Winboard Forum

Just want to make a note that I've put a 2.0 beta of Neurosis on my website.
It's the version that played in Leiden last November, but my interest has been low since then, and haven't worked on it since. The 2.0 version of my 3D-gui isn't ready, so it's just a Winboard-executable. You can use it when you have or download the 1.91 zipfile and replace the older Neurosis.exe (the Winboard-compatible file) file with the newer one.
Tactically it's almost the same as 1.91, but I had retuned my evaluation. (most static terms multiplied by two, and other things better balanced.) Tactically it's slower/extremely slow now, (the reason it loses most of it's games..) but positionally it's not the worst. Also it's very stable and I'm happy with it.
I might still finish up my Gui for release again, (It can ponder in N3D now. Tried that last November in Leiden too but then I found a bug in it and had to go back to Winboard again. ) and enter with it in Leiden in May, if not, then this is the last of Neurosis.

http://www.geocities.com/tasteofinsanity81/stanspage

I did start on a new chessprogram a while ago, but that might never come off the ground or take a really long time. Atleast the point now is to be extremely selective and unsafe in search to reach big depths, instead of being extremely safe and conservative, as with Neurosis. (that didn't work too well for chess-strength. )

Stan

Stan Arts wrote:I did start on a new chessprogram a while ago, but that might never come off the ground or take a really long time. Atleast the point now is to be extremely selective and unsafe in search to reach big depths, instead of being extremely safe and conservative, as with Neurosis. (that didn't work too well for chess-strength. )

Hi Stan,
I think it's possible to make a very strong program without using forward pruning. For example, Delfi 4.5 just uses null-moving and lazy evaluation (IIRC). Fabio actually told me that most of Delfi's strength comes from its evaluation.

Hi Alessandro,

Indeed I guess you can make a very strong program with nullmove alone, but when you use an agressive nullmove scheme I consider that to be forward pruning, (atleast you'll delay finding many tactics) Neurosis is more conservative then the average nullmover, my point with it was to almost never delay finding tactics or positional info.
As for Delfi, it seems extremely fast to me (reported nps atleast) and that alone gives 1-2 ply extra already, but usually the difference is even bigger, (Delfi is one of my favourite engines, because it's written in Pascal/Delphi. ) and it seems tactically very strong to me. (Kiwi too, by the way!) I do equal depth matches sometimes (for fun) then indeed Delfi is still really strong but Neurosis then has some chances too.

My new program is also just for experimentation and fun, and trying to be very selective will be interesting, but probably won't be very strong either. (so far it's not even playing chess yet)

Goodluck with Kiwi or Hamsters! May it not get stuck in a loop. (Hamster wheel)



Stan