My apologies... that message was only intended for John.



________________________________
From: Eugene Deon <[email protected]>
To: [email protected]
Sent: Monday, January 26, 2009 8:20:38 PM
Subject: Re: [quackle] maximal leaves


Hey John,

Thank you for the latest leaves.  I just googled superleaves and pulled a text 
file off of Jason's MIT site, i think.  I will rerun with this.

python fucking rocks.  That's all I have to say.  :)

If you have anything I can do to help improve Quackle or Scrabble knowledge in 
general, I have 5 fast cores I can be running on 24/7.  I'm windows only, 
though.  But I can usually get C/C++ code and now some python ported, if need 
be.  To compute the custom playability lists and the hookability list, I just 
edited quackletest to spit out the info I needed, and had a python script parse 
it in 2000 game chunks.  It's not the fastest way, but worked fairly well.

I'd love to quickly get a grip on leave worth, but it doesn't seem nearly as 
compressable as I had hoped.  I'm tempted to compute a "leaveability" list to 
try to sort the list of leaves so i could at least try to study some of them.  
I'd like to be able to guess any leave to within 5 points.  Did your analysis 
reveal any useful tips for a newbie like me?

thanks,
Eugene




________________________________
From: John O'Laughlin <olaugh...@gmail. com>
To: quac...@yahoogroups .com
Cc: crossword-games- p...@yahoogroups. com
Sent: Monday, January 26, 2009 7:09:34 PM
Subject: Re: [quackle] maximal leaves


Eugene,

Nice work! I did some analysis like this a while ago but haven't had
time recently. I think the values you're using aren't the latest. I
apologize for not making a human readable file of the leaves more
widely available (and for continuing to let the original one
propogate, but that's hard to stop). Quackle's current values for TWL,
which are themselves almost two years old now, can be found here

http://pages. cs.wisc.edu/ ~o-laughl/ leaves.zip

Basically these are bootstrapped, using the original set of
superleaves in place of "static" values. If I had lots of computer
(and me) time to spare it would be nice to continue the process until
something of an asymptote is reached.

John
    

Reply via email to