[ View Thread ] [ Post Response ] [ Return to Index ] [ Read Prev Msg ] [ Read Next Msg ]

BGonline.org Forums

Whatever happened to Zare, ZBot, Chow..?

Posted By: Timothy Chow
Date: Monday, 29 June 2026, at 1:15 p.m.

In Response To: Whatever happened to Zare, ZBot, Chow..? (MK)

R. B. Sahi wrote:

Why don't you invite your "colleague" here to start up a thread about it?

There's nothing much to discuss until he has concrete results to share. I will say, though, that because he was interested in the doubling cube, he did one preliminary experiment, with an extremely simplified version of backgammon where there was very little to decide other than if and when to double. He found that if he coded up the problem naïvely, AlphaZero was not able to learn good cube handling. It would double randomly, which meant doubling a lot. That meant that the scores would vary wildly from one game to the next, creating so much "noise" that it drowned out the "signal." In order to get any meaningful results, he had to tweak some parameters, e.g., I think he tried setting the initial default probability of doubling to some low number, to discourage it from doubling too aggressively.

This simple experiment already demonstrates that if R. B. Sahi has some romanticized idea that AlphaZero has "zero influence from human expertise" then that expectation should be tempered somewhat. There is, at minimum, a lot of jiggling around of hyperparameters and training algorithms that goes into developing a world-class bot. This is one reason why, even after the AlphaZero paper was published, it took the chess world a long time to replicate the performance of AlphaZero Chess. Of course, part of it was that DeepMind had a ton of computing resources that it could throw at the problem, but it was also because the AlphaZero algorithm is a lot more complicated than press releases might have you believe. The paper did not fully document all the human expert choices that went into the actual program that they ran. It is true that they did not directly use human expert heuristics for playing each game, but that is not quite the same as zero human input. I'm sure there were many, many iterations of "try this, and if the bot plays like crap, change something and try again."

Messages In This Thread

 

Post Response
Your Name:
Your E-Mail Address:
Subject:
Message:

If necessary, enter your password below:

Password:
 

 

[ View Thread ] [ Post Response ] [ Return to Index ] [ Read Prev Msg ] [ Read Next Msg ]

BGonline.org Forums is maintained by Stick with WebBBS 5.12.