<shach matov: <Elo for the 2284 master if you value the Rybka that was whitewashed at a mere 2900 Elo>

You have been talking about this all the time but I already replied: the match took place 6+ years ago and according to the chessbase article they used <an intermediate version> of Rybka (that is, it wasn't even the strongest Rybka version for 2008 or 2007 when the match was played). So we know one thing: it was weaker than 2900 Elo, though nobody specified it's exact rating. Thus, we can't possibly make any conclusions based on this one match since we simply don't know that Rybka's Elo. It could have been rated 2600. How would that help us in comparing a 2900 player with a PDP which has solved chess completely and has "divine" judgement? So unless you can show evidence about that Rybka's Elo, this example is not very useful.>

2600 is a joke when you consider the best players struggling in engine matches years before that. You keep taking intermediate to mean that it's terrible, but it could very well simply mean that it wasn't given a release version, so it was just Rybka plus whatever improvements they had so far, but without the others they'd include in the next version. It <really> doesn't make sense that they'd test with a crappy version; that defeats the point. The idea was to find out just how above humans Rybka was. You don't do that by using a garbage version and giving the human high odds on top of that just to make the results as ambigious as possible.

<<Daisuki>

And I must most definitely warn you to stop all the silly personal attacks, for otherwise I can also start those and you will feel very uncomfortable indeed! So lets keep our attenuation glued to the argument at hand and not constantly talk about our characters. I can also play that game and I assure, you will not like it!>

An intermediate version means it was not the best they had even at that time! And we don't know what it means exactly apart from it being weaker than best versions. So was it 2600, 2700? We simply don't know. Unless you can provide proof of it's actual strength, you simply do not have an argument there.>

Why <must> extreme <human-style> (i.e. not based on engines blundering by running out of only moves after the horizon; this can in theory go on forever without resulting in huge increases in strength against humans) ratings be "possible", and even if so, why would they matter if there's no clear way to overcome the <fat stack> of advantages that the Chessgames community overwhelmingly agrees is <amazing> for an elite player (yes, I know you disagree and imagine that a perfect player simply "finds a way", but that's wishful thinking so long as you have no evidence, and weighs poorly on the scale next to the collective chess knowledge of Chessgames; sorry, but this will go on forever without "adjudication", since you have a 0% of changing your mind and a 0% chance of convincing much of the Chessgames community, it seems)? Simply assuming that Carlsen is too weak isn't good enough; in practice skill should scale in the way metatron2 mentioned, where you know how to win a <very clear> win, but would always lose in a fair game. On average, of course. Carlsen may not win all the games...perhaps he wins 50%, draws 30%, and loses 20%. Good enough for the match, and losing 20% doesn't make much sense to me anyway given that he doesn't blunder <<<material>>> <from a position of strength and comfort> nearly that often, or so it seems...

Well I have to again repeat myself: Rybka was definitely not even close to 2900 not to speak 3200! Come on!

<Even 4500 Elo only requires ~1600 points for the knight in Carlsen's hands...a number that's quite possible.>

No way 1600 is possible. My 4500 is very conservative actually and thus reasonable. But where does 1600 come from? We have to realize that human rating stays at 2900 while PDP's is highest possible! Which means that PDP's ability to defend against a knight is much stronger than human's.

First of all, you have to stop repeating this! Argue with facts, numbers, ideas, not what somebody else may say! And what the hell does it mean cg community? Did you ask everybody? Three or four people agreed with you concerning an engine (not even PDP) and couple agreed with my point of view, calling my arguments here as elaborate and deep. This is hardly the whole of cg community!

We absolutely can't say for sure whether even a night is enough (meaning we can't really prove). However, you have not shown why a knight is enough either. You just said things like it will help Carlsen force exchanges; that's not a definitive answer, since I can also say the queen will be able to force exchanges much better than a knight. Neither answer is definitive in a mathematical way. So your answer is no better than mine.

And I already gave a long post (on WCC page) for my reasons to believe that exchanges will not always be beneficial for the human since he is much weaker tactically and will not be able to look as deep at the tactical consequences of the exchanges. I gave an elaborate answer to this issue.

Well, we can definitely agree about that.

Although, I am not really forcing your hand here. I said before that nothing is provable in such an argument, but my common sense, intuition and knowledge tell me that against a PDP any human will have major problems, even with a knight.

So, sure, we can end it right here, but I am willing to continue indefinitely. I like this subject matter, it makes me think.

Well I have to again repeat myself: Rybka was definitely not even close to 2900 not to speak 3200! Come on!>

Implausible, I'm afraid. Can you counter what I said with anything other than your personal belief regarding what "intermediate" <must> mean?

<<Even 4500 Elo only requires ~1600 points for the knight in Carlsen's hands...a number that's quite possible.>

No way 1600 is possible. My 4500 is very conservative actually and thus reasonable. But where does 1600 come from? We have to realize that human rating stays at 2900 while PDP's is highest possible! Which means that PDP's ability to defend against a knight is much stronger than human's.>

Yes way is it possible. And your 4500 comes from nowhere at all...you have no evidence for it, as there is no such player.

<Consider: simplify and say that for two 2000 Elo players, a knight is about 500 Elo.>

No.

<Now add 50 Elo for a knight for each 100 Elo increase in players' rating, resulting in about 1000 Elo (or even 1200, say) for a knight at 2900 Elo.>

Too small.

<Now, when we compare 2900 player to PDP, the 2900 stops while PDP is still going for quite a while after 2900! Which means that the defensive ability of PDP is increasing while the attacking ability 2900 player is not.>

Doesn't matter so long as in general only material mistakes can fail to win rather than delay the win.

<So if anything, that initial 1200 Elo for a knight should decrease in value relative to a PDP. So it can't possibly be as high as 1600 against the PDP. I emphasize: the defensive ability of PDP is much greater than of a 2900 player, so even if a knight is worth 1200-1400 for two 2900 players, it's value must necessarily decrease when a 2900 plays a PDP.>

Do you have <any> evidence that the passive, <lost> player just gets all this control by merit? We <know> that in greater extremes that the lost player's skill is irrelevant so long as the won player is good enough to win. So in the end you're simply effectively assuming that a knight is not enough and using that to justify the idea that the PP can mitigate the advantage of the knight...kind of circular there.

<<shach matov: <Chessgames community overwhelmingly agrees>

First of all, you have to stop repeating this! Argue with facts, numbers, ideas, not what somebody else may say! And what the hell does it mean cg community? Did you ask everybody? Three or four people agreed with you concerning an engine (not even PDP) and couple agreed with my point of view, calling my arguments here as elaborate and deep. This is hardly the whole of cg community!>

Chessgames' overall thought is a statistical number, and one with far more basis than your pessimistic knight values and your arbitrary PP values. It definitely wasn't three or four people. I recall <many> people all calling your conclusion crazy. Can you find anyone who clearly agreed with you (not just someone admitting that you <could> be right; anyone reasonable does that even if they still use my working conclusion instead of yours).

<So lets stop this silliness about cg community! Argue with me if you have real arguments, don't tell me what somebody else thinks!>

The problem doesn't seem to be my lack of arguments, as you still haven't upset any of the reasoning I've laid out. The problem is that you don't have any evidence that comes from an actual place other than your mind, so why use your working conclusion instead of the one that makes sense?

<<shach matov: <Why is the queen enough but not the knight?>

We absolutely can't say for sure whether even a night is enough (meaning we can't really prove). However, you have not shown why a knight is enough either. You just said things like it will help Carlsen force exchanges; that's not a definitive answer, since I can also say the queen will be able to force exchanges much better than a knight. Neither answer is definitive in a mathematical way. So your answer is no better than mine.>

Can I not just substitute queen for knight in that paragraph (substitute "two queens" for "the queen" first)? If not, why not? Without a real answer the main advantages of the knight, as explained quite well, can quite plausibly shine through.

<And I already gave a long post (on WCC page) for my reasons to believe that exchanges will not always be beneficial for the human since he is much weaker tactically and will not be able to look as deep at the tactical consequences of the exchanges. I gave an elaborate answer to this issue.>

You believe this, but I have no reason to. Carlsen is no chump; he will find good enough squares for his pieces and understand when the PP has done well for its pieces. The PP is the one with the hard choices, not Carlsen. Can you tell me exactly how the PP is going to prevent Carlsen from placing his pieces well <while> avoiding trades? It's just implausible.

<So your idea of just exchanging and trying to control space is simply an idea, you can't possibly prove such a scheme will work for human's benefit. Say he goes for an exchange that looks good, but PDP calculates 30-50-100 moves deeper and sees major problems for this move, either tactically or strategically. Where does that live the human? He may easily lose material and get a terrible position. So it's absolutely not as simple as you're trying to make it look!>

It's an idea that makes sense to the Chessgames community because it is based on actual strategic principles of chess, and is able to be weighed according to its ease of use for an elite player. If you had upset the idea then <maybe> you'd have something, but you haven't. You still just refuse to understand that with six pieces against seven <everything> is tilted Carlsen's way, and the PP is simply <suffering>. Tactics in general won't work as Carlsen has all the control. Infinity-ply deep plans don't matter so long as you <can't> maintain positional gains and in fact just keep getting squeezed harder so long as you have this overwhelming material deficit. It's just easy street for Carlsen. Given that you have no reason for the knight being insufficient but not the queen it's only reasonable to just use the same plan, which has all the same upsides as for the queen. Just establish that Carlsen hangs 2+ pawns worth of material for nothing in over half of his games if you don't like that.

<<shach matov: <Daisuki: <shach matov>, it sure takes a while to post this much. I'm sure we both have better things we could be doing>

Well, we can definitely agree about that.

Although, I am not really forcing your hand here. I said before that nothing is provable in such an argument, but my common sense, intuition and knowledge tell me that against a PDP any human will have major problems, even with a knight.>

I've kept telling you that it's all just working conclusions. Yours just seems quite arbitrary, that's all. Anyway, common sense is a terrible argument (especially when you're about the only one here that seems to have what you call "common sense"), intuition is not an argument, and knowledge is something you're not expressing here in a form that makes a lot of sense to me.

<So, sure, we can end it right here, but I am willing to continue indefinitely. I like this subject matter, it makes me think.>

I'm not sure what there is to think about anymore. I've mostly been able to say the same things forever, as you haven't disrupted them at all. Only recently did I realize that on top of everything else seven vs. six pieces was yet another crushing advantage in Carlsen's favor. Carlsen of course would likely need <very> little time to understand that, and he'd know about twenty other important things I couldn't even mention because I'm not good enough.

<I'll make a last point for tonight: I mentioned before the fact that initially I talked about a match, say 15 games during a few weeks. This is another major advantage for the PDP or even an engine, as I explained on WCC page. Any game can last as long as 100+ moves, the human will almost always start making mistakes and even blunders sometimes. Now imagine after such a game having to play another one tomorrow or a day after tomorrow (which also may be 100+ moves)! With each game, the mistake and blunder rate must go up, and we know these will be punished ruthlessly by the PDP. So for all these reasons I am very skeptical that a human has a chance. Opinion? Yes, but I think not a baseless one>

Hmmm, I though you wanted to stop the debate, but I guess not.

But as you can see, at this point it's my opinion against yours (as I stated from the start), and you can't prove any of your statements! I never claimed I could, always said that it's my opinion...

While regarding the stamina question, our opinions are 180 degrees opposites. Regardless whether one side has an initiative or the other, playing 100+ moves against a PDP is extremely difficult, especially psychologically and it must inevitably cause a major increase of the mistake/blunder rate of the human player. While the idea that the human will always have an initiative is plain silly, surely even you must agree that in some games he must lose the initiative early in the game.

And now imagine that he has to do this for a month or more, everyday or every other day. However strong one is, the strain will be completely overwhelming! To me this is very clear.

On the other hand, if we're given long enough time after the match to analyze the comp's play (on the bases of the games played during the match), maybe we can find some weaknesses; though how to exploit those against such a superior overall chess machine in a real game is not clear. Also, the comps are constantly improving themselves, if we find a weakness, the programmers will immediately try to get rid of it. IMO it will be a hopeless competition for us; if we could easily find weaknesses in computer chess, they would not be getting stronger every day for the last 20+ years.

Hmmm, I though you wanted to stop the debate, but I guess not.>

Well, I don't mind discussing new things. I wish you had better answers to some of my main issues, since it would be interesting if you did.

<But as you can see, at this point it's my opinion against yours (as I stated from the start), and you can't prove any of your statements! I never claimed I could, always said that it's my opinion...>

My statements are based on evaluation of actual chess principles relevant to the position and on actual events and what they mean (you could try to evaluate the master-Rybka result differently, but so long as the 50% odds standard for a 100%-happened occurrence is good and so long as you can't support a comically low Rybka rating your low knight values are simply automatically implausible). Yes, they are not proof, but they are not simple opinions, either. They are attempts at reasoning, not statements of personal preference, the latter being what an opinion is.

<While regarding the stamina question, our opinions are 180 degrees opposites. Regardless whether one side has an initiative or the other, playing 100+ moves against a PDP is extremely difficult, especially psychologically and it must inevitably cause a major increase of the mistake/blunder rate of the human player. While the idea that the human will always have an initiative is plain silly, surely even you must agree that in some games he must lose the initiative early in the game.>

The game may be difficult while all pieces are developed and the piece ratio is 7:6, 6:5, or 5:4 (just merge lower complicated positions into the higher ones that aren't actually complicated for simplicity's sake). Given that trading is inevitable, lest the PP cede everything Carlsen aims for to Carlsen, this will not likely be above 20 moves per game. These moves would be during and right after the completion of development, so not several hours in when Carlsen is more tired. After this phase passes Carlsen's ease of play and plan and even more extreme number of winning move choices per move just make it a case of mopping up to him.

The initiative needs to be held (if it's not held for very long then the PP obviously isn't making progress rather than defending) by continuing to put greater threats on the enemy than vice versa. This is seriously not easy when down a piece with basically no compensation (okay, we know that castling queenside is faster for the PP, but castling queenside is generally inferior anyway, and it's easier to safely develop the kingside knight than the queen (bishops being "equal") to boot, which makes kingside castling still faster than queenside castling in general). To improve a position one fundamentally has to position one's pieces well while denying the same to the opponent. This is a given for the PP in a normal game against Carlsen. In a knight odds game, however, the strict disparity of force changes this, along with the fact that trading is also bad (which is a huge problem given how many viable piece trades there are in chess, and how occupying good squares is not overall possible without putting your pieces somewhere where Carlsen can oppose them with his own pieces; the reverse is also in Carlsen's favor, as he can more brazenly occupy good squares knowing that trading is good for him). Carlsen can <mathematically> control more squares and/or control them better (in the extreme case he can attack a pawn with 6 pieces + 2 pawns and the PP has <no> way to defend it, being able to defend it with only 5+2p) by virtue of having more pieces.

<And now imagine that he has to do this for a month or more, everyday or every other day. However strong one is, the strain will be completely overwhelming! To me this is very clear.>

It sure makes sense if you ignore just how great the extra knight is, how it makes the PP holding small positional gains unlikely, how easy it makes positional gains for Carlsen, and thus how the game is simply not fair in general.

<The rest of the posts here seems to be just repetition of what we said before, and again just opinions. I can live with it though, no problem.>

I still disagree that it's fair to call our positions opinions. Opinions are completely subjective, which means that they're based purely on personal preferences. Now, maybe this is how you look at your position, but this is a problem in the context of what I'm looking at, which is the most rational working conclusion (an operating conclusion based on evidence and reasoning that exists outside of preferences, not proof of course, but this is ultimately just shades of grey (although this is admittedly murkier than many things are) as proof is never going to be perfect so long as we are forced to use axioms to support all reasoning). Opinions don't really counter such a thing at all, and just don't make sense to hold when they are opposed directly by reasoning. Basically, you may prefer that a PP win such a match, but if you don't have any idea <how> it will happen (the PP being stronger is not enough given that we know it has to be enough stronger to overcome knight odds, and that even being possible for knight odds against Carlsen is not remotely a given) and can't solve the major chess problems it has in any clear way then it just seems like wishful thinking.

I guess I just have two main things:

Why should extreme ratings be possible, and why should they matter if there's no clear way to disrupt the extreme advantages having the knight gives? We know that the lost player's rating is irrelevant given enough skill on the part of the winning player, so why is Carlsen not skilled enough? An argument that the PP just <must> have enough skill is simply assuming the conclusion; i.e. "the knight must not be enough for Carlsen, so the PP has enough skill to overcome the knight, so the knight must not be enough for Carlsen!". In the end it's really about whether or not Carlsen has enough skill, not the PP. The PP is irrelevant so long as you agree that the starting position is theoretically won for Carlsen. What exactly is Carlsen failing to do well while up the extra piece, and why is he failing to do it? Again, try to focus less on the PP, whom you can't estimate anyway. Tell me exactly how Carlsen, with huge pressure, simple, crushing plans, and many winning move options on most every move that isn't a recapture, is failing.

Why is the queen enough but not the knight? Every time you've answered it's been possible to substitute QQ for Q and then Q for N in your sentences. This implies that you are just making an assumption, rather than specifying anything that uniquely contrasts Q and N. Q and N are functionally identical for all plan purposes in all ways except for the queen having higher mobility and the ability to control more <extra> squares. This would be relevant if you established that Carlsen blunders 2+ pawns of material in the majority of his games (a number that brings the knight's net square/material advantage into question but not the queen's), but you haven't.

------

<matmzc: Instead of silly arguments about playing a computer with knight odds why don't you guys simply do the experiment. Play some games with an engine and see how it goes. I am at 2012 rating and I can beat 2600 rated Shredder about two to one with some draws if I remove Shredder's QN.>

Were these classical games, and is your rating from FIDE? Anyway, your performance rating is almost 700 points above your rating (although how many draws there were is unclear). One would expect that the master, nearly 300 points above you, would do better, and for that to be so Rybka (which is generally known as better than Shredder, is it not?) only needs the same crappy 2600 rating...

------