My take on the whole "End of an era" thing.

[hgm]

More important is that the NN-based engines aren't really very good at Chess in general: they seem to be quite inferior for analyzing arbitrary positions. The one thing they are good at is playing games from the opening position, because that allows them to avoid the large fraction of positions where they would suck. But who wants that?

[chrisw]

More important is that the NN-based engines aren't really very good at Chess in general: they seem to be quite inferior for analyzing arbitrary positions. The one thing they are good at is playing games from the opening position, because that allows them to avoid the large fraction of positions where they would suck. But who wants that?

Er, people who play chess from the usual opening position?

It’s true that current NNs are not good at playing lost positions nor good at playing wildly wonpositions but the former is lost anyway and the latter won, so so what?

[jp]

More important is that the NN-based engines aren't really very good at Chess in general: they seem to be quite inferior for analyzing arbitrary positions. The one thing they are good at is playing games from the opening position, because that allows them to avoid the large fraction of positions where they would suck. But who wants that?

Er, people who play chess from the usual opening position?

More to the point, NN engines are good in the opening phase and get worse the longer the game goes on. That would be good for users who want an analysis engine if they suggested genuinely good opening novelties, but so far they haven't.

It’s true that current NNs are not good at playing lost positions nor good at playing wildly wonpositions but the former is lost anyway and the latter won, so so what?

But as we've discussed in other threads, in endgames they can turn wins into draws and draws into losses.

[Laskos]

More important is that the NN-based engines aren't really very good at Chess in general: they seem to be quite inferior for analyzing arbitrary positions. The one thing they are good at is playing games from the opening position, because that allows them to avoid the large fraction of positions where they would suck. But who wants that?

Well, Chess as a board game is defined as from default starting position. Number of positions reached in standard Chess is IIRC larger than a square root of total number of arbitrary positions using an allowed number of Chess pieces. But you are right, the vast majority of these arbitrary positions is badly analyzed by NN-based engines. I don't know if that can be corrected, hand-crafted eval is simply more generalist and possesses a higher level of abstraction.

[corres]

More important is that the NN-based engines aren't really very good at Chess in general: they seem to be quite inferior for analyzing arbitrary positions. The one thing they are good at is playing games from the opening position, because that allows them to avoid the large fraction of positions where they would suck. But who wants that?

Obviously the inferiority of NN engines for analyzing (that is in arbitrary position) depends on the dimension and structure of NN. In the case of an NN engine with bigger and better structured NN this issue is smaller.
We can hope the development of NN engine will reduce this issue.

[hgm]

More important is that the NN-based engines aren't really very good at Chess in general: they seem to be quite inferior for analyzing arbitrary positions. The one thing they are good at is playing games from the opening position, because that allows them to avoid the large fraction of positions where they would suck. But who wants that?

Er, people who play chess from the usual opening position?

The point is that people tend to play their own moves from that position, unless they happen to be operator for a NN engine in a computer tournament. And then they would like to get expert advice from an engine on the positions that result. LC0 is not very good at that; it will often come up with tactical blunders.

[Laskos]

More important is that the NN-based engines aren't really very good at Chess in general: they seem to be quite inferior for analyzing arbitrary positions. The one thing they are good at is playing games from the opening position, because that allows them to avoid the large fraction of positions where they would suck. But who wants that?

Er, people who play chess from the usual opening position?

The point is that people tend to play their own moves from that position, unless they happen to be operator for a NN engine in a computer tournament. And then they would like to get expert advice from an engine on the positions that result. LC0 is not very good at that; it will often come up with tactical blunders.

Not only that. On weird, unusual, "constructed" positions, Lc0 can be pretty bad both positionally and tactically, and I am not sure that can be corrected in the zero or quasi-zero framework. Large disbalances, endgames are other cases where Lc0 often sucks, but maybe that can be corrected.

[jp]

Obviously the inferiority of NN engines for analyzing (that is in arbitrary position) depends on the dimension and structure of NN. In the case of an NN engine with bigger and better structured NN this issue is smaller.
We can hope the development of NN engine will reduce this issue.

Whether there's a better structure for an NN is an interesting question, but just making the NN bigger and bigger isn't a better solution than just making a traditional engine search deeper and deeper.

[chrisw]

More important is that the NN-based engines aren't really very good at Chess in general: they seem to be quite inferior for analyzing arbitrary positions. The one thing they are good at is playing games from the opening position, because that allows them to avoid the large fraction of positions where they would suck. But who wants that?

Er, people who play chess from the usual opening position?

The point is that people tend to play their own moves from that position, unless they happen to be operator for a NN engine in a computer tournament. And then they would like to get expert advice from an engine on the positions that result. LC0 is not very good at that; it will often come up with tactical blunders.

Not only that. On weird, unusual, "constructed" positions, Lc0 can be pretty bad both positionally and tactically, and I am not sure that can be corrected in the zero or quasi-zero framework. Large disbalances, endgames are other cases where Lc0 often sucks, but maybe that can be corrected.

A/B engines with hand crafted evaluations working on general principles are very good at applying their principles onto all kind of chess positions. But, at same time, there are huge holes in hand crafted evaluations based around a) fortresses and similar and b) marrying together the forcedly material part of the evaluation with other aspects relating to the king and similar dynamic concepts. Result is a sometimes complete blindness and stupidity.

NN engines have other problems, but they are quite good at fortresses and king aspects.

I mean what do you guys want? Endlessly expressing dissatisfaction.

[jp]

NN engines have other problems, but they are quite good at fortresses and king aspects.

But are they? Has anyone looked? What eval does Lc give for the fortress positions that are shown in this forum like this one: viewtopic.php?f=2&t=70507 ?