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

BGonline.org Forums

Side question: definition of blunder

Posted By: Timothy Chow
Date: Tuesday, 21 February 2012, at 4:03 p.m.

In Response To: Side question: definition of blunder (Carlo Melzi)

I haven't tampered with XG's settings, but for GNU, I use 0.020 for doubtful, 0.050 for bad, and 0.100 for very bad. I analyze using World Class.

Setting 0.000 for doubtful tends to generate "noise," for example in straight races where the bot sometimes thinks some bizarre move is 0.001 better than the move any human would play. Even if it's correct, which often I think it isn't, who cares?

Regarding Stick's initial question, the answer will vary A LOT from one match to the next. However, at a guess, I'd hypothesize that the person with the 16th lowest PR on the planet makes on average two blunders exceeding 0.080 in a 9-point match, as well as a bunch of smaller errors.

I don't think one can figure this out just from knowing the ER. That just gives you an overall average. But humans tend to make a few big mistakes and a lot of little mistakes, so one also needs some sense of the total distribution.

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.