Speculation

 Currently I'm watching the comments on the FIDE candidates tournament 2026. The commentators have an average rating of 2670.

During the comments, an eval bar of the position generated by Stockfish is shown. Sometimes (often?) the eval bar shows a different judgement of the position than the commentators. Sometimes the eval bar shows a stubborn 0.0 judgement of the position, while the commentators have a clear preference for one  side which they wish to play. And at other moments the eval bar swings to one side, while the commentators are struggling for minutes to find where the tactic might be hidden in the position.

It takes a bit away the mystification that we tend to surround the titles with. Which is a good thing, because we get a chance to get a more realistic and objective view of what is going on.

When I started this blog back in 2005, I always was convinced that there was one specific trick that stood out where a grandmaster relied on. Over the years, I unearthed what that trick is, albeit I must admit that there are more layers to the problem.

When I observe the commentators, I'm starting to become more and more convinced that even the difference between grandmasters and super grandmasters is based on the very same trick.

A lot has been said about instinct. I always have been reluctant to talk much about instinct. The eval bar shows that even the instinct of grandmasters often leaves a lot to be desired. Which is good news. Both for us and for plateauing grandmasters.

The very existence of this trick is of course no more than mere speculation as long as it is not proven. For me that is no problem, because I see proof enough. But mr. Elo must be the final judge.

Now I'm speculating, why not take this a whole lot further?

The evaluation function of Stockfish is quite meager. It leads the pieces only globally in the right direction. Because it calculates future positions, it can only guarantee that it is still in good shape after 35 ply. And even this meager evaluation function is already good enough to get a rating of 3600+.

I showed the all importance of LoAs (lines of attack). All technicalities are geared around LoAs and their pivot points and sitting ducks. When the evaluation function of Stockfish would be focusing on these technicalities, then it would play better chess, and it would produce evaluations that would be more comprehensible for humans, I speculate. Our curse is that we are so bad at chess, that even a mediocre evaluation function is enough to crush even the best super grandmasters with ease.