World Mind Sport Games 2012

[Saltie]

WMSG in 2008 were covered by Asian press and tv.
Not by other countries.
China did a huge sponsorship too.

Let's see what happens this year.

[HermanHiddema]

Some numbers from http://wmsg2012.org/

79 players in the men's individual
39 players in the women's individual
28 teams of three players in the team event
21 pairs in the pair go event
44 players in the under 21 competition

A total of 30 countries are represented, though sadly neither China nor Korea are sending a team, so I guess the main contenders will be Taiwan and Japan then.

[HermanHiddema]

Well, Taiwan definitely dominated the Men's Individual, taking all four semi-final places.

The Women's semi-final is more varied, with contenders from Taiwan, Japan, Canada and the UK.

[RobertJasiek]

Let me try to reverse engineer part of the used tournament system of the Men's Individual from its final result table:

- Players are required to play all rounds of the first stage. Seeded players are required to play all rounds.
- The first stage is a 5 rounds McMahon bands system. The second stage is a KO for four players with a game for place 3.
- The bands are 7d-5d (initial MMS 3, 41 players), 4d-3d (MMS 2, 17 players), 2d (MMS 1, 11 players), 1d-16k (MMS 0, 9 players). (Total 78 players. Note: 41 players in the top group is a possible number for 5 rounds to distinguish between places 4 and 5 by means of tiebreakers.)
- The final placement criteria are MMS - Wins - SOS - SOSOS. (Notes: As the tied places indicate, there is no apparent further tiebreaker after SOSOS; but how would a tie on place 4 have been broken?! This system gives initial MMS 2 players, i.e. 3d or 4d, the chance to qualify to the KO by winning all 5 games against on expected average tournament-weaker opponents than the qualifying initial MMS 3 players. Very interesting idea. However, unfortunately, initial MMS 0 or 1 players have no principle chance to qualify. The randomiser SOS - SOSOS determined places 4 and 5 indeed.)

It is very unfortunate that the tournament system has to be reverse engineered and has not been published for everybody before the tournament start.

Which pairing strategies were used?

[RobertJasiek]

Reverse engineered part of the tournament system of the Women's Individual: like Men's with these changes:

- Bands 6d-4d (initial MMS 2, 15 players), 3d-1d (MMS 1, 7 players), 1k-13k (MMS 0, 16 players). (Total 38 players.)

Luckily, tiebreaking was not necessary for seeding. Exactly four players ended with MMS 6 (4 wins).

Which pairing strategies were used?

[HermanHiddema]

Actually, in the Men's individual, some 4d also started in the top group (e.g. 43. Lucas Neyrink, 38. Jesse Savo). I believe this is the case for those 4d that are the strongest participant from their country.

Pairing strategy is probably the usual as implemented by Gerlach's MacMahon, I see little reason to deviate from that.

[RobertJasiek]

Taiwan definitely dominated the Men's Individual, taking all four semi-final places.

The first two players are determined clearly. The following three players happen to be those happy enough to lose against one of the first two players. The relative order within the following three players is randomised by tiebreakers. So players 3 to 9 performed very similarly and players 5 to 9 simply had the worse pairing luck. Nevertheless, it is indeed impressive that four of five Taiwanese 7d have 4 or 5 wins.

The Women's semi-final is more varied, with contenders from Taiwan, Japan, Canada and the UK.

Jin Yu beat me in the EGC. In that game, I had some chances, but her reading is clearly better, without being overwhelming. So I would estimate her playing strength as about European 5.5d. If this is taken as at least some indication of the Women tournament's top playing strength, there appears to be a significantly lower world amateur top women level compared to the world amateur top men level. If you have further information on the women's strength, please share it! OTOH, I cannot know how many possibly stronger players simply did not participate because of the later announcements or travel expenses.

[RobertJasiek]

Pairing strategy is probably the usual as implemented by Gerlach's MacMahon, I see little reason to deviate from that.

That would mean fold pairing, which would be very good in rounds 3+. Rounds 1 and 2 might deserve more careful study.

[HermanHiddema]

Unless the program was instructed to use rank/rating, which I guess is somewhat unlikely with so many participants from entirely different backgrounds, this would mean that the first two rounds are random, as there is no sorting possible yet. This seems to be the case, given e.g. pairings in round one like 11 vs 44 (5d-5d) and 2 vs 37 (7d-6d). Although there is something to be said for pairing based on rank/rating, there is also something to be said for treating all players in the top group exactly the same.

[RobertJasiek]

Sorting in rounds 1 and 2 is not possible using only tournament-internal information. A few world championships used earlier country success as sorting criterion. I prefer equal treatment because that best meets the event character of determining the currently best from all countries. For that reason, McMahon style is a dubious choice.

[RobertJasiek]

http://www.wmsg2012.org/results/teams

The team tournament's placement criteria appear to have been


1. Team Points (0 = loss, 2 = win)

2. Board Wins

3. Sum of Opponents Scores (Team Points)

4. Board Wins on 2 Upper Boards

5. Board Wins on 1st Board


While 1. and 2. are rather natural (although an objection, which I do not share, says that getting easy opposing teams leads to many board wins; 1 for a team win and 1/2 for a tie would also work). The order of 3., 4. and 5. seems somewhat arbitrary, but any order of further tiebreakers would. Something more complicated instead of SOST, such as a hybrid of 2. and 3., has not been chosen. If one uses SOST, then why not also SOB2U and SOB1U...? I.e., if one trusts opponents' results and top board wins at all, then, by assumption, any such tiebreaker would be more meaningful than lottery. IOW, the used tiebreakers pretty much amount to 1. Team Points - 2. Board Wins - 3. lottery.

[HermanHiddema]

SOST is an obvious choice, and a good one. SOS is widely considered the best tiebreaker there is, when tiebreakers are needed.

I do not think higher boards are more meaningful in general, but using them as tiebreakers discourages/punishes strategic player distribution (i.e. to put the weakest play on board 1 to hopefully get more wins on boards 2 and 3 that way, thus increasing overall chance of team wins).

[RobertJasiek]

SOS is widely considered the best tiebreaker there is, when tiebreakers are needed.

Of course not. To simplify discussion, playing more (fast) games is a better tiebreaker.

[HermanHiddema]

SOS is widely considered the best tiebreaker there is, when tiebreakers are needed.

Of course not. To simplify discussion, playing more (fast) games is a better tiebreaker.

Yes, but there never seems to be time for that. My statement applies to the commonly used tiebreakers, which do not involve playing extra games.

[RobertJasiek]

there never seems to be time for that.

"seems", exactly:)

My statement applies to the commonly used tiebreakers, which do not involve playing extra games.

For discussion of direct comparison, SOS and other nonsense, see elsewhere:)