Interesting read about Fat Fritz 2

[gonzochess75]

I predict the whitewash will go about like this:

ASilver will argue that he has nothing to do with misleading in marketing of FF.
...

Check.

[acepoint_de]

For those who didn't get to see the FB post before it was hidden:

classical reversal of perpetrator and victim by him, seen thousands of times in political business and in internet «discussions». Not worth wasting more time on it...

Ciao

acepoint

[glennsamuel32]

Nakamura's take on this -- more the better

[glennsamuel32]

And chess.com

https://www.chess.com/blog/the_real_gre ... ontroversy

[Collingwood]

https://www.chess.com/blog/the_real_gre ... ontroversy

This part is amusing.

Never trust an author's claims about his own engine. Wait until someone else tests it.

There are a number of reasons why ChessBase's results might be misleading. First is the strange number of games; 1552 is a bizarre (and suspicious!) number of games. Ideally it would have (for example) 10,000 games, and that number would have been determined before the test.

Second, Silver picked his own set of test openings.

He's right. A strange number like 1552 suggests that person kept running games until the random fluctuations gave a result that looked better for him and then stopped.

[Collingwood]

Nakamura's take on this -- more the better

It does show how unfamiliar pro chess players are with computer chess, although they rely heavily on chess software. It sounds like the bigger issue was all new to Nakamura.

Wow! Pretty serious stuff here.

Wow! Very sneaky.

I don't really know what this means.

[gaard]

https://www.chess.com/blog/the_real_gre ... ontroversy

This part is amusing.

Never trust an author's claims about his own engine. Wait until someone else tests it.

There are a number of reasons why ChessBase's results might be misleading. First is the strange number of games; 1552 is a bizarre (and suspicious!) number of games. Ideally it would have (for example) 10,000 games, and that number would have been determined before the test.

Second, Silver picked his own set of test openings.

He's right. A strange number like 1552 suggests that person kept running games until the random fluctuations gave a result that looked better for him and then stopped.

Note: the reason for the odd number of games is that the suite used has 776 positions, which with reversed colours leads to 1552 games.

The games were attached to the article. It is plausible that they did exactly what they said they did. That doesn't, however, imply that all other testing conditions were fair.

[Collingwood]

The games were attached to the article. It is plausible that they did exactly what they said they did. That doesn't, however, imply that all other testing conditions were fair.

That makes some sense.

[gaard]

The games were attached to the article. It is plausible that they did exactly what they said they did. That doesn't, however, imply that all other testing conditions were fair.

That makes some sense.

The results I would like to have clarified are here: https://en.chessbase.com/post/how-a-neu ... rk-is-made



AS explained to me that the results indicate SF13 is 16 Elo behind FF2, not 32. I cannot reproduce these results, although I have tried, for example, by adjusting contempt to 24 from 0 for both players. Otherwise, FF2 consistently places below SF13 in my testing.

[the_real_greco]

https://www.chess.com/blog/the_real_gre ... ontroversy

This part is amusing.

Never trust an author's claims about his own engine. Wait until someone else tests it.

There are a number of reasons why ChessBase's results might be misleading. First is the strange number of games; 1552 is a bizarre (and suspicious!) number of games. Ideally it would have (for example) 10,000 games, and that number would have been determined before the test.

Second, Silver picked his own set of test openings.

He's right. A strange number like 1552 suggests that person kept running games until the random fluctuations gave a result that looked better for him and then stopped.

Note: the reason for the odd number of games is that the suite used has 776 positions, which with reversed colours leads to 1552 games.

The games were attached to the article. It is plausible that they did exactly what they said they did. That doesn't, however, imply that all other testing conditions were fair.

Either the 776-game book note was added after I published the blog, or I simply missed it when I was writing.

In any case using your personal opening suite is comically bad test design. Any results that don't use some standard book should be ignored.

But honestly? Tons of authors report results on SALC, low-draw books, etc. This community is really bad at self-testing.