Author: Robert Hyatt
Date: 11:42:11 06/09/04
Go up one level in this thread
On June 09, 2004 at 08:50:30, Uri Blass wrote: >On June 09, 2004 at 08:35:04, Peter Berger wrote: > >>On June 09, 2004 at 08:16:49, Uri Blass wrote: >> >>>On June 09, 2004 at 05:36:44, Peter Berger wrote: >>> >>>>On June 09, 2004 at 03:54:25, Mark Winands wrote: >>>> >>>>>2. The tournament will be an 11-round Swiss-system event, using standard >>>>>(non-accelerated) Swiss pairings. The provisional playing schedule was given on >>>>>page 293 of the December issue. >>>> >>>>In case there were only 13 participants I wonder if it would really be possible >>>>to do legal Swiss pairings for 11 rounds - probably not. >>>> >>>>Someone sent me an interesting thought experiment some time ago. >>>> >>>>Assume a Swiss event with six players, A-F - A is the highest seed. >>>> >>>>1st round: >>>> >>>>A-D 1-0 >>>>E-B 0-1 >>>>C-F 0-1 >>>> >>>>2nd round: >>>> >>>>A is paired first, F gets downfloat and has already played C, thus: >>>> >>>>B-A 1/2 >>>>F-E 1/2 >>>>D-C 1/2 >>>> >>>>Standings: >>>> >>>>A:1.5 (WB) >>>>B:1.5 (BW) >>>>F:1.5 (BW,d) >>>>C:0.5 (WB) >>>>D:0.5 (BW) >>>>E:0.5 (WB,u) >>>> >>>>F already had downfloat, so it's B's turn. A gets F, B gets C, so: >>>> >>>>3rd round: >>>> >>>>A-F >>>>C-B >>>>E-D >>>> >>>>Whatever the results of the 3 round now, there are no legal pairings for round >>>>4. >>>> >>>>It is possible that there is some mistake in the above, but I don't think the >>>>principal problem can be avoided. >>>> >>>>Peter >>>It means that >>>A-F >>>C-B >>>E-D >>> >>>is illegal pairing after >>> >>>A-D >>>E-B >>>C-F >>> >>>B-A >>>F-E >>>D-C >>> >>>I expect every intelligent program that calculates pairing not to do that >>>pairing. >>> >>>What is bad about the following pairing for the 3th round? >>>B-F >>>A-C >>>D-E >> >>It isn't the pairing that you will get if you follow the FIDE rules, that's the >>only problem. >> >>Peter >> >>PS: The rules are very complicated so it's possible that I miss something. A >>FIDE arbiter would be helpful ;). > >I think that the first rule should be to allow possible pairing for all the >rounds. > >You can define priorities between possibilities that allow pairing for all the >rounds but allow pairing for all the rounds seems to me logical and it seems to >me logical that every program that gives a pairing should check that there are >possible pairing for all the rounds later and to cancel the pairing as illegal >if there is not legal pairing for the next rounds. > >Uri The problem is too many rounds. There is no possible Swiss pairing rules that will handle 11 rounds with 13 players. Why? Because the Swiss system was created to handle log2(#players) rounds. The minute you pass that number of rounds, pairing problems hit. With 13 players, you are one round short of a half-round-robin. Swiss rules simply don't cover that territory. IE one golden rule is you _never_ play the same player twice. If you play 13 rounds you have to. In this event, if two more programs drop out, things will get ugly. Rounds beyond the point where the top half of the players have already played are pointless. 6-7 reaches that point easily. Leaving 4 rounds where top programs have no viable pairings left except against bottom programs. All that does is extend the event without providing any really interesting games nor useful results. With this number of players, and insisting on playing 11 rounds, I would go to 12 rounds and for each pairing, play two games alternating colors. That would be a much fairer event, with more interesting games, and maintain interest into the last four rounds easily since each program only plays 1/2 of the total, and you don't end up with bottom vs top pairings at the end, or at least not nearly so many...
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.