Author: Vincent Diepeveen
Date: 12:18:46 03/02/04
Go up one level in this thread
On March 02, 2004 at 14:22:35, Gerd Isenberg wrote: I feel the basic point made is that swiss pairing is already a very complicated way to do pairing. Pair first 3 rounds upon seeding and rounds 3..11 upon sum of opponents (buchholz in german called or weerstandspunten in dutch). The swiss rules are so very complicated already and take *so many* fair considerations into account that it is very complicated to write a correct pairing swiss program. It's more difficult than writing a perfect mate solver that solves all mates up to 20 moves at current hardware. >On March 02, 2004 at 09:53:15, Omid David Tabibi wrote: > >>Based on the feedback to the previous thread >>(http://talkchess.com/forums/1/message.html?352100), it seems that the majority >>prefer 11 rounds Swiss to the knockout with reentry option. >> >>Gerd Isenberg suggested an interesting modification to standard Swiss: >> >>-------- >>http://talkchess.com/forums/1/message.html?352264 >> >>What about following slightly modified [Swiss system], a kind of "delayed" swiss >>system to keep the tension a bit longer? >> >>For the first (4-6) rounds playing swiss with two about equal strong groups with >>about (even) quantity and quality, eg. >>(1,4,5,8,9,12,13,16)<->(2,3,6,7,10,11,14,15) or similar. >> >>Then reunion both groups and continue with standard swiss, and there are still >>some thrilling rounds to go. >>-------- >> >>In other words, a simple 11 rounds Swiss, but keeping some stronger pairings for >>the later rounds. >> >>What do you think would be better? Standard 11 rounds Swiss (like in Graz WCCC), >>or the delayed Swiss as suggested by Gerd? > >Obviously there are some practical considerations, as Vincent pointed out. >One problem might be the initial ranking (even in the lower part of the field), >the other problem is whether a pairing program is able to handle such pairing >restrictions during the first m of 11 rounds. > >Next problem is to determine the "optimal" m, the number of rounds with disjoint >groups, to delay important matches - or whether it introduces more dependencies >as suggested. And as already mentioned, all is dependent on the number of >participants N. > >If N is 16, both groups are even (8) and there is no group internal "by". >Then m == 7 is the maximum because it's already group internal round robin. >If N is not divisable by four it becomes complicated, odd/even groups, one or >two internal "by" (in case of two they may play each other as an other >exception). > >Seems all too complicated and probably exhausting for the TD ;-) >Anyway, thanks for considering the idea. > >I strongly consider to participate, independent on delayed or classical swiss >system. > >Regards, >Gerd
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.