Author: Vincent Diepeveen
Date: 05:01:02 10/30/03
Go up one level in this thread
On October 29, 2003 at 07:57:27, Mogens Larsen wrote:
>On October 28, 2003 at 21:46:57, Vincent Diepeveen wrote:
>
>>the matter is not whether sjeng is better in 1.d4 or 1.e4 only,
>>the matter is that the 1.d4 book for sjeng is also very bad prepared.
>>I can compare some lines it plays with 1.d4 with the Arturo book and see
>>clearly that there a lot of lines played are not exactly very good.
>
>Basically there are two options. Either build a new repetoire for Sjeng or try
>to adopt the book(s) used by Rebel and Tiger. I would probably prefer the former
>option even though it is a time consuming method. Probably more rewarding if the
>engines differ significantly in preferences. Either way, I'm sure that something
>was prepared with 1.e4 and it seemed like a good opportunity to test it.
>Especially if the opponent isn't a top engine.
About option 1 i can clearly explain a number of things having gained some
experience there:
1a) i had to write a lot of new knowledge which wasn't in diep before and
certain bugfixes still are happening. See game DIEP-Tiger
1b) it takes at least 3 years to make a good 1.d4 book
1c) i personally doubt Sjeng will ever play 1.d4 well because of lacking
knowledge which i can enter without problems in DIEP (and even then
have problems tuning it)
1d) Consider that Rebel is more suited towards 1.d4 than Sjeng, so why
choose Sjeng? Rebel uses the same stuff like Tiger
1e) Theory in 1.e4 is moving faster than 1.d4, so for those who already
keep up to date with 1.e4, they can beat 'amateurs' pretty easily after
having gained so much experience and especially already having prepared
all the other lines.
Jeroen already has shown he has a pragmatic approach towards using the first
move. Remember that he has tried temporarily 1.c4 with Tiger a few tournaments.
I would suggest using the same approach here.
>>Then we see great lines played by the books used by rebel + tiger, which is very
>>undeniable.
>>
>>So i'm not saying that what Sjeng used is worse than a random PGN book or
>>whatever, but i'm noticing that what's used for Rebel + Tiger (2 completely
>>different styles of engines) is very similar and much better than the total
>>different book used for Sjeng.
>
>OTOH, it would be even more suspect if the same book was used for all three
>engines. There should be credit for developing different books. It's the
>inability to learn from the games, I focus on.
>
>>Though i commercially can understand the idea, from sportive viewpoint i find it
>>very disgusting.
>
>As mentioned by Hyatt in a different subthread, a bookauthor responsible for
>more than one engine is inherently suspect. There is too much room for
>shenanighans.
>
>>VERY VERY VERY DISGUSTING, especially if you hear leaked out scores of dual
>>sjeng when playing tiger (when it is equipped with the same book like sjeng).
>
>Well, leaked scores are as reliable as astrology. The Tiger strength is
>stagnating, so I wouldn't be surprised if Deep Sjeng becomes the strongest
>Lokasoft engine in a not too distant future.
>
>Regards,
>Mogens
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.