Author: Robert Hyatt
Date: 11:28:53 01/18/00
Go up one level in this thread
On January 18, 2000 at 12:47:44, Will Singleton wrote: >On January 18, 2000 at 09:27:32, Robert Hyatt wrote: > >>On January 17, 2000 at 13:34:10, Will Singleton wrote: >> >>> >>>I was thinking about the initial seeding for the ICC tourney (starting Jan 29, >>>ICC, don't miss it), and it seems to me that the hardware should be taken into >>>account. For example, some folks will be running multiple cpus, at least one >>>will be on a 1ghz machine, and some will be as slow as 300mhz. :) >>> >>>So, I'd like to see a list of the hardware for this tourney. And, for those who >>>are familiar with such things, what is the relative importance of the initial >>>seeding in a tourney such as this? >>> >>>+------------------------------------------------+ >>>|Program Main Author ICC Handle | >>>+--------------+------------------+--------------+ >>>|Amateur Will Singleton Amateur | >>>| | >>>|Averno Jos? C M Gal?n | >>>| | >>>|Bringer Gerrit Reubold Bringer | >>>| | >>>|Crafty Robert Hyatt Crafty | >>>| | >>>|Diep Vincent Diepeveen DiepX | >>>| | >>>|EXchess Dan Homan EXchess | >>>| | >>>|Ferret Bruce Moreland Ferret | >>>| | >>>|Galahad James Swafford GalahadX | >>>| | >>>|GnuChess Stuart Cracraft GnuChess | >>>| | >>>|Grok Peter Kappler | >>>| | >>>|Hossa Steffen A. Jakob Hossa | >>>| | >>>|Insomniac James Robertson | >>>| | >>>|LambChop Peter McKenzie LambChop | >>>| | >>>|Little Goliath Michael Borgstaedt | >>>| | >>>|Nimzo Chrilly Donninger Varguz | >>>| | >>>|PostModernist Andrew Williams PostModernist | >>>| | >>>|Shredder Stefan Meyer-Kahlen | >>>| | >>>|Tinker Brian Richardson Tinker | >>>| | >>>|UruChess Nicol?s Carrasco UruChess | >>>| | >>>|ZarkovX John Stanback ZarkovX | >>>+------------------------------------------------+ >> >> >>Seeding doesn't matter if the number of rounds is >= log2(number of players). >>In big events, this isn't possible, so you end up with a log-jam at the top, >>with a bunch of 5 point finishers. Seeding ensures that the highest rated >>player generally has the highest tie-break (sum of opponent's scores or >>whatever is used). When number of rounds is >= log2(players) the probability of >>this happening goes down. In our case, 8 rounds is way larger. It won't matter >>at all... >> >>It would be fun to let everyone seed the thing just before it starts, just to >>see who comes closest, at least in the top half.. > >Yes, perhaps we can get someone to put up a prize for the closest guess. > >Glad to hear the seeding is unimportant in this case. I assume you mean that, >while it remains important to try to order the programs, the *exact* order is >not critical. Or would a random seed be equivalent? > >Will With 8 rounds and 32 or less players, seeding can be 100% random and still produce the best program. And tie-breaks still mean something. Think about an event with 32 players, 16 are GM and 16 are class-A. If you play only 4 rounds, the 16 GMs will win in round 1, then 8 will win in round 2, then 4 in round 3 and finally 2 in round 4. How to break the tie? Higher-rated player should win. Since he will also have the higher tie-break. But if seedings are random, tiebreak will be random. With 8 rounds, that doesn't happen, because the guys at the top, rise to the top, and then knock each other out, since there are enough rounds. It is impossible to have 8 rounds, 32 players, paired properly, and have two undefeated players at the end. Of course, you can have 4 tied for first. :) But the tie-breaks work pretty well even with random seeding. It is always sensible to seed as accurately as possible, but more rounds mean mistakes don't influence the tournament standings as much...
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.