Life In 19x19

Posted: **Thu Nov 18, 2010 2:56 am**

tapir, for some world champion results and effects of SOS on top players, see

http://groups.google.com/group/rec.game ... ode=source

http://groups.google.com/group/rec.game ... ode=source

Posted: **Thu Nov 18, 2010 3:19 am**

For many years, I have criticised the low quality of SOS especially in 5 rounds Swiss or McMahon tournaments. This could be observed in many national or European weekend tournaments in ordinary, EGF or EGC events. SOS can at best show only what has happened but, in such low number rounds tournaments, the final top wins players have often played only too few games among each other, especially when there are relatively many (strong) participants or a too great initial top group in relation to a (by far) too small number of rounds. Even with perfect pairings, SOS can never in general solve a fundamentally flawed combination of tournament system and round number.

Better pairing strategies can reduce only the symptoms. Women championships have a great variety of player strengths. Therefore it is even far from obvious what could constitute a good pairing strategy in early rounds. Each of the following reference parameter is doubtful for such national or European women championships: rating, rank, earlier tournament performance, earlier values of other same-nation players. Doubtful because a too great fraction of participants have played too few recent tournament games or because same-nation players are too different in strength, too.

There is some good news though: For the European (Open) Champions, the EGF Rules Commission is suggesting (and I hope it will be adopted) for the currently revised EGC rules draft recommended pairing strategies (subject to the pairing programs' capabilities): Cross Pairing by rating (#1-#3, #2-#4) in rounds 1 and 2; Fold Pairing by MMS - SOS (#1-#4, #2-#3) in all later rounds. Thereby Alexander's 3d opponents will be history of the past (unless he loses his early rounds, of course). In the German Championship Preliminaries (6 rounds Swiss, SOS tiebreaker, player field 3d-6d, to qualify previously 6 or now 4 players), we have had very good experience with almost this pairing strategy combination. It does not fix SOS nor a constant round number, but given both it pairs pretty much as well as possible.

Posted: **Sun Nov 21, 2010 2:45 am**

RobertJasiek wrote:
There is some good news though: For the European (Open) Champions, the EGF Rules Commission is suggesting (and I hope it will be adopted) for the currently revised EGC rules draft recommended pairing strategies (subject to the pairing programs' capabilities): Cross Pairing by rating (#1-#3, #2-#4) in rounds 1 and 2; Fold Pairing by MMS - SOS (#1-#4, #2-#3) in all later rounds. Thereby Alexander's 3d opponents will be history of the past (unless he loses his early rounds, of course). In the German Championship Preliminaries (6 rounds Swiss, SOS tiebreaker, player field 3d-6d, to qualify previously 6 or now 4 players), we have had very good experience with almost this pairing strategy combination. It does not fix SOS nor a constant round number, but given both it pairs pretty much as well as possible.

Can you publish the complete proposel (so we can all comment on it ;-)

)

I think cross or slide pairing is better than fold pairing in the first rounds

But also that accelerated pairings would be even better :blackeye:

Posted: **Sun Nov 21, 2010 3:11 am**

willemien wrote:But also that accelerated pairings would be even better

There is no good reason to ever use accelerated pairings in a McMahon tournament. ;-)

Posted: **Sun Nov 21, 2010 3:41 am**

HermanHiddema wrote:
willemien wrote:But also that accelerated pairings would be even better

There is no good reason to ever use accelerated pairings in a McMahon tournament.

That is true but many tournaments are not McMahon (for example the one we are discussing here

)

also it the topgroup of a McMahon tournament is big accelerated pairing is a good idea for this group.

The main disadvantage of a McMahon tournament is that players outside the topgroup by definition are prohibited from winning the title.
With accelerated pairing they have (at least) a (theoretical) chance

Posted: **Sun Nov 21, 2010 3:56 am**

willemien wrote:
HermanHiddema wrote:
willemien wrote:But also that accelerated pairings would be even better

There is no good reason to ever use accelerated pairings in a McMahon tournament.

That is true but many tournaments are not McMahon (for example the one we are discussing here )

Robert's post referred to the EGC, which is McMahon.

also it the topgroup of a McMahon tournament is big accelerated pairing is a good idea for this group.

No, if the top group is too big, it should be made smaller. In any situation where the top group is sufficiently large and has sufficiently large differences in skill that AP can be argued for, the same arguments can be used to support making the top group smaller, which is then the better solution.

The main disadvantage of a McMahon tournament is that players outside the topgroup by definition are prohibited from winning the title.

No, they are not. If a player at the EGC starts outside the top group and scores 10/10, he will very probably win the tournament (only way to lose would be a if a top group player had 9/10 and better SOS).

I've seen it happen in an actual tournament that a player from just outside the top group scored 5/5 and shared first place with a 4/5 top group player (the prize money was shared equally between them).

With accelerated pairing they have (at least) a (theoretical) chance

Posted: **Sun Nov 21, 2010 9:09 am**

willemien, the current draft is being consulted by the EGF Committee. Therefore the timing for your request is unsuitable. I can state the prior Rules Commission's major intentions though:

- The text "Fill any remaining places with Europeans according to the EGF rating list." shall be deleted so that, in case of too few strong non-Europeans etc., a supergroup has 24+ instead of up to ca. 32 Europeans.
- Recommended Cross Pairing for McMahon rounds 1 + 2, Fold Pairing for later rounds.
- On the second Wednesday, i.e., after round 7 and before the new KO for the EC title, relegation games can occur as follows: 1) Number the top 16 Europeans by MMS-SOS-rating-lottery. 2) Fold Compare them. 3) Of each compared pair, seed the player with greater MMS and send back the other player to the main T. 4) Compared pairs with equal MMS play a relegation game; winner qualifies for KO, loser rejoins the main T.
- Minimize repeated pairings.
- KO has also 3rd place game.
- Losers of relegation and KO quarter finals play all remaining main T rounds (to give non-Europeans enough interesting opponents).
- Only non-Europeans can become EuOpenCh and only if they are in top MMS group after round 10.
- Details for main T final tiebreakers.

Posted: **Sun Nov 21, 2010 4:21 pm**

RobertJasiek wrote:willemien, the current draft is being consulted by the EGF Committee. Therefore the timing for your request is unsuitable. I can state the prior Rules Commission's major intentions though:

I disagree with you here, is not better to have a lot of discussion about it?
(Or should we all join the rules comitee?)

t

...

- On the second Wednesday, i.e., after round 7 and before the new KO for the EC title, relegation games can occur as follows: 1) Number the top 16 Europeans by MMS-SOS-rating-lottery. 2) Fold Compare them. 3) Of each compared pair, seed the player with greater MMS and send back the other player to the main T. 4) Compared pairs with equal MMS play a relegation game; winner qualifies for KO, loser rejoins the main T.

what do you mean here?
you take the top 16 european players in the tournament, (not just the top 8)

you compare 1-16 2-15 3-14 4-13 5-12 6-11 7-10 and 8-9(this is fold comparing)

and then take the better in each pair?
(why not take just the top 8 straigt away?) :shock:

- Minimize repeated pairings.
- KO has also 3rd place game.

This differs from what you said in the rules forum (but that was not related to this here off course ;-)

)

- Losers of relegation and KO quarter finals play all remaining main T rounds (to give non-Europeans enough interesting opponents).
- Only non-Europeans can become EuOpenCh and only if they are in top MMS group after round 10.

In the last EGC for the first time an european won the open title, and now we make it impossible. (that is no good publicity) :blackeye:

correction in the fold comparing (sorry there was a mistake)

Posted: **Sun Nov 21, 2010 6:35 pm**

willemien wrote:I disagree with you here, is not better to have a lot of discussion about it?

You don't want to discuss all the linguistic typos anyway.

what do you mean here?
you take the top 16 european players in the tournament, (not just the top 8)

you compare 1-16 2-15 3-14 5-13 6-12 7-12 8-11 and 9-10 (this is fold comparing)

You got it, except for a typo of omitting #4 and the ensuing typos.

and then take the better in each pair?

Yes, IF there is a better one by MMS.

Here are some samples:

The first row of every block of two rows denotes how many wins less than player #1 the players have.

+ = qualified automatically
X = winner of same letter pairing qualifies
- = not qualified automatically

Code: Select all

0 1 1 1 1 1 1 1 2 2 2 2 2 2 2 2
+ + + + + + + + - - - - - - - -

0 1 1 1 1 1 1 2 2 2 2 2 2 2 2 2
+ + + + + + + A A - - - - - - -

0 1 1 1 1 1 2 2 2 2 2 2 2 2 2 2
+ + + + + + A B B A - - - - - -

0 1 1 1 1 2 2 2 2 2 2 2 2 2 2 2
+ + + + + A B C C B A - - - - -

0 1 1 1 2 2 2 2 2 2 2 2 2 2 2 2
+ + + + A B C D D C B A - - - -

0 1 1 2 2 2 2 2 2 2 2 2 2 2 2 2
+ + + A B C D E E D C B A - - -

0 1 2 2 2 2 2 2 2 2 2 2 2 2 2 2
+ + A B C D E F F E D C B A - -

0 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2
+ A B C D E F G G F E D C B A -

0 0 0 0 1 1 1 1 2 2 2 2 2 2 2 2
+ + + + + + + + - - - - - - - -

0 0 0 0 1 1 1 2 2 2 2 2 2 2 2 2
+ + + + + + + A A - - - - - - -

0 0 0 0 1 1 2 2 2 2 2 2 2 2 2 2
+ + + + + + A B B A - - - - - -

0 0 0 0 1 2 2 2 2 2 2 2 2 2 2 2
+ + + + + A B C C B A - - - - -

0 0 0 0 2 2 2 2 2 2 2 2 2 2 2 2
+ + + + A B C D D C B A - - - -

Cases with some 3s instead of 2s would look similar.

(why not take just the top 8 straigt away?)

For these reasons:
- With the Fold Comparison and the possible relegation games, each seeded player has won at least one game more than each not seeded player. This is a very good seeding criterion!
- EDIT: It becomes much less likely that players with 2 wins less than the top seeded Europeans could qualify. :EDIT
- SOS as seeding criterion would be much worse than depending only on greater numbers of wins (aka greater MMS).
- SOS would invite easy cheating as allegedly happended similarly in the EGC 1987 when same-nation players were strongly suspected to have lost intentionally against other-nation competitors for the title.

:shock:

Feel the joy of seeing the best possible criterion (greater wins / MMS) to be meaningfully and exclusively used at the decisive seeding in Europe's most important tournament!

This differs from what you said in the rules forum (but that was not related to this here off course

I am not exactly sure what you refer to.

In the last EGC for the first time an european won the open title, and now we make it impossible. (that is no good publicity)

Would you prefer the 5th strongest European to get the Open title while Europeans 2 to 4 do not get a title?

That the Open title does not make much sense for a European holder any longer is a drawback of the last AGM's decision to invent the European KO finals while keeping the schedule pretty much unchanged.

Hm, maybe there is an alternative: The KO game results need to be copied into the McMahon anyway for the opponents' SOS values. So one might allow the top European in the McMahon to get the Open title. - IMO, such does not make much sense, either: The separation of the top 4 Europeans is too great to compare their own final SOS meaningfully with the top non-Europeans' SOS.

So the sheer honour (eh, and prize money) of a European being placed above the non-European Open EC must be good enough a reward.

Posted: **Tue Nov 23, 2010 5:09 am**

RobertJasiek wrote:
willemien wrote:I disagree with you here, is not better to have a lot of discussion about it?

You don't want to discuss all the linguistic typos anyway.

I agre e lets bnot do that :cool:

RobertJasiek wrote:
willemien wrote:what do you mean here?
you take the top 16 european players in the tournament, (not just the top 8)

you compare 1-16 2-15 3-14 5-13 6-12 7-12 8-11 and 9-10 (this is fold comparing) << LTER CORRECTED

You got it, except for a typo of omitting #4 and the ensuing typos.

and then take the better in each pair?

Yes, IF there is a better one by MMS.

Here are some samples:

The first row of every block of two rows denotes how many wins less than player #1 the players have.

+ = qualified automatically
X = winner of same letter pairing qualifies
- = not qualified automatically
Code: Select all
0 1 1 1 1 1 1 1 2 2 2 2 2 2 2 2
+ + + + + + + + - - - - - - - -

0 1 1 1 1 1 1 2 2 2 2 2 2 2 2 2
+ + + + + + + A A - - - - - - -

0 1 1 1 1 1 2 2 2 2 2 2 2 2 2 2
+ + + + + + A B B A - - - - - -

0 1 1 1 1 2 2 2 2 2 2 2 2 2 2 2
+ + + + + A B C C B A - - - - -

0 1 1 1 2 2 2 2 2 2 2 2 2 2 2 2
+ + + + A B C D D C B A - - - -

0 1 1 2 2 2 2 2 2 2 2 2 2 2 2 2
+ + + A B C D E E D C B A - - -

0 1 2 2 2 2 2 2 2 2 2 2 2 2 2 2
+ + A B C D E F F E D C B A - -

0 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2
+ A B C D E F G G F E D C B A -

0 0 0 0 1 1 1 1 2 2 2 2 2 2 2 2
+ + + + + + + + - - - - - - - -

0 0 0 0 1 1 1 2 2 2 2 2 2 2 2 2
+ + + + + + + A A - - - - - - -

0 0 0 0 1 1 2 2 2 2 2 2 2 2 2 2
+ + + + + + A B B A - - - - - -

0 0 0 0 1 2 2 2 2 2 2 2 2 2 2 2
+ + + + + A B C C B A - - - - -

0 0 0 0 2 2 2 2 2 2 2 2 2 2 2 2
+ + + + A B C D D C B A - - - -
Cases with some 3s instead of 2s would look similar.

I don't understand this (Or do i try not to understand this)
you introduce here an extra round to decide who of the A B C and sometimes D goes to the final KO rounds?

But also there is a problem with the sorting allready.

How do you sort the players even before you decide who is welcome to this (extra qualification round)
If you introduce an extra round anyway there are possibly better ways. (let every european join with just enough wins)

(why not take just the top 8 straigt away?)

For these reasons:
- With the Fold Comparison and the possible relegation games, each seeded player has won at least one game more than each not seeded player. This is a very good seeding criterion!
- EDIT: It becomes much less likely that players with 2 wins less than the top seeded Europeans could qualify. :EDIT
- SOS as seeding criterion would be much worse than depending only on greater numbers of wins (aka greater MMS).
- SOS would invite easy cheating as allegedly happended similarly in the EGC 1987 when same-nation players were strongly suspected to have lost intentionally against other-nation competitors for the title.

Feel the joy of seeing the best possible criterion (greater wins / MMS) to be meaningfully and exclusively used at the decisive seeding in Europe's most important tournament!
[/quote]

If only i saw it that way

BTW i guess you mean by seeded the players who are in the top 8?

This differs from what you said in the rules forum (but that was not related to this here off course

I am not exactly sure what you refer to.

This one : viewtopic.php?f=45&t=2410

In the last EGC for the first time an european won the open title, and now we make it impossible. (that is no good publicity)

Would you prefer the 5th strongest European to get the Open title while Europeans 2 to 4 do not get a title?

That the Open title does not make much sense for a European holder any longer is a drawback of the last AGM's decision to invent the European KO finals while keeping the schedule pretty much unchanged.

Hm, maybe there is an alternative: The KO game results need to be copied into the McMahon anyway for the opponents' SOS values. So one might allow the top European in the McMahon to get the Open title. - IMO, such does not make much sense, either: The separation of the top 4 Europeans is too great to compare their own final SOS meaningfully with the top non-Europeans' SOS.

So the sheer honour (eh, and prize money) of a European being placed above the non-European Open EC must be good enough a reward.

I hope the champion will see it that way. But you see the problem. :grumpy:

Maybe it is an idea to leave the decision to decide who is the Strongest European (professional) go-player to the [sl=EuropeanGoProfessionals]European Go Professionals[/sl] and only bother on the congress about the open Champion (who may also be an european)

Posted: **Tue Nov 23, 2010 5:48 am**

willemien wrote:you introduce here an extra round to decide who of the A B C and sometimes D goes to the final KO rounds?

Yes.

But also there is a problem with the sorting allready.

How do you sort the players even before you decide who is welcome to this (extra qualification round)

I repeat: Number the top 16 Europeans by MMS-SOS-rating-lottery.

If you introduce an extra round anyway there are possibly better ways. (let every european join with just enough wins)

You speak of ways in plural. Which ways do you consider better and why better?

Why would "let every european join with just enough wins" be a better way? We need to determine EXACTLY 8 seeded players in (for practical reasons) AT MOST 1 extra round. Under these requirements, it is impossibe to "let every european join with just enough wins".

If only i saw it that way

Take your to time to acquire and share that opinion:)

BTW i guess you mean by seeded the players who are in the top 8?

From the McMahon to the KO, there is a seeding process after which the exactly 8 participants of the KO are the seeded players.

This one : viewtopic.php?f=45&t=2410

There I meant the minimization once the KO has started. I imply that you are pointing to the aspect that playing relegation games at all increases the average expected minimal number of repeated pairings in the KO. Right. Now if you ask me which I prefer - seeding of only such players with more wins than not seeded players versus yet further minimizing repetitions, then I prefer the former by far.

Maybe it is an idea to leave the decision to decide who is the Strongest European (professional) go-player to the [sl=EuropeanGoProfessionals]European Go Professionals[/sl] and only bother on the congress about the open Champion (who may also be an european)

The EGF does not need to resign its standard of issuing the EC title just because there will be also League Professionals now. They might choose to create another title like Top League Professional, if they like. The more titles there are, the better everybody can try to judge who indeed might be the currently strongest player.

***

Can you please not make very long quotations but save me from doing your work of compressing your messages? Thanks.

Posted: **Tue Nov 23, 2010 10:57 pm**

RobertJasiek wrote:I imply that you are pointing to the aspect that playing relegation games at all increases the average expected minimal number of repeated pairings in the KO. Right. Now if you ask me which I prefer - seeding of only such players with more wins than not seeded players versus yet further minimizing repetitions, then I prefer the former by far.

You successfully confused me:) The problem does not even exist because the relegation games do not increase the number of games that could be repeated in the KO!

Posted: **Wed Nov 24, 2010 1:02 am**

RobertJasiek wrote:
RobertJasiek wrote:I imply that you are pointing to the aspect that playing relegation games at all increases the average expected minimal number of repeated pairings in the KO. Right. Now if you ask me which I prefer - seeding of only such players with more wins than not seeded players versus yet further minimizing repetitions, then I prefer the former by far.

You successfully confused me:) The problem does not even exist because the relegation games do not increase the number of games that could be repeated in the KO!

If the pairing of the Quarter Finals can influence the probability of repeat pairings in the Semi Finals, then the pairing of the relegation games can influence the probability of repeat pairings in the QF and SF, I would think?

Posted: **Wed Nov 24, 2010 4:41 am**

Oh... right! If the relegation kills the potential heavy repetition pairing opponents suitably, then fewer expected repetitions will occur in the KO - and vice versa. IOW, on average relegation is pretty much neutral WRT to repeated pairings in the KO.

But... ultra-meticulous relegation pairing can help the KO yet further...

Posted: **Mon Nov 29, 2010 3:12 am**

RobertJasiek wrote:
You speak of ways in plural. Which ways do you consider better and why better?

Why would "let every european join with just enough wins" be a better way? We need to determine EXACTLY 8 seeded players in (for practical reasons) AT MOST 1 extra round. Under these requirements, it is impossibe to "let every european join with just enough wins".

I still prefer the hidden option see
viewtopic.php?p=14403#p14403 (and following)
your knock-out and extra round can be the final part of it.

(The main thing is that only european - eurpean games count and that the games are also part of the open tournament)

RobertJasiek wrote:
Can you please not make very long quotations but save me from doing your work of compressing your messages? Thanks.

I will

Can you refrain from using abbrevations?

Life In 19x19

EWGC-2010

Re: EWGC-2010

Re: EWGC-2010

Re: EWGC-2010

Re: EWGC-2010

Re: EWGC-2010

Re: EWGC-2010

Re: EWGC-2010

Re: EWGC-2010

Re: EWGC-2010

Re: EWGC-2010

Re: EWGC-2010

Re: EWGC-2010

Re: EWGC-2010

Re: EWGC-2010

Re: EWGC-2010