Welcome to my talk page! Please begin new discussion threads by pressing the New section tab above.

Good Work

edit

I got the message that you left on my talk page regarding references for the etymology for the term Round-Robin. I am glad that you were able to add the proper references yourself. thanks Robin klein (talk) 17:24, 15 January 2014 (UTC)Reply

Feedback needed on using special characters

edit

Hello. Thank you for using VisualEditor! Having editors use it is the best way for the Wikimedia Foundation to develop it into the best tool it can be.

While we always welcome general feedback (please report any issues in Bugzilla in the "VisualEditor" product or drop your feedback on the central feedback page on MediaWiki.org), the developers are especially interested right now in feedback on the special character inserter. This new tool is used for inserting special characters (including symbols like , IPA pronunciation symbols, mathematics symbols, and characters with diacritics). It is intended to help people whose computers do not have good character inserters. For example, many Mac users prefer to use the extensive "Special Characters..." tool present at the bottom of the Edit menu in all applications or to learn the keyboard shortcuts for characters like ñ and ü.

The current version of the special characters tool in VisualEditor is very simple and very basic. It will be getting a lot of work in the coming weeks and months. It does not contain very many character sets at this time. (The specific character sets can be customized at each Wikipedia, so that each project could have a local version with the characters it wants.) But the developers want your ideas at this early stage about ways that the overall concept could be improved. I would appreciate your input on this question, so please try out the character inserter and tell me what changes to the design would (or would not!) best work for you.

 
The "insert" pulldown on the task bar of VisualEditor will lead you to the '⧼visualeditor-specialcharacterinspector-title⧽' tool.
 
This is the ⧼visualeditor-specialcharacterinspector-title⧽ inserter as it appears on many wikis. (Some may have customized it.) Your feedback on this tool is particularly important.

Issues you might consider:

  • How often do you normally use Wikipedia's character inserters?
  • Which character sets are useful to you? Should it include all 18 of the character sets provided in the wikitext editor's newer toolbar at the English Wikipedia, the 10 present in the older editor toolbar, or some other combination of character sets?
  • How many special characters would you like to see at one time?
    • Should there be a "priority" or "favorites" section for the 10 or 12 characters that most editors need most often? Is it okay if you need an extra click to go beyond the limited priority set?
    • How should the sections be split up? Should they be nested? Ordered?
    • How should the sections be navigated? Should there be a drop-down? A nested menu?
  • The wikitext editor has never included many symbols and characters, like and . Do you find that you need these missing characters? If the character inserter in VisualEditor includes hundreds or thousands of special characters, will it be overwhelming? How will you find the character you want? What should be done for users without enough space to display more than a few dozen characters?
  • Should the character inserter be statically available until dismissed? Should it hover near the mouse? Should it go away on every selection or 10 seconds after a selection with no subsequent ones?
  • Some people believe that the toolbar already has too many options—how would you simplify it?

The developers are open to any thoughts on how the special character inserter can best be developed, even if this requires significant changes. Please leave your views on the central feedback page, or, if you'd prefer, you can contact me directly on my talk page. It would be really helpful if you can tell me how frequently you need to use special characters in your typical editing and what languages or other special characters are important to you.

Thank you again for your work with VisualEditor and for any feedback you can provide. I really do appreciate it.

P.S. You might be interested in the current ideas about improving citations, too. Whatamidoing (WMF) (talk) 00:20, 18 February 2014 (UTC)Reply

Non responsive matrices (reciprocal)

  1   2  3  4  Weight
  1   7  2  7  0,480
      1 49  9  0,444 <<<<
         1  9  0,059
            1  0,018
  1   2  3  4  Weight
  1   7  2  7  0,411
      1 49  1  0,451 ++++ (less favourable judgement, better overall priority
         1  9  0,081
            1  0,057

non-monotonic behaviour

edit

Less favourable judgement (1/5 instead of 5), better rank (1 instead of 2)

                           -A'(x=5) CR=9,9%-    -A(x=⅕) CR=46% -
 9   1 2 3 4 5 6 7 8 9     Priority Rk f2/fk    Priority Rk f2/fk
 ----------------------------------------------------------------
 1   1 5 1 1 1 2 2 2 3      0,209   1  0,969     0,151   2  1,010 <<< non-monotonic
 2     1 3 3 3 3 3 3 x <<   0,202   2  1,000     0,153   1  1,000
 3       1 1 1 1 1 1 9      0,100   3  2,016     0,107   3  1,423
 4         1 1 1 1 1 9      0,100   3  2,016     0,107   3  1,423
 5           1 1 1 1 9      0,100   3  2,016     0,107   3  1,423
 6             1 1 1 9      0,090   6  2,246     0,102   6  1,501
 7               1 1 9      0,090   6  2,246     0,102   6  1,501
 8                 1 9      0,090   6  2,246     0,102   6  1,501
 9                   1      0,019   9 10,663     0,070   9  2,179

LS-method rating

edit

Let G(nxn) be regular bipartite and complete with two disjoint parts. Let n > 1 be the number of nodes. Let m > 0 be the number of comparisons between Xi and Xj. Let s be the score vector. Let q be the rating vector. Let h = n/2.

Rating vector q[1] equals:

   qi = (2.si +  Σsk / h) / m.n, for all Xk not in the part of Xi

The solution for GRM[2] equals:

   qi = ( γ.si + m. ∑sk ) / ( ϵ-1 + m.n/2) ,  for all Xk not in the part of Xi, and  γ = (ϵ-1+ m.n)

For ϵ-1 = 0 this amounts to the LS-solution, up to a factor (n.m)-1.

References

edit
  1. ^ Csató, László (27 Aug 2015). "A graph interpretation of the least squares ranking method". Social Choice and Welfare. 44 (1): 51-69. doi:10.1007/s00355-014-0820-0.
  2. ^ Cheboratev, Pavel; Shamis, Elena (1997-01-01). Constructing ab Objective Function for Aggregating Incomplete Preferences.

Elo rating system anomalies

edit

PMcGarrigle (talk) 18:49, 18 March 2022 (UTC)Did you ever figure out where the irregularities in Elo's expectancy table come from?Reply

No I have not. I still intend to visit the Euwe library in Amsterdam.
I think there are still old versions of the table lying around somewhere.
KP (talk) 21:22, 18 March 2022 (UTC)Reply

The table values seem to be Q( 7D / 2000 ) rounded to two decimal places, except for 54, 343, 344, 358, 392, and 620, which are rounded the wrong way.

Elo may have switched from Q( D / 200 sqrt(2) ) to Q( 7D / 2000 ) in order to use tabulations of the Q function, with linear interpolation. (It seems unlikely that he would have calculated the values any other way, since he did this in the late 1950's, I think.) This can explain 392 and 620: 7 x 392 / 2000 = 1.372, Q(1.37) = 0.9147, Q(1.38) = 0.9162, and via linear interpolation we get 0.915, which could be rounded up to 0.92, matching the table.

However, this fails to explain 54, 343, 344, and 358 (and perhaps may introduce other discrepancies).

I'm reluctant to assume Elo made computational errors. But what other explanation is there?

Irregularities in The Percentage Expectancy Table

edit

Tables of Normal Probability Functions U. S. Department of Commerce National Bureau of Standards Applied Mathematics Series #23 Issued June 5, 1953[1]

Fide Table Handbook Fide, https://handbook.fide.com/chapter/B022022
The Rating of Chessplayers. [2]

Below we find the irregularities in the table:

Rounding errors in The Percentage Expectancy Table
σ=2000/7
Dif. H z-score Table y 50% + y / 2 Calc P
54 57% 0,1890 0,14990 71832 09998 0,57495 35916 04999 0,574953591604999 57%
88% 1,2000 0,76986 06595 56583 0,88493 03297 78292 0,884930329778292
343 1,2005 Interpolate 0,885027364557095 89%
88% 1,2010 0,77024 87986 71795 0,88512 43993 35897 0,885124399335898
344 88% 1,2040 0,77141 04214 54696 0,88570 52107 27348 0,885705210727348 89% *
358 90% 1,253 0,78979 42933 03286 0,89489 71466 51643 0,894897146651643 89% *
392 92% 1,372 0,82993 65621 86873 0,91496 82810 93437 0,914968281093436 91% *
620 99% 2,170 0,96999 31540 52536 0,98499 65770 26268 0,984996577026268 98% *
edit

Hi. Thank you for your recent edits. An automated process has detected that when you recently edited Elo rating system, you added a link pointing to the disambiguation page USCF. Such links are usually incorrect, since a disambiguation page is merely a list of unrelated topics with similar titles. (Read the FAQ • Join us at the DPL WikiProject.)

It's OK to remove this message. Also, to stop receiving these messages, follow these opt-out instructions. Thanks, DPL bot (talk) 06:09, 16 April 2022 (UTC)Reply

  1. ^ Tables of Normal Probability Functions. Applied Mathematics Series. Vol. 23. U. S. Department of Commerce. 1953.
  2. ^ Elo, Arpad E (2008) [1978]. "8.1.2 Table of conversion of difference in rating, D, into scoring probability PD, for the higher, H, and the lower, L, rated player respectively.". The Rating of Chessplayers, Past&Present. Bronx NY 10453: ISHI Press International. ISBN 978-0-923891-27-5.{{cite book}}: CS1 maint: location (link)