Doneth! (If that's not a word yet, give me a few minutes at Wiktionary :) Note: you have way too much focus on article-building and collaboration, you don't seem to be addressing the critical shortage of drahma and self-absorption we're trying to counter ;) Franamax (talk) 04:29, 16 December 2008 (UTC)

Add me in too! - Mailer Diablo 08:12, 16 December 2008 (UTC)

Done! Edit total from the API, start date from edit history (>45000 edits = wannabekate overflow) Franamax (talk) 12:40, 16 December 2008 (UTC)

Thanks! No wonder katewannabe died on me too.. - Mailer Diablo 14:16, 16 December 2008 (UTC)

Can I be added, thanks a bunch! Miguel.mateo (talk) 09:40, 6 February 2009 (UTC)

Something wonky about the sorting

I popped back in to be further embarassed at my paucity of mainspace contribs, and decided to see just how bad it was compared to others... So I went to User:Franamax/Ucontribs-0.3b#Other_Listings and clicked the sort arrow on mainspace edits. Wonkiness! I don't think it handles Colspanning in headings very well. Perhaps use a first table for the spanned headings (with the widths pinned so things line up) and put all the sorts in teh second row of the heading? If that's making no sense please advise. I use Firefox 2 on Win XP. ++Lar: t/c 16:46, 16 December 2008 (UTC)

Lar, if you want to give it a shot, please do and I will learn from your example. Cool Hand Luke fixed the Candidates table for me (or at least partly fixed it) and I think I just copied him, but it sure does go nuts. None of the examples at Help:Sorting deal with multi-spans, and I believe the Help desk advised me something like you've suggested. Franamax (talk) 00:49, 17 December 2008 (UTC)

I chickened out, and just got rid of the multispans... looks uglier but works. Revert if too ugly! :) I will take a gander again tomorrow, to try to do what I suggested above. ++Lar: t/c 05:27, 17 December 2008 (UTC)