Stockfish Natural TB loses heavily to Stockfish master

[Ras]

1. Nf3 {-0.01/16 0.52s} e5 {+120.00/17 0.35s}

Why is the black eval (SF-NTB) so strange?

[syzygy]

I think Marco is agent for commercial engines, and is trying his best to hinder the further development of Stockfish. TCEC is coming, and what a perfect opportunity to make Stockfish play bad. among its rivals...

Never attribute to malice that which is adequately explained by ...

[whereagles]

not sure it's productive for such a small community to get into conflict..

marco might be stubborn but at least he's trying for something.. im sure he'll make it work.. whether or not it's an improvement we'll see. So far seems not.

[syzygy]

not sure it's productive for such a small community to get into conflict..

marco might be stubborn but at least he's trying for something..

Trying to do what?

I fixed the issue in a few lines, but it is not being considered. (Because it does not decrease the line count by the same amount as Marco's patch that rips out useful functionality. Seriously.)

I fixed another issue last year after someone complained that TBs were not working properly with multipv > 1, but Marco refused to read my explanations (complaining that I had not explained anything).

So whatever there is to be done has been done already, it seems.

Well, I am not saying that things are perfect as they are (or rather would be if committed). But before someone starts to "improve" things, it might be nice to know exactly what still needs to be improved. Then we could consider whether a solution is feasible.

[Damir]

Marco is wasting his time, not to mention Kai's CPU time as well...

[Michel]

I fixed another issue last year after someone complained that TBs were not working properly with multipv > 1, but Marco refused to read my explanations (complaining that I had not explained anything).

Unfortunately this is another one of Marco's very effective debating tactics. Strongly insisting on very detailed technical explanations but then making no effort whatsoever to understand them.

[duncan]

Well, I am not saying that things are perfect as they are (or rather would be if committed). But before someone starts to "improve" things, it might be nice to know exactly what still needs to be improved. Then we could consider whether a solution is feasible.

what things would you say still need to be improved?

[syzygy]

Well, I am not saying that things are perfect as they are (or rather would be if committed). But before someone starts to "improve" things, it might be nice to know exactly what still needs to be improved. Then we could consider whether a solution is feasible.

what things would you say still need to be improved?

I am asking you as a user

[BBauer]

As a user I use

// Step 4a. Tablebase probe
if (!rootNode && TB::Cardinality)
{
int piecesCount = pos.count<ALL_PIECES>();

if ( piecesCount <= TB::Cardinality
&& (piecesCount < TB::Cardinality || depth >= TB::ProbeDepth)
&& pos.rule50_count() == 0
&& !pos.can_castle(ANY_CASTLING)
&& alpha > -VALUE_KNOWN_WIN // Stop probing if we have
&& beta < VALUE_KNOWN_WIN) // reached a won position. joster
{


This gives me what I want.
The game of chess is not "natural". Think about studies or moves like Bu's last move against Carlsen with his rook. This move was not "natural".
Kind regards
Bernhard

[syzygy]

Hi Bernhard, you are using SF for analysis, right?

Did you try the early_mate modification?
https://github.com/official-stockfish/S ... /pull/1241
https://github.com/syzygy1/Stockfish/tree/early_mate

Do you use SF+TBs also for analysing positions with 6 pieces or less, so positions that are in the TBs?