I suppose you don't like his conclusion that the evidence is strong but not overwhelming; I don't either, frankly. What I find fascinating is simply the notion of using engine-computed histograms for this purpose.

If a score is wrong we should expect to see lots of zig-zag motions (or at least one or two really exaggerated zig-zags) just like we see in the ...b5 histogram. If a score is right, the slope should be smoother. That doesn't mean every single blunder is a bad score sheet, but when the score goes up and down in a predictable sawtooth fashion it's a dead giveaway that you're looking at a wrong move that is affecting the subsequent moves. Like yourself, I think that zanzibar underestimates the quality of the evidence he produced, but his work is to be applauded.

Speaking of technology-based ideas to delete junk, one great way to reduce spambot-related cleanup work is by blocking posts (especially by new users) that involve coding - especially link coding - that the site does *not* use.

I just see now that you are already blocking html links - good work. ;) Banning link codes that other common message boards use, will also prevent a lot of junk.

Some of those that I can recall - if they still exist - may be:

[link]...[/link] (old UBB board types, and their descendants) and

\link{} (old Disqus board types).

New <Rogoff> conspiracy theory: H. Ross Perot has taken over <chessgames.com>.

No, it's not that I disagree. I think 2...b6 is clearly the more likely correct.

But is it more likely enough for <CG> to alter the score?

And what criteria are you using?

FSR's esthetics?

Sorry, but to alter a scoresheet that makes sense (the point of my histograms is that both 2...b6 and 2...b5 "make sense") takes a higher level of evidence, afaic.

As I said on my blog, there is a difference between "guilty beyond a reasonable doubt" and a "clear preponderance of the evidence".

To change a scoresheet requires a higher level of burden of proof, especially when <CG> strikes out from the pack.

I have a lot more irons in the fire, and can strengthen this argument a little more later.

(If I were to vote for the correct move, I'd agree with FSR, certainly. If I were to vote to change the move in the database, I would vote against it. What's the <CG> process??

<cg> Pretty please forum for Grenke Chess Classic?

Of course that's the best suggestion... for this specific instant. The debate on the general principles would remain of course.

However...

The World Team has already played him twice.

The guy that would be great to see vs the team is....

Gata Kamsky.

I can dream, can't I?

<Like it or not, Tab, you have friends...>

But I would say that the certainty of your winning is not 100%.

I still don't think people understand fully what my point is... it's about the process, not the particulars of this one case.

Actually, I think they do understand. This is evidenced when, for example, <cg.com> posts that the histogram could be useful in finding questionable (note I didn't say "wrong") scoresheets.

The only point of disagreement seems to be about this particular game. You are sticking to the laudable principle of demanding a high standard of evidence, and your interlocutors are saying that (in this case) your evidence is even stronger than you give yourself credit for.

To me, in this instance, there is simply no question. <FSR> is right, and we will very probably soon have confirmation of it.

For any player above perhaps class "B," the ...b5 stands out as dubious. And when White's reply is as inappropriate a response to ...b5 as what appears in the scoresheet, it's not at all hard to conclude that the scoresheet is almost certainly mistaken.

What you have (supposedly) in Lein-Shulman is far stranger. This game is not a famous example of double blunders, and time pressure cannot have been a factor given that the blundering started on <move 2>, from a position doubtless quite familiar to both players. From move 2 on, the players (both strong GMs) played a long series of blunders. Shulman hung a pawn for essentially nothing. Lein failed to take it, far and away the strongest move, or even play 3.e4, which would have given him the advantage. Instead, he played a feeble move that would have given Shulman approximate equality after either move of his b-pawn. But no, Shulman played the ridiculous 3...Bb7?, now allowing Lein to take a free pawn <two> different ways. Lein played neither capture, although those are easily the two strongest moves on the board, instead inexplicably preferring the milquetoast 4.Nf3? Once again, Shulman could save his hanging pawn, either by 4...b4 or 4...bxc4 (which are again easily the two strongest moves). But again he played neither. And so on, and so on.

In sum, you have two strong GMs again and again playing feeble moves and avoiding obvious strong moves that any half-decent player would play. Both simply ignore Black's hanging b-pawn, and White's hanging c-pawn, for move after move - as though Black's pawn weren't even on b5! Why not? Because it's not! The circumstantial evidence overwhelmingly supports the inference that Black played 2...b6, not 2...b5.

<I will gladly bet my $100 against your dollar that he played 2...b6, not 2...b5.>

Miles vs Christiansen, 1987

< Miles actually spent some time polishing the e2-square with his finger (until Christiansen's face had assumed a suitable shade of red) before making the move 6.♘xe4. The game was of course agreed drawn in advance. >

It's possible that Shulman, at 2559, offered Lein, at 2231, pawn's odds in the game. They agreed, and Shulman, after playing 2...b4 looked up to see Lein winking at him while playing 3.Nc3. After that, as an inside joke, the pawn stayed played on b5.

Unlikely, sure. Possible, who's to say.

Or perhaps they got into an extended discussion over a few beers about computer databases. And they got to wondering how accurate the scores of the databases were. Some of the discussion involved how easily the games could be modified. They decided to test this by leaving a pawn hanging to test who puts it back on b6.

Or maybe they got sick of people who always say a GM has to play engine best. And they decided to play a game where you couldn't just pick the engine best move - since they decide to throw a monkey wrench into the eval by leaving the b-pawn hanging.

Three scenarios, increasingly unlikely, but still possible. And each has a creative bent, warped though it may be.

Admittedly, all unlikely, sure. But how unlikely is <Miles--Christensian>?

Or even this game:

A Zapata vs Anand, 1988