[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: computer-go: Grading Go Problems
On Tuesday, July 31, 2001, at 07:16 AM, David Fotland wrote:
We could start the grading by looking at which problems are solved by
existing programs. If I get access to the
problem set,
Thomas sent me an email saying he gave you access.
I can tell you which problems Many Faces gets correct. It will take
about a couple of days to run through
such a big problem set. Many Faces does about 1000 life and death
problems an hour.
Exactamundo. I was hoping that those of you with working, or partially
working go or tsume-go programs could sign up to get access to Thomas's
full set and give me a list like this:
Problem Rating Comment
5003002 65 Ko
5003003 - Can't solve
The - in rating means it couldn't be solved. The rating could be as
simple as the time to solve the problem in seconds, or number of nodes
or something, just make larger numbers harder. You can report negative
numbers for hard problems you couldn't solve.
Comment can be anything you want to report that might help me choose
problems like "Eye Shape #67 or any sort of information about the
problem you have.
Then I can use that information both in the final set, and while
choosing problems. For instance, if God smiles on me and 3! people come
forward and do this, I could slice the problems such that there are 5
difficulty grades (by rating), and then inside each grade are three sets
based on how many programs could solve that set (at this moment, I bet
after the final set comes out, most people will be able to solve the
problems).
For those of you that don't want to get access to the full set, but are
willing to rate the final set, I can include the rating along with the
problem in a later index file, so that set users can rearrange however
they like.
Perhaps even a pro will volunteer to grade the problems, and we can have
a "human difficulty" slice...
Then again, I'm an optimist.
Pierce