RobertJasiek wrote:Currently the EGF Rules Commission consists of Matti Siivola (chairman, FI) and me.
Because both are interested in playing in the supergroup themself, it's not honest to allow you and Matti to decide the supergroup size and members. Am I wrong?
Like we say in Russian " Лес охраняешь - дрова имеешь": If you are guarding the forest - you have the wood
breakfast, let me repeat for the 20th time: The EGF Rules Commission does not (let me repeat: NOT) set by itself an important rule like for the supergroup size. The one and only change to that rule since at least 1996 (before I do not know by heart) was by the AGM 2008. Before it was exactly 32 players. I made the proposal, which was accepted by acclamation, that changed that to ca. 32 players (and the number of Europeans to ca. instead of exactly 24 and the number of non-Europeans to ca. instead of exactly at most 8). From 1996 (or earlier) to 2008 no such changes were made; all proposals of changes of the rules were rejected by the AGM. As you see, it is not (let me repeat: NOT) the Rules Commission that changes such important rules by itself. Also the 2008 rule change was not (let me repeat: NOT) set by the Rules Commission on its own but granted by the AGM. (The 2008 change was an important improvement because earlier problems like partly randomly choosing 8 out of 9 top non-Europeans is not necessary any more. Now we may include all 9 players.)
The supergroup size has been tradition for many years and the AGM or EGF Committee has maintained it. The Rules Commission merely applies this state of the rules, except when making proposals (let me repeat: PROPOSALS) to the AGM. It is the AGM (let me repeat: IT IS NOT THE RULES COMMISSION) to adopt such important changes, if any. So stop criticising the (what normally is) executive (EGF Committee and Rules Commission) when you mean to criticise the (what normally is) legislative (AGM)!
You might criticise the current members of the rules commission though (quite like you might also criticise all delegates of the AGM incl. Russian delegates and your own failure to convince them and the EGF Committee for the same reason) that there have not come PROPOSALS / MOTIONS to reduce the supergroup size for the current system. But do not confuse the major executive with the major legislative.
Not honest? Wrong. Quite contrarily: The Tournament Supervisors on request of the Rules Commission (often members of it) have applied the valid rules of the supergroup size. When it was 32, we have set 32 (except for exceptional circumstances due to, IIRC, delayed strong players). When it was ca. 32, we have set ca. 32.
If you want to suggest that there is a conflict of interests between the strong players among the Tournament Supervisors and objective supergroup forming: Yes, there is. We have been aware of it and usually have let third persons double check when some of us supervisors was at the lower boundary of a supergoup. If you consider that insufficient, then please propose capable and rules-firm weak players who want to do the job of the Tournament Supervisors! It would save me of 8 - 20 hours of work every year.
Your proposal of supergroup size reduction should be worked out. Which size? With non-Europeans or not? European-non-European games or not? If with non-Europeans, how many? What would such a rules change have meant it the past or what, by empirical or theoretical studies, would it mean in the future? Why not abandon the young / strong local players condition, which has never provided really strong supergroup members anyway? Why, and this I find particularly interestimg, would the winner not depend on opponent-score tiebreakers when the supergroup size would be reduced?! Currently there are several problems: too few European-only games, too few strong-strong games (really?), too great tiebreaker impact. How does your suggested change solve ALL these problems?
"A super group should be formed of up to 32 players. Up to 8 of the 32 players may be chosen for particular reasons by the EGC Organisers (for instance for publicity, strong youth-players, visiting strong players). Otherwise the group is chosen by the Rules and Ratings Commission in consultation with the EGF Executive."
How do you understand this? They say "up to", not exactly 32. Do you have enough power to reduce the supergroup size without any additional EGF decisions? If so, how about we vote for this decision here?
I suggest 16 players supergroup: 8 best Europeans, 8 best Asians, according to EGF rating. We can also accept 2 strong Finnish players, if there are less than 8 strong Asians. We will see all important games played and the European champion will not be decided on board #50 in last round as we usually see.
kirkmc, first let me clarify the context: In this context, "forum" means "topic section of a discussion medium", e.g., "Announcements" is the forum in which this thread is.
I have not said that I would not like forums. I have said that I dislike the (current) structure of forums on GD and the fact that it is a fixed rather than a dynamic structure.
breakfast, thank you for pointing out the 1996 AGM's statement! This was a few months before I joined the Rules Commission. Of the 1996 AGM, I attended as a kibitz only for a couple of minutes because of playing in a tournament. At that time, AFAIK, AGM decisions were not published on the web yet. In the Rules Commission, none of the older members told me about that AGM decision. During tournament supervising, I think nobody recalled that decision. What we did recall was the EGF Fujitsu GP Guidelines / Regulations, which specified the supergroup and had pretty much the same contents, except that somehow the "up to" was lost over the years (or not in the Fujitsu rules; I would have to look up them to find out). Also nobody in the EGF Committee would notice the difference.
Now all that is history. Currently we have the ca. 32 and ca. 8 decision of the AGM 2008, which modifies the long-term practice of (usually) exactly 32 and up to ca. 8, provided a supergroup is formed at all.
So currently the Tournament Supervisors of an EGC have the power to interpret how far the "ca." can be applied. I guess that practice has been conservative (30 - 34) rather than liberal (say, 24 - 40). But in principle Tournament Supervisors might be somewhat liberal now, without needing a new AGM decision. "ca." would not mean "possibly 50% or less" though; there the rules would have to say "extremely roughly 32".
In practice, there would have to be reasons to depart from the standard. E.g., if we have a players field with 24 European or non-European players rated 2500+, then a gap and then lots of players with rating 2440-, I am pretty sure we would draw the line at the gap. However, in practice gaps near the bottom supergroup rating tend to be much smaller, maybe like 15 rating points.
If Matti and I are the Tournament Supervisors, we will probably usually be conservative about the "ca." interpretation. Maybe other Supervisors would be very liberal by default? The Rules Commission in the executive background would tolerate this, I think. We would not tolerate sudden extremes like 16 instead 32 though. This is not covered by "ca. 32" is I assume was also never intended by the 1996 text "up to 32".
Your suggestion 8 + 8 in the supergroup is bad, IMO. First of all, it is not guaranteed in every year to have 8 strong non-Europeans. Secondly, 8 Europeans is by far too small for a 10 rounds McMahon-with-supergroup. If you wanted to have such a tiny number, then you must ensure also a revision of the seeding criteria to have a much higher quality for them. If you suggest "Ca. 16 Europeans, up to ca. 8 non-Europeans, decided by AVERAGE rating during [some suitable period] and with a minimal number of X rated games during that period.", you increase chances that the AGM might consider a proposal of yours more seriously. Otherwise it might just be dismissed as "Alexander's election system":)
Even with such a system, you do not prevent place 1 - place 50 games. E.g., a strong player might have bad luck (very strong opponents) in the first two rounds and he would meet a weak opponent then.
What do you think about flexible number of rounds to determine the winner? Like 9 to 11 rounds, until exactly 1 player has the greatest McMahonScore? Then some top players' too weak opponents in a few rounds don't matter that much.
Something that has not been considered at all but maybe should: For the supergroup opponents, one might set a lower opponent rating bound, e.g., the rating of the (S*3)rd player in initial rating order, where S is the supergroup size. It would mean quite some manual tournament director interference though because I think pairing programs do not offer such a thing.
Robert, if something works badly it's always possible to find people who is responsible and punish them. For example, after Russian team got only few medals during the last Olympiad, lot of directors and presidents lost their places in Local Sport Federations. How do you think, who is responsible for the lottery we have on EGC every year? If it's not you, please tell me his name. I think it's good to prevent him from making any EGC-related decisions in future.
The "lottery" aspect of the current EGC system is caused by several subaspects:
Relevant aspects:
A - Usage of final result tiebreakers that are only marginally better than lottery: SOS etc. (The various(!) side effects of opponent-dependent final result tiebreakers do not have to be considered in detail if only one does acknowlege their principle impact.) B - Top Europeans can be paired against different numbers of top non-Europeans. Whether this is a lottery problem (preference whether top Europeans like or dislike many such games is an unrelated aspect) changes every year. In some years, top non-Europeans have very similar strengths to top Europeans and with a very similar distribution. In other years, top non-Europeans might be super-strong and then it really matters how many such opponents a European has to play against. Since the latter is possible at all, we need to consider this lottery aspect. C - Some supposedly top non-Europeans do not have an EGF rating. This creates uncertainty whether a) they should be in the supergroup and b) how strong a European's opposition really will turn out to be.
Immaterial aspects:
- The seeding to the supergroup relies on weak data (any number of rated games suffices, only the moment just before the tournament is considered). Although this is relevant to some extent, if the supergroup is big enough, them we can as well assume that it is almost irrelevant for the stronger half of the supergroup members. - Europeans having lived overseas might need to earn a new EGF rating. If they really are strong, then that should be easily possible for them. - Some players cannot afford to travel to the tournament. This is a relevant issue but only marginally on the level of tournament system. - The top players' field is a bit too thin or thick: All players are pretty equally affected by that. - I think that the varying strength of the players' opposition can be neglected because the typically used pairing programs try to make SOS values of players similar. To really see clear causes of still remaining differences requires lots of research, which has not been done yet. So we may as well simplify here and ignore this lottery aspect. (In tournaments with only a few rounds, matters are very different though. An EGC with a too big number of top non-Europeans would also still create problems, but this is a special case.) - A player is paired against his angst opponent. This is the player's psychological responsibility rather than a fault of the tournament system. - A player gets the "wrong" color against a specific opponent. It is the player's responsibility to play well with both colors. - A player prefers a different komi, scoring, thinking time, starting time of a round etc. Maybe relevant for the individual but we cannot do much by trying to replace the tournament system.
Responsibility for the relevant aspects:
It is rather nebulous who might be responsible for maintenance of aspects B and C. Besides these aspects are not as relevant as A. So the major aspect for which responsibility should be studied is A. Responsible are all those delegates and their direct or indirect supporters that have rejected, have delayed or reject tournament systems for that A plays a role at all. IMO, this lets most delegates and the EGF Committee be responsible. Also many supporters of maintaining SOS as the first tiebreaker.
It is easier to say who is not responsible: Those having made system change proposals so that opponent-dependent final result (or relevant seeding to a next tournament stage) tiebreakers are not needed.
SOS, SOS-1, SOS-2, can I get a link to the different kind of tiebreakers these are?
is SOS-2 SODOS?
Breakfast: I dunno, you've been in more situations where tiebreakers count, which method of tie-breaking do you prefer?
(P.S. I love the quote...)
I only ask because, my impression is, as a player who's been on the low end of American supergroups. It's hard to reduce the supergroup without also reducing excitement about the tournament. People like having a hometown hero in the mix, even if they know they're going to get smashed.
I feel like a good implementation of SOS would be less damaging to the format than cutting players out of the running...
# SOS-1 = SOS, where 1 round with the smallest value is ignored. # SOS-2 = SOS, where 2 rounds with the smallest values are ignored.
***
"a good implementation of SOS" is: Do not use it for the final results ordering. A player should always be responsible for his final placement, which contradicts that he cannot influence his SOS value well.
I could understand taking out earlier rounds, but taking out the lowest round feels weird.
Suppose you had a very long (but not round robin) tournament, with a large number of players, each player would converge to their appropriate placing pretty quickly no? even with incomplete information. But suppose you stop it before it's "done", and then weight the earlier round information the same as the later round information, then you're overcompensating for earlier rounds.
Why? By virtue of the fact that the existing information that can be gleaned about each player's performance is already, in a sense, locked up in who they're playing in later rounds.
for example, suppose we have the following performance:
(1's are wins, 0's are losses. Number below is current rank of opponent)
1 1 0 1 1 10 8 4 6 3
Now, I've heard of people subtracting the first round, but it seems to me that if we know that this player is playing against player number 4 in the third round, then that already tells us a lot of information about the player, the result of this game is more significant than the game preceeding it.
So why not establish a statistic which starts with:
Tiebreak one: Sum of scores (weighted by significance) Tiebreak two: Sum of opponent's weighted scores etc...
Look at some earlier EGC final result tables for top players! You will notice that low SOS values of weakly performing opponents occur in up to ca. 2 rounds and do NOT always occur for the first two rounds (for some players, their lowest SOS contribution is in some seemingly unpredictable rounds). Therefore SOS-2 is the opponent-score dependent tiebreaker with the greatest meaning. SOS-1, SOS-3, SOS-R1, SOS-R2, SOS-R3 (Rx meaning: drop first x rounds), SOS each has less meaning.
What do you mean with "weighted by significance"? Which significance?
