Houdini 5 in TCEC

[M ANSARI]

There was a time where the strongest engines had a weakness that could be exploited and I guess an engine could be tuned to get such positions. I remember Rybka 3 underestimated the bishop pair and Naum would rack up nice wins against it. Also Rybka could get nice wins against Houdini as it over estimated the value of a queen over pieces ... apparently queen value worked well at fast time controls but at long time controls the pieces would contain the dynamic activity of the queen and then over power it. Today, I just don't think there are any weaknesses that are apparent to a human player. Sure an engine might misplay a rook endgame or other endgame ... but in general the top engines have incredibly well tuned evaluations, and most of the wins will be due to slightly stronger evaluations by one engine, or the ability to go one ply deeper in some positions.

[Hai]

It will be very interesting to see if Houdini 5 can get full points in the tactical test suite.

[Nay Lin Tun]

Some people expect a lot of elo jump when the old programmers come back. In reality, both Vas and Houdart are commercial programmers and they did not come back for a long time probably because they cant do reasonable elo jump to sustain their business. So similar story like Vas´ Rybka 5(fritz 15), I expect Houdart`s H5 elo jump no more than 15-20.

[JJJ]

Some people expect a lot of elo jump when the old programmers come back. In reality, both Vas and Houdart are commercial programmers and they did not come back for a long time probably because they cant do reasonable elo jump to sustain their business. So similar story like Vas´ Rybka 5(fritz 15), I expect Houdart`s H5 elo jump no more than 15-20.

I think houdart wants more than 15-20 elo jump.

[Dicaste]

Some people expect a lot of elo jump when the old programmers come back. In reality, both Vas and Houdart are commercial programmers and they did not come back for a long time probably because they cant do reasonable elo jump to sustain their business. So similar story like Vas´ Rybka 5(fritz 15), I expect Houdart`s H5 elo jump no more than 15-20.

I think houdart wants more than 15-20 elo jump.

Exactly. It's nice to see Houdini in competiton and Houdart doesn't release newer version unless +50 elo before previous version. Although can't have a chance against Stockfish and Komodo.

[JJJ]

Some people expect a lot of elo jump when the old programmers come back. In reality, both Vas and Houdart are commercial programmers and they did not come back for a long time probably because they cant do reasonable elo jump to sustain their business. So similar story like Vas´ Rybka 5(fritz 15), I expect Houdart`s H5 elo jump no more than 15-20.

I think houdart wants more than 15-20 elo jump.

Exactly. It's nice to see Houdini in competiton and Houdart doesn't release newer version unless +50 elo before previous version. Although can't have a chance against Stockfish and Komodo.

No chance if the elo jump is 50. Some chance if the elo jump is 100. It's been 3 years. Houdart didn't probably work 3 years in a row on Houdini, but maybe 6 month and until the 3nrd stage come, there is some times. So maybe jump will be big, maybe not. I m hoping for a nice surprise here.

[Ralf Müller]

Houdart doesn't release newer version unless +50 elo before previous version.

Houdini 4 was only +30 Elo more than Houdini 3.

[Ozymandias]

You're being generous, it only stands at +14 on CCRL 40/40.

[whereagles]

there's way too much speculation on Houdart.. it's just another 2 weeks to start and then we'll see

[Ozymandias]

In two weeks, we'll see H4, so the speculation will go on for a while yet.