Author: Don Beal
Date: 15:27:15 02/28/99
Go up one level in this thread
On February 28, 1999 at 12:32:57, Bruce Moreland wrote: >On February 24, 1999 at 21:31:57, Peter McKenzie wrote: >>On February 24, 1999 at 07:57:48, Don Beal wrote: >>>The 9th World Computer Chess Championship takes place in Paderborn >>>June 14-20 1999. >>>This event incoporates the 1999 World Microcomputer Chess Championship - the >>>highest placed program running on a single generally-available >>>microprocessor wins the WMCCC title. >> >>I think this is less than ideal for the reasons outlined below. Not sure what >>can be done about it, probably nothing :-) Perhaps some discussion will throw >>up some ideas. Anyway, the reasons: >> >>1) The swiss tournament system is specifically designed to find a fair >>tournament winner. Once you start looking at the minor places it becomes less >>clear who really 'deserves' second place etc. For a given field size, there is >>a recommended number of rounds to find a fair tournament winner. If you want to >>fairly hand out Nth (where N > 1) prize, that recommended number of rounds will >>typically increase. This stuff is well documented somewhere, probably the FIDE >>rules. Perhaps someone could post the relevant numbers for a 32 player >>tournament? >> >>The whole thing for WCCC/WMCCC is complicated by the fact that we don't even >>know what N will be! Maybe the best micro will win it, in which case there >>won't be a problem. But maybe the best micro will place 5th, in which case >>there is decent chance that it will have actually performed worse than whatever >>placed 6th. >> >>2) There is a dilemma for people who can run on SMP machines (I'm picking that >>at least 4 strong micro programs will face this dilemma). Do they run SMP, >>thereby having a better chance for the WCCC title, but rule out their WMCCC >>chances completely? >> >>Actually both of the above points actually favour my program (if it is >>accepted). Point (1) effectively adds some randomness to the result of WMCCC >>which helps my program as it won't be a favourite. And point (2) will surely >>weaken the competition for WMCCC. >> >>> >>>This a reminder to anyone who was intending to enter, or contemplating >>>entering, that the deadline for making an application is very near (1 March). >>>More information about the event, and the application form, are on the ICCA >>>web pages: http://www.dcs.qmw.ac.uk/~icca/ >>>We expect an exciting tournament, and hope to see as many of you as >>>possible there. >> >>Yes, should be good. Thanks for organising it! >> >>> >>>Don Beal. > >I am disappointed that this didn't receive any response. The proposal here is >to do a 7-round Swiss (probably with 24 entrants) and use it to determine two >champions. > >Assuming that one program doesn't win both the titles, this means that one of >the other places in the tournament would have to "count". Does anyone know >enough about the Swiss system to know if this is really possible? > >bruce Hi Bruce, I'm not sure what response you hoped for. I'd like to be able to post that a friendly millionaire rang the ICCA after reading these messages and offered to pay the costs of a second, separate, tournament, plus extra money to allow more rounds in both. But that didn't happen. Here's the only response I can offer: The swiss tournaments aren't exact science. They are something of a lottery for first place, let alone 5th, or 10th. (Program developers are likely to know that the results from 7 test games could easily be *very* misleading, since the variability is high. It's possible to calculate exact odds given sufficient assumptions about ELO differences.) The championships are not a scientific measurement. They are (in a sense) a biased lottery, in which programs may participate if they wish. The event has to run for a fixed (not too long) length of time, to keep costs down to that which participants and sponsors can accept. Because it's a biased lottery, the better the program one shows up with, the better the chances of winning. But there are significant chances for upsets. I think this has advantages as well as disadvantages. It drives up the motivation to squeeze the very best from the software and hardware. It offers newcomers more motivation too - they might get lucky on the day. It increases excitement for spectators (at the expense of stress for participants). Given all this, it seems to me that it isn't worth worrying about the somewhat greater uncertainties associated with a "best micro" title compared to the "best of all" title. Yes, we could cut out the micro title from this championship, but as we can't offer a separate micro championship this year, we would be reducing the opportunities for those people who have deliberately aimed at that title as opposed to the open title. (And reducing the excitement level.) We can't please everyone. (Unless that millionaire can be found :-)) Don Beal.
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.