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

BGonline.org Forums

Clarification about Janowski's formulas

Posted By: Timothy Chow
Date: Saturday, 23 January 2010, at 8:39 p.m.

I was a bit puzzled as to why NJ felt compelled to post (on 18 January) a clarification of his views on Janowski's formulas, but I understand now, having finally checked my Hotmail email account. For future reference, if you want me to respond quickly to your email, please use this email address instead of my Hotmail address (but please don't post that email address here because I'm trying to limit spam).

In case anyone has misinterpreted my view of Janowski's work, I will state clearly that it is extremely valuable. Plainly Janowski's formulas still form the basis for the cubeful evaluations of today's strongest bots. This goes without saying, which is why I didn't bother saying it. At some point, the value of certain contributions becomes so obvious that we take them for granted. I probably think about Edison at most once every several thousand times I use an incandescent light bulb.

The point I've been trying to make is that cubeful evaluations based on Janowski's formulas can't compare in accuracy to rollouts. Making an educated guess about the cube efficiency and applying Janowski's formula is inherently a kludge. It's an extremely clever kludge, of course, and I don't know of any better way to get a cubeful evaluation without a rollout, but compared to a full live-cube rollout it's still a kludge.

Conversely, if we are willing to invest the computational time to do a full live-cube rollout with checker play ATS, then it strikes me that we should really be looking at the cube statistics of that rollout to understand cube efficiency and recube vig. Why guess at the cube efficiency when the bot can give you direct estimates of it from its rollout? This is why I am lobbying for better support for the "View Statistics" feature of GNU.

Hopefully these remarks will help clarify what I have been trying to say. I apologize for any misunderstanding that may have been caused by poor choices of words in my previous posts.

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.