Computers do not get tired, and they are able to play thousands of games while maintaining the same quality. If the purpose of these competitions are to determine the strongest Go AI, shouldn't they have the AI play as much as 1000 games per round and declare the one with the higher win percentage the winner? Why leave anything to chance with a low sample size?
Between humans, there is a sense of being a competitor and "winning when it matters", but nothing like that matters for AI, so I feel like this makes more sense.
What do you think?
Why do AI competitions play so few games?
-
lightvector
- Lives in sente
- Posts: 759
- Joined: Sat Jun 19, 2010 10:11 pm
- Rank: maybe 2d
- GD Posts: 0
- Has thanked: 114 times
- Been thanked: 916 times
Re: Why do AI competitions play so few games?
Yeah, most tournaments are not about identifying the strongest player. They're more about having a social gathering, providing a spectacle for entertainment/commentary/competition/suspense, etc. This also includes human tournaments in many sports or games or activities, which not uncommonly also have tournament structures that are suboptimal for identifying the strongest player or team, even within constraints on the maximum number of rounds and/or games that could be afforded.
-
ez4u
- Oza
- Posts: 2414
- Joined: Wed Feb 23, 2011 10:15 pm
- Rank: Jp 6 dan
- GD Posts: 0
- KGS: ez4u
- Location: Tokyo, Japan
- Has thanked: 2351 times
- Been thanked: 1332 times
Re: Why do AI competitions play so few games?
At the moment it does not seem that there is enough interest/support for a Go version of the computer chess championship, which has a couple thousand games (if I am reading the website correctly).
Dave Sigaty
"Short-lived are both the praiser and the praised, and rememberer and the remembered..."
- Marcus Aurelius; Meditations, VIII 21
"Short-lived are both the praiser and the praised, and rememberer and the remembered..."
- Marcus Aurelius; Meditations, VIII 21