Wisdom from NagaSkaki homepage

[Matthias Gemuh]

Geschrieben von: / Posted by: Matthias Gemuh at 09 March 2004 18:19:25:


"Don't take things too seriously - if you get tired, stop programming until you're motivated again (even if it takes years!)"

I feel like obeying immediately !!
/Matthias.




BigLion + Taktix

[Tord Romstad]

Geschrieben von: / Posted by: Tord Romstad at 09 March 2004 19:45:54:
Als Antwort auf: / In reply to: Wisdom from NagaSkaki homepage geschrieben von: / posted by: Matthias Gemuh at 09 March 2004 18:19:25:

"Don't take things too seriously - if you get tired, stop programming until you're motivated again (even if it takes years!)"
I feel like obeying immediately !!

This is indeed wisdom, but unfortunately it is virtually impossible to follow,
once you have made your engine public. All these pesky users keep sending us
nice games, bug reports, requests for new features, as well as asking when the
next version will be available. I often wish I could let my engine die and
never again touch my ugly and rotten old source code, but it's too late by
now.
And you think we'll let you do that? Dream on.
Tord

[Uri Blass]

Geschrieben von: / Posted by: Uri Blass at 09 March 2004 20:29:45:
Als Antwort auf: / In reply to: Re: Wisdom from NagaSkaki homepage geschrieben von: / posted by: Tord Romstad at 09 March 2004 19:45:54:

"Don't take things too seriously - if you get tired, stop programming until you're motivated again (even if it takes years!)"

This is indeed wisdom, but unfortunately it is virtually impossible to follow,
once you have made your engine public. All these pesky users keep sending us
nice games, bug reports, requests for new features, as well as asking when the
next version will be available. I often wish I could let my engine die and
never again touch my ugly and rotten old source code, but it's too late by
now.

If you look at all the games that people give you then you have no time to work on the engine and the natural result is that your engine die.
I do the mistake of looking too much in games of my engine.
I do not look at all the games that are available but what I do is too much and it means less time to work on the engine.
Unfortunately it is more interesting for me to look at games than to write code for my engine.
Uri

[Matthias Gemuh]

Geschrieben von: / Posted by: Matthias Gemuh at 09 March 2004 21:04:07:
Als Antwort auf: / In reply to: Re: Wisdom from NagaSkaki homepage geschrieben von: / posted by: Tord Romstad at 09 March 2004 19:45:54:

"Don't take things too seriously - if you get tired, stop programming until you're motivated again (even if it takes years!)"
I feel like obeying immediately !!

This is indeed wisdom, but unfortunately it is virtually impossible to follow,
once you have made your engine public. All these pesky users keep sending us
nice games, bug reports, requests for new features, as well as asking when the
next version will be available. I often wish I could let my engine die and
never again touch my ugly and rotten old source code, but it's too late by
now.
And you think we'll let you do that? Dream on.
Tord

BigLion + Taktix

[Matthias Gemuh]

Geschrieben von: / Posted by: Matthias Gemuh at 09 March 2004 21:37:05:
Als Antwort auf: / In reply to: Re: Wisdom from NagaSkaki homepage geschrieben von: / posted by: Tord Romstad at 09 March 2004 19:45:54:

"Don't take things too seriously - if you get tired, stop programming until you're motivated again (even if it takes years!)"
I feel like obeying immediately !!

This is indeed wisdom, but unfortunately it is virtually impossible to follow,
once you have made your engine public. All these pesky users keep sending us
nice games, bug reports, requests for new features, as well as asking when the
next version will be available. I often wish I could let my engine die and
never again touch my ugly and rotten old source code, but it's too late by
now.
And you think we'll let you do that? Dream on.
Tord

Hi Tord,
thanks for the Gothmog with fritz-hash option.
That means I can now use my favorite UCI engine under my favorite GUI.
As for programming, I am too lazy to start from scratch again and too stupid to fix the mighty bugs in my search. Frequently, when I search twice under identical conditions, I get 2 different "best" moves and miss the best move.
Against all opponents below Elo 2250, the performance of BigLion can be quickly determined by simply tossing a coin.
I don't program much now. When my search shall be stable, I shall limit my
chess programming to "Evaluation and Fun Features". No search improvements.
Thanks again for Gothmog046,
Matthias.


BigLion + Taktix

[Slobodan R. Stojanovic]

Geschrieben von: / Posted by: Slobodan R. Stojanovic at 09 March 2004 22:15:15:
Als Antwort auf: / In reply to: Re: Wisdom from NagaSkaki homepage geschrieben von: / posted by: Uri Blass at 09 March 2004 20:29:45:

"Don't take things too seriously - if you get tired, stop programming until you're motivated again (even if it takes years!)"

This is indeed wisdom, but unfortunately it is virtually impossible to follow,
once you have made your engine public. All these pesky users keep sending us
nice games, bug reports, requests for new features, as well as asking when the
next version will be available. I often wish I could let my engine die and
never again touch my ugly and rotten old source code, but it's too late by
now.

If you look at all the games that people give you then you have no time to work on the engine and the natural result is that your engine die.
I do the mistake of looking too much in games of my engine.
I do not look at all the games that are available but what I do is too much and it means less time to work on the engine.
Unfortunately it is more interesting for me to look at games than to write code for my engine.
Uri

Not "unfortunately", it is _fortunately_. This passion to look at chess games, common to most of us here, is fortunate. That´s why we can say: GENS UNA SUMUS.
But when we put on the top of our agendas OTHER things, then the story is very, very, different.
SL.

[Tord Romstad]

Geschrieben von: / Posted by: Tord Romstad at 10 March 2004 11:29:34:
Als Antwort auf: / In reply to: Future of BigLion geschrieben von: / posted by: Matthias Gemuh at 09 March 2004 21:37:05:

"Don't take things too seriously - if you get tired, stop programming until you're motivated again (even if it takes years!)"
I feel like obeying immediately !!

This is indeed wisdom, but unfortunately it is virtually impossible to follow,
once you have made your engine public. All these pesky users keep sending us
nice games, bug reports, requests for new features, as well as asking when the
next version will be available. I often wish I could let my engine die and
never again touch my ugly and rotten old source code, but it's too late by
now.
And you think we'll let you do that? Dream on.
Tord

Hi Tord,
thanks for the Gothmog with fritz-hash option.
That means I can now use my favorite UCI engine under my favorite GUI.
As for programming, I am too lazy to start from scratch again and too stupid to fix the mighty bugs in my search.
Frequently, when I search twice under identical conditions, I get 2 different "best" moves and miss the best move.
Against all opponents below Elo 2250, the performance of BigLion can be quickly determined by simply tossing a coin.
I don't program much now. When my search shall be stable, I shall limit my
chess programming to "Evaluation and Fun Features". No search improvements.

Thanks for the kind words, Matthias! I wish I had the pleasure of playing
with your engines more often, too. As you know, I live in a Mac OS/Linux
world and don't have many engines to play against.
I have tried to start from scratch a couple of times myself lately, but
the magnitude of the task is scary. I had forgotten how much work it
takes to produce an engine which plays half decent chess. In a way, I
find it to be even more difficult the second time. When Gothmog played
its first ugly moves, I was happy to see it play chess at all, and I
enjoyed watching the games, even though it lost virtually all of them.
But now, when I already have a moderately strong engine, creating a new
and (initially) awfully weak engine and watching it improve slowly will
be unbearably painful. Nevertheless, sooner or later the difficulties
of maintaining my current, badly written code and struggling with old,
fundamentally brain-damaged design decisions will force me to make a
fresh start.
Concerning the mighty bugs in your search, the best way to fix them is
probably to make your search as simple as possible. Throw away all tricks,
and just use plain alpha-beta. It will be very slow, of course, but
the bugs should be easier to locate and fix. When everything appears to
work, proceed to add all the stuff you removed, but do it very slowly.
Add one thing at a time, and test it very thorougly and try to make sure
there are no bugs before you attempt to add further tricks.
This is very common to me, too. And if I make just a minor, apparently
harmless change to my move ordering, this often results in wildly different
best moves and scores from the same position.
Again, I often feel the same.
That's where I plan to concentrate my attention in the near future, too.
The last half year, I have been working almost exclusively on my search.
It's definitely time for "Evaluation and Fun Features".
Do you have any ideas for fun features, by the way? Is there anything
you would like to see in future versions of Gothmog?
Of course, Matthias is not the only person who is allowed to answer these
questions.
Tord

[Matthias Gemuh]

Geschrieben von: / Posted by: Matthias Gemuh at 10 March 2004 12:03:26:
Als Antwort auf: / In reply to: Re: Future of BigLion geschrieben von: / posted by: Tord Romstad at 10 March 2004 11:29:34:

It's definitely time for "Evaluation and Fun Features".
Do you have any ideas for fun features, by the way? Is there anything
you would like to see in future versions of Gothmog?

I was planning to contact you some day about this, anyway.
I still need a few weeks before my fun ideas are fit to be published.
Regards,
Matthias.

[Tord Romstad]

Geschrieben von: / Posted by: Tord Romstad at 10 March 2004 12:41:39:
Als Antwort auf: / In reply to: Future of Gothmog geschrieben von: / posted by: Matthias Gemuh at 10 March 2004 12:03:26:

It's definitely time for "Evaluation and Fun Features".
Do you have any ideas for fun features, by the way? Is there anything
you would like to see in future versions of Gothmog?

I was planning to contact you some day about this, anyway.
I still need a few weeks before my fun ideas are fit to be published.

Cool -- I look forward to hearing about them.
On the top of my own list of future "Fun Features" is the capability of
emulating human tactical mistakes. What I intend to do is simply to prune
random moves everywhere in the search tree. The probability that a move
will be pruned should be bigger at nodes deep in the search tree than at
nodes close to the root, and moves which tend to be easily overlooked by
humans (moves which appear to lose material, long moves, and so on)
should be pruned more often than obvious moves. The frequency of mistakes
will be configurable from the init file.
Tord

[Matthias Gemuh]

Geschrieben von: / Posted by: Matthias Gemuh at 10 March 2004 12:53:37:
Als Antwort auf: / In reply to: Re: Future of Gothmog geschrieben von: / posted by: Tord Romstad at 10 March 2004 12:41:39:

It's definitely time for "Evaluation and Fun Features".
Do you have any ideas for fun features, by the way? Is there anything
you would like to see in future versions of Gothmog?

I was planning to contact you some day about this, anyway.
I still need a few weeks before my fun ideas are fit to be published.

Cool -- I look forward to hearing about them.
On the top of my own list of future "Fun Features" is the capability of
emulating human tactical mistakes. What I intend to do is simply to prune
random moves everywhere in the search tree. The probability that a move
will be pruned should be bigger at nodes deep in the search tree than at
nodes close to the root, and moves which tend to be easily overlooked by
humans (moves which appear to lose material, long moves, and so on)
should be pruned more often than obvious moves. The frequency of mistakes
will be configurable from the init file.
Tord

BigLion + Taktix