new hype again ?

[jefk]

So on PLaychess many operators now seem to have abandoned 'Brainfish'
for yet another toy, the (private?/development) 'Toltec' engine (could
only google an older version Tt2 or so but download didn't work)

Sometimes behind such hypes there nevertheless can be interesting
developments, eg the announced Sirius interface (with Cerebellum);
so maybe this Toltec thing is more than yet another SF clone!?!?!

What i find interesting are engines with other eval methods,
apparently Hannibal, Zurichess and maybe a few others (Gull?)
seem to be promising, and in longer games sooner or later
Sf can be surpassed i predict, whereby it remains to be
seen if Komodo can keep its hegemony.

[APassionForCriminalJustic]

So on PLaychess many operators now seem to have abandoned 'Brainfish'
for yet another toy, the (private?/development) 'Toltec' engine (could
only google an older version Tt2 or so but download didn't work)

Sometimes behind such hypes there nevertheless can be interesting
developments, eg the announced Sirius interface (with Cerebellum);
so maybe this Toltec thing is more than yet another SF clone!?!?!

What i find interesting are engines with other eval methods,
apparently Hannibal, Zurichess and maybe a few others (Gull?)
seem to be promising, and in longer games sooner or later
Sf can be surpassed i predict, whereby it remains to be
seen if Komodo can keep its hegemony.

They will always be clones...

Good luck surpassing Stockfish. Hannibal and Stockfish don't even belong in the same sentence.

[jefk]

They will always be clones...

so you say this Toltec is a clone ? (if it would be roughly
based on other code wouldn't be a problem imho)

NB my example with other engines vs Stockfish
was meant to indicate various possible methods
in the still ongoing search for improving engines;
we know there still are many inaccuracies, recognizing
fortresses, and more; if some of such other engines,
eg Andscac or whatever would develop fast whereas
Stockfish progress would halt, surpassing SF
is certainly not possible.

Once in a while you get a breakthrough. Rybka albite
(controversially/partially) a (Fruit) clone, was such a breakthrough,
AlphaGo was a breakthrough for Go engines; it's
a pity i think that development of the Giraffe engine was
stopped; i looked at some otherr similar systems
and they seem to play well (better than SF) in
the opening/middle phase, but very bad still tactical.
But some of these things are open source, so hopefully
some developments can be picked up there again
later by some talentful programmers/researchers.

Don't know how the Alphago team solved that problem
(in Go you also have tactics!) precisely (they only suggest
to combine various modules) but then such a method might
also work for chess. But as it is now, Toltec doesn't seem
to be a breakthrough (yet?) in the Playchess rankings.

[jefk]

[quote="APassionForCriminalJustic"] Hannibal and Stockfish don't even belong in the same sentence.[/quote]

well it's my impression that the Hannibal eval is sometimes better;
but SF is faster/is searching deeper ofcourse.
Then we have Zurichess which recently made a big leap,
and sofar it's only 1 thread; so with some good smp
method they could make quite some progress maybe.
jef
PS yes the SF opening play is improved with this 'brain'thing
but then there always are lines possible where it get's
out of book relatively early, still in early middle game.

[Dann Corbit]

So on PLaychess many operators now seem to have abandoned 'Brainfish'
for yet another toy, the (private?/development) 'Toltec' engine (could
only google an older version Tt2 or so but download didn't work)

Sometimes behind such hypes there nevertheless can be interesting
developments, eg the announced Sirius interface (with Cerebellum);
so maybe this Toltec thing is more than yet another SF clone!?!?!

What i find interesting are engines with other eval methods,
apparently Hannibal, Zurichess and maybe a few others (Gull?)
seem to be promising, and in longer games sooner or later
Sf can be surpassed i predict, whereby it remains to be
seen if Komodo can keep its hegemony.

They will always be clones...

Good luck surpassing Stockfish. Hannibal and Stockfish don't even belong in the same sentence.

The problem with clones is that generally the cloners are as dumb as a box of ball peen hammers and so the code never advances. Of course, that is not always true.
As for SF and Hannibal in the same sentence, you put them there yourself.

My turn:
Hannibal and Stockfish are two excellent, top of the line, chess engines.

[jefk]

problem with clones

ok found sources of Toltec 2, a Nayeem clone;
and Nayeem is a .... Stockfish clone !

So.. ergo... Toltec is a ...
(SF) clone indeed.
QED

PS nayeem sources are available so who's next

[Rochester]

The playchess operator are the clowns. Know nothing about the chess. Know nothing about the math. Can sell them the eifel towers (two) still then they buy.

[Edsel Apostol]

So on PLaychess many operators now seem to have abandoned 'Brainfish'
for yet another toy, the (private?/development) 'Toltec' engine (could
only google an older version Tt2 or so but download didn't work)

Sometimes behind such hypes there nevertheless can be interesting
developments, eg the announced Sirius interface (with Cerebellum);
so maybe this Toltec thing is more than yet another SF clone!?!?!

What i find interesting are engines with other eval methods,
apparently Hannibal, Zurichess and maybe a few others (Gull?)
seem to be promising, and in longer games sooner or later
Sf can be surpassed i predict, whereby it remains to be
seen if Komodo can keep its hegemony.

They will always be clones...

Good luck surpassing Stockfish. Hannibal and Stockfish don't even belong in the same sentence.

We don't have as much development time and testing resources to compete with SF development. We only develop Hannibal like a total 2 months in a year. There are tens of developers working on SF, and they are probably 100 years CPU hours worth of testing ahead of Hannibal, so yeah, it will be very hard surpassing SF.