Reflections on the WCCC

[Harvey Williamson]

So ideas to make tie breaks better?

Mark

Arm wrestling between engine authors.

I did suggest that Johannes should play a match against Larry as a tie break. That would have been an interesting match. Johannes has IM norms.

[Rochester]

I think the pity of Mr Baron can not be allowed anymore.
Only the best 6 engines now in the play.
Also pity at the live show I can't see the engine score.
The TEC website is the good example. Lets follow!

[Modern Times]

I certainly like to see the no holds barred monster hardware in the open section, that is very special ! And I value the equal hardware tournament as well. So as long as this event attracts some top engines then it is worthwhile.

[Steve Maughan]

I have a different take on the WCCC.

Last year I entered with Maverick. For those not familiar with Maverick, it's rated about 2500 ELO on the CCRL scale i.e. 800 ELO below the top engines. I knew I had ZERO chance of winning. But as a husband, dad, boss, supplier I have little time for computer chess. So for one week I got to immerse myself in my geeky hobby. I thoroughly enjoyed watching Maverick being crushed by the top engines. I met some people I had "known" for decades but never met. I chatted at the bar about different aspects of computer chess, everything from lazy SMP, to test methodologies - something I never done before. Also everyone was super friendly. There was no tension between the programmers (which I wondered about before the event). All in all, a great week, and having followed this year's event I'm encouraged to work more on Maverick.

In light of my experience I also have some different ideas as to the path I'd like to see the event take. As so many have pointed out, the CCRL doesn't find the definitively best chess engine (nor do the Olympics necessarily find the definitively best athletes). I see the event as a good way of attracting talent into the world of Computer Chess.

So I'd suggest a couple of extra titles / prizes.

1. Best Under 21 Programmer
2. Best New Entrant

The talents needed to create a decent chess program are the same talents which are becoming increasingly valuable to companies working with machine learning and analytics in general. How about trying to get Google / Facebook / Microsoft et al to offer a prize / internship for the best young entrant i.e. sell the sponsorship of the event as a talent acquisition investment?

Anyway I loved taking part, and I hope I can take part again. I encourage anyone with a half decent engine to enter, simply for the experience. I'd certainly be disappointed if the WCCC simply went away.

Steve

[bob]

Robert, how about a 3-1-0 score for WDL?

For the rest, I agree of course.

I've heard of it, but NEVER have played in any event where it was used. I'd likely defer to Larry Kaufman, he would probably understand how this might affect tournaments and whether there is any room for manipulation via pairings...

[Dirt]

Robert, how about a 3-1-0 score for WDL?

Why not 3-2-0? Is fewer losses less important than more wins?

It's a solution, but so is flipping a coin. I'd prefer co-champions, but computers are getting so good that co-champions would be the norm.

[bob]

Robert, how about a 3-1-0 score for WDL?

Why not 3-2-0? Is fewer losses less important than more wins?

It's a solution, but so is flipping a coin. I'd prefer co-champions, but computers are getting so good that co-champions would be the norm.

The point of 3 1 0 is to try to encourage players to play for a win, rather than playing for a safe draw. 3 2 would seem to eliminate that encouragement. A win being as good as three draws is a pretty good incentive. But then a little manipulation of the pairings might allow mischief.

[Dirt]

The point of 3 1 0 is to try to encourage players to play for a win, rather than playing for a safe draw. 3 2 would seem to eliminate that encouragement. A win being as good as three draws is a pretty good incentive.

I think 3 1 0 makes sense for human tournaments, where more wins (and more losses) make for more entertainment. I don't think that is justified for computer vs computer matches. If anything, I think number of losses is a better measure of engine strength.

[bob]

The point of 3 1 0 is to try to encourage players to play for a win, rather than playing for a safe draw. 3 2 would seem to eliminate that encouragement. A win being as good as three draws is a pretty good incentive.

I think 3 1 0 makes sense for human tournaments, where more wins (and more losses) make for more entertainment. I don't think that is justified for computer vs computer matches. If anything, I think number of losses is a better measure of engine strength.

Isn't that the same thing as counting wins and draws significantly more than a loss???

[Dirt]

I think 3 1 0 makes sense for human tournaments, where more wins (and more losses) make for more entertainment. I don't think that is justified for computer vs computer matches. If anything, I think number of losses is a better measure of engine strength.

Isn't that the same thing as counting wins and draws significantly more than a loss???

Roughly, 3 1 0 scoring means a tiebreak goes to the most wins. A 3 2 0 scoring nearly means a tiebreak goes to the fewest losses. More wins are entertaining; fewer losses means, perhaps, a stronger engine.

This is a bit theoretical, though, as I really favor 1 1/2 0 scoring with no tiebreak.