Fritz 13: “Let’s Check”

[Albert Silver]

I mean, do the GMs use Najdorfs and Scheveningens in something like half of their games, spending most of their creative effort on them, like computer buffs using engine analysis of openings (the way new Fritz database will work) are doing on playchess?

Some do of course. You don't imagine that all GMs play everything under the sun after all?

Anyway, as I said, engines need opening books for a reason, don't you think? Therefore the usefulness of the new Fritz opening database is very doubtful.

Your logic baffles me.

According to your reasoning, since engines use analysis that is not theirs on a move-to-move basis, the usefulness of their analysis is doubtful. By the same reasoning, I should not bother with a GM's analysis since they are no less fallible, and they too are incapable of coming up with perfect moves on a move-to-move basis.

There is only one conclusion: I should take up golf!

[Laskos]

I mean, do the GMs use Najdorfs and Scheveningens in something like half of their games, spending most of their creative effort on them, like computer buffs using engine analysis of openings (the way new Fritz database will work) are doing on playchess?

Some do of course. You don't imagine that all GMs play everything under the sun after all?

Can you show an opening novelty found in playchess engines room (there are many found) and popular among GMs?

Anyway, as I said, engines need opening books for a reason, don't you think? Therefore the usefulness of the new Fritz opening database is very doubtful.

Your logic baffles me.

According to your reasoning, since engines use analysis that is not theirs on a move-to-move basis, the usefulness of their analysis is doubtful. By the same reasoning, I should not bother with a GM's analysis since they are no less fallible, and they too are incapable of coming up with perfect moves on a move-to-move basis.

There is only one conclusion: I should take up golf!

You seem to lack some logic reasoning. The problem is the following: the engines are much better in the middlegames than in the openings compared to very strong humans. The new Fritz database will cover, as expected, some openings. These openings will be analyzed by _engines_ in this database. I reasoned, as I said, that the new Fritz database will be useful in engine-engine playchess room. Its usefulness to a GM (used as move-to-move or whatever) is very doubtful, as he probably doesn't need an engine evaluation of some kind in the opening (besides maybe blunders)

Kai

[AdminX]

Shall we call this "Cloud Chess"?

Anyone super excited about this?

“Let’s Check” is a world first, it is the innovation with which Fritz 13 will have the chess world holding its breath

It is with the help of this new function that Fritz 13 users world-wide will create a gigantic database of knowledge. No matter which position you happen to be analysing, should you wish to Fritz 13 will send the evaluation of your engine totally automatically to the “Let’s Check” server. The best analyses will be included in the database of chess knowledge. And best of all: after a short time you will find for almost every position (and without having to spend time calculating) high level analysis produced by a strong chess program or even several analyses produced by various programs for you to compare. And in live broadcasts on playchess.com there will be several hundred computers following world class games simultaneously and feeding the “Let’s check” database with profound analysis. Well, if your analysis goes deeper you can actually gain more positions from other players and see your name enter the roll of honour for “Let’s Check”.

"No matter which position you happen to be analysing"
"The best analyses will be included in the database of chess knowledge"

and the best

"you will find for almost every position (and without having to spend time calculating) high level analysis"

LOL

Kai

You can laugh, but it isn't inaccurate IMHO. It is a bit premature, but the description will correspond fairly quickly. A position will keep up to three computer evaluations (different engines to be exact), and each evaluation is upgraded to a new and deeper one if someone chooses to use the same engine and go deeper. To register analysis, you must run it to a certain depth, which varies according to the computer you are using and the number of cores.

For example, my computer is a 4 x 4.2GHz machine, and only registers a position after at least one minute of continuous analysis if I am using all four cores.

I have personally contributed about 1000 analyses of Houdini 1.5a, including perhaps 50 original positions (i.e. novelties). In over 100 cases, I'd estimate the time analyzed was anywhere between three minutes to an hour (only twice did I run an hour). If I run for an hour non-stop, the position is declared as having undergone "Deep Analysis". Users can also contribute written comments which are then voted upon by others for their quality.

If you are running an engine while watching a game broadcast on Playchess, these too can be automatically added to the LiveBook if you so choose. You can even try to anticipate a novelty in these cases, by playing the move on the board while watching, and getting credit for the analysis first. I suspect there will be a number of users having fun gunning for first credit for analysis in such cases.

His comment was about the "quantity". Exactly how many different positions are there to store? Even the CAP project Dann worked on did not have enough positions to touch even 1% of available book moves. And book moves don't touch even .00000000000000000000000000000000001% of possible games...

Sure, but CAP was mostly 5-6 computers running on it, and the positions were analyzed for some 10 seconds for one core per position on average. As to quantity, IIRC CAP still covered some 10 million positions which is probably a bit more than 1% of the moves in the opening books. Now take into consideration a massively collective effort, with comments and room for constant updating and improvement, and I think the scale and potential is on quite another level.

Reuben Fine wrote a book called "The Ideas Behind the Chess Openings" (ISBN 0-8129-1756-1). I don't see how this helps an amateur chess player if he/she does not understand why a move is played. Once out of book they are lost and have no idea what to do. Still I might get it to renew my Playchess account.

[Albert Silver]

I mean, do the GMs use Najdorfs and Scheveningens in something like half of their games, spending most of their creative effort on them, like computer buffs using engine analysis of openings (the way new Fritz database will work) are doing on playchess?

Some do of course. You don't imagine that all GMs play everything under the sun after all?

Can you show an opening novelty found in playchess engines room (there are many found) and popular among GMs?

As I said before, even if I could, it would be impossible to prove. In any case, I don't think it is a reference, since it presumes that a GM knows abut the novelties, and decided they lacked value. Instead of the far more likely option: he has no knowledge whatsoever about novelties from the Playchess engine room.

Anyway, as I said, engines need opening books for a reason, don't you think? Therefore the usefulness of the new Fritz opening database is very doubtful.

Your logic baffles me.

According to your reasoning, since engines use analysis that is not theirs on a move-to-move basis, the usefulness of their analysis is doubtful. By the same reasoning, I should not bother with a GM's analysis since they are no less fallible, and they too are incapable of coming up with perfect moves on a move-to-move basis.

There is only one conclusion: I should take up golf!

You seem to lack some logic reasoning.

If so, your reply didn't show where.

The problem is the following: the engines are much better in the middlegames than in the openings compared to very strong humans. The new Fritz database will cover, as expected, some openings.

Not some openings: all openings.

These openings will be analyzed by _engines_ in this database. I reasoned, as I said, that the new Fritz database will be useful in engine-engine playchess room. Its usefulness to a GM (used as move-to-move or whatever) is very doubtful, as he probably doesn't need an engine evaluation of some kind in the opening (besides maybe blunders)

Kai

All I can say is that I think you have little idea how much engines are relied upon in opening research nowadays.

[Albert Silver]

Reuben Fine wrote a book called "The Ideas Behind the Chess Openings" (ISBN 0-8129-1756-1). I don't see how this helps an amateur chess player if he/she does not understand why a move is played. Once out of book they are lost and have no idea what to do. Still I might get it to renew my Playchess account.

Absolutely, and this is merely a told to be added, not something meant to replace study and understanding.

[Mike S.]

Most chess players aren't GMs or IMs, but less. - Note that the whole process of position selection, where to put analysis effort into, etc. will be "human-guided", by chess players.

As for engines being better in the middlegame relative to the opening, that's true, but we can expect that the most interesting novelties, and maybe re-discovered old continuations too etc. will not be in the early opening, but later. Actually, the opening theory covers much longer variations than the opening itself. That is of course a matter of definition. A 20th or 30th "book move" is not an opening move really, to me.

So, there we'll not have that problem. Also, there are some engines which can handle the early opening better than other engines. The knowledgable chess engine users will have a sense for which engines to trust more, and which to trust less, in such positions. - Of course, some gambits may remain a problem respectively to that.

As I see it from a hobby player's perpective, it's not just about to find and evaluate the absolute strongest moves each (although that will probably be the major purpose), but also which "second rate" moves or even crazy ideas may be useful, because the opponents are most probably not prepared against them. Some moves are 2nd rate only if the opponent finds the best replies. The Let's Check analyses could e.g. provide me with deeply analysed continuations against the less good replies, to such extravagances.

Baselines: The longer I ponder about Let's Check, the more I think it may indeed be one of the greatest inventions in chess software, since very many years. It could bring computer chess out of "nerd corners" (like the engine room), and put it into the middle of the general chess life, providing more notable contributions than before.

[Albert Silver]

Reuben Fine wrote a book called "The Ideas Behind the Chess Openings" (ISBN 0-8129-1756-1). I don't see how this helps an amateur chess player if he/she does not understand why a move is played. Once out of book they are lost and have no idea what to do. Still I might get it to renew my Playchess account.

Absolutely, and this is merely a told to be added, not something meant to replace study and understanding.

A tool to be added. (the error galled me so much I had to fix.)

[Laskos]

Can you show an opening novelty found in playchess engines room (there are many found) and popular among GMs?

As I said before, even if I could, it would be impossible to prove. In any case, I don't think it is a reference, since it presumes that a GM knows abut the novelties, and decided they lacked value. Instead of the far more likely option: he has no knowledge whatsoever about novelties from the Playchess engine room.

Therefore the new database will popularize these novelties? And then GMs, using the new Fritz database, will suddenly start to play Najdorfs and related, where the computer analysis was the most productive on playchess?

The problem is the following: the engines are much better in the middlegames than in the openings compared to very strong humans. The new Fritz database will cover, as expected, some openings.

Not some openings: all openings.

Wow, even the wrong ones? First it was "almost every position", now it is "all openings", next it will probably be "all first five moves from the starting position" LOL.

These openings will be analyzed by _engines_ in this database. I reasoned, as I said, that the new Fritz database will be useful in engine-engine playchess room. Its usefulness to a GM (used as move-to-move or whatever) is very doubtful, as he probably doesn't need an engine evaluation of some kind in the opening (besides maybe blunders)

Kai

All I can say is that I think you have little idea how much engines are relied upon in opening research nowadays.

Yes, using mostly blunder check and databases of human games. You yourself admitted that GMs are unaware of the many opening novelties on playchess engine room. How the engine analyzed new Fritz opening database will be useful to strong human players (as you said) is very unclear, it's probably just a new blunder-checked database of openings, and a database not of human games, but a tree of openings with/without holes in it.

Kai

[Albert Silver]

Can you show an opening novelty found in playchess engines room (there are many found) and popular among GMs?

As I said before, even if I could, it would be impossible to prove. In any case, I don't think it is a reference, since it presumes that a GM knows abut the novelties, and decided they lacked value. Instead of the far more likely option: he has no knowledge whatsoever about novelties from the Playchess engine room.

Therefore the new database will popularize these novelties?

If they were appear in it and are good, yes.

Not some openings: all openings.

Wow, even the wrong ones? First it was "almost every position", now it is "all openings", next it will probably be "all first five moves from the starting position" LOL.

Why do you act so surprised and add "LOL"? It is what I told you from the very beginning.

These openings will be analyzed by _engines_ in this database. I reasoned, as I said, that the new Fritz database will be useful in engine-engine playchess room. Its usefulness to a GM (used as move-to-move or whatever) is very doubtful, as he probably doesn't need an engine evaluation of some kind in the opening (besides maybe blunders)

Kai

All I can say is that I think you have little idea how much engines are relied upon in opening research nowadays.

Yes, using mostly blunder check and databases of human games.

You really shouldn't make such statements when you don't know what you are talking about.

You yourself admitted that GMs are unaware of the many opening novelties on playchess engine room. How the engine analyzed new Fritz opening database will be useful to strong human players (as you said) is very unclear, it's probably just a new blunder-checked database of openings, and a database not of human games, but a tree of openings with/without holes in it.

Kai

I didn't "admit it", I pointed it out. How the openings database will be useful to strong and weak players, is actually extremely clear, just not to a non-player such as yourself.

[Laskos]

Wow, even the wrong ones? First it was "almost every position", now it is "all openings", next it will probably be "all first five moves from the starting position" LOL.

Why do you act so surprised and add "LOL"? It is what I told you from the very beginning.

What did you write from very beginning? That is what I wrote:

--"you will find for almost every position (and without having to spend time calculating) high level analysis"

LOL

Kai

And in your first post you replied:

--You can laugh, but it isn't inaccurate IMHO.

All I can say is that I think you have little idea how much engines are relied upon in opening research nowadays.

Yes, using mostly blunder check and databases of human games.

You really shouldn't make such statements when you don't know what you are talking about.

Please release your secret information about how GMs are using comps for openings LOL.

You yourself admitted that GMs are unaware of the many opening novelties on playchess engine room. How the engine analyzed new Fritz opening database will be useful to strong human players (as you said) is very unclear, it's probably just a new blunder-checked database of openings, and a database not of human games, but a tree of openings with/without holes in it.

Kai

I didn't "admit it", I pointed it out. How the openings database will be useful to strong and weak players, is actually extremely clear, just not to a non-player such as yourself.

Well, completely a non-player I am not, are you a hidden GM or a less hidden... well, you know? Anyway, your weak argument does not prevent me from making valid statements about the new, _engine_ analyzed, opening tree with holes in it.

Kai