Search depths of Chess Challenger 7

[mclane]

Without dedicated chess computers our hobby would not have had so many good programmers such as spracklens, kittinger, ed schroeder, richard lang, frans morsch, david broughton, mark taylor, julio kaplan, mike johnson, ron nelson, ulf rathsmann, thomas nitsche and elmar henne, David Lindsay, … johan de koning, chrilly donninger,

[AlexChess]

Without dedicated chess computers our hobby would not have had so many good programmers such as spracklens, kittinger, ed schroeder, richard lang, frans morsch, david broughton, mark taylor, julio kaplan, mike johnson, ron nelson, ulf rathsmann, thomas nitsche and elmar henne, David Lindsay, … johan de koning, chrilly donninger,

It's why I'm still on Talkchess. Some myths like Ed Schroder (Gideon Tasc Chess Machine) , Chris Whittington, Robert Hyatt, (Richard Lang I think is Mephisto here, I talked to him in 1987 when my company Intelligent Games organized THE 1987 WCCC in Rome) are still here!!!

Mclane It's always a great pleasure to talk with you, remembering the History of Chess Computers together. If moderators ban me for defending my reputation from Eduard Nemeth, please register on the BanksiaGui forum , where I'm admin

[towforce]

Without dedicated chess computers our hobby would not have had so many good programmers such as spracklens, kittinger, ed schroeder, richard lang, frans morsch, david broughton, mark taylor, julio kaplan, mike johnson, ron nelson, ulf rathsmann, thomas nitsche and elmar henne, David Lindsay, … johan de koning, chrilly donninger,

+1 well said!

I doubt that anyone is going to do this, but it might be possible to create a test suite to work out the full width search-depth of a chess engine (or at least a good approximation to it). Find some positions in which selective search would normally miss the key move, and try these out to see roughly how deep the full width search is going.

If the request was to determine the maximum SELECTIVE search depth using a black-box technique (without looking at the source code), I don't know how you'd go about accomplishing this right now, because the moves selected for the selective search can vary between engines.

[AlexChess]

Without dedicated chess computers our hobby would not have had so many good programmers such as spracklens, kittinger, ed schroeder, richard lang, frans morsch, david broughton, mark taylor, julio kaplan, mike johnson, ron nelson, ulf rathsmann, thomas nitsche and elmar henne, David Lindsay, … johan de koning, chrilly donninger,

+1 well said!

I doubt that anyone is going to do this, but it might be possible to create a test suite to work out the full width search-depth of a chess engine (or at least a good approximation to it). Find some positions in which selective search would normally miss the key move, and try these out to see roughly how deep the full width search is going.

If the request was to determine the maximum SELECTIVE search depth using a black-box technique (without looking at the source code), I don't know how you'd go about accomplishing this right now, because the moves selected for the selective search can vary between engines.

At that time there was some tests like Europe Echess's Pierre Nolot ones. Now modern engine solve all them is seconds

[mclane]

I am still the opinion that NN chess engines are better then normal engines but not really capable to create a plan the way strong chess players did like fischer.
Ok i know the point that programmers say: why do we need a plan when we can beat humans without a plan ?!
But the intention is to beat opponent chess engines with a long time plan.

Now i hear the point: plans were made by humans because they cannot calculate and do search as deep as stockfishor any other modern engine.

So why do we need this primitive method PLAN when we can do it our way.

Ok.

I guess there will always be a „reason“ not to make progress.

In the moment the first engine does a long range plan , and beats the other NN engines, the programmers will maybe think different.

Nolot was a collection of positions with very difficult solutions.
Unsolved positions.

There had been many many test suites over the ages.

My point is that computer chess at universities was important in the ages where no hardware was universally in the homes of the common people.

In the moment first 8 bit homecomputers with z80 or 6502 came out, hobbyists and normal people like schroeder, broughton, richard lang etc etc. Showed that university guys are not better then normal hobby programmers with talent.

That when the university guys could only win with big hardware and even that changed.
In 1986 ed schroeder was close to win against the big boys with an 8 bit cpu and 32 KB engine with 8 kb ram.


I often think these hobby programmers were like Don Quichotte against the universal windmills.

They showed very early that hardware is not everything but software is the key to strength.

[AlexChess]

I am still the opinion that NN chess engines are better then normal engines but not really capable to create a plan the way strong chess players did like fischer.
Ok i know the point that programmers say: why do we need a plan when we can beat humans without a plan ?!
But the intention is to beat opponent chess engines with a long time plan.

Now i hear the point: plans were made by humans because they cannot calculate and do search as deep as stockfishor any other modern engine. [...]

Interesting comment, thanks!

[towforce]

In the moment the first engine does a long range plan , and beats the other NN engines, the programmers will maybe think different.

Very interesting point, but I'm not sure I am fully in agreement with it. I think that picking the best move based on the data trumps making plans in chess.

"No plan survives contact with the enemy" - Helmuth von Moltke (link)

[AlexChess]

In the moment the first engine does a long range plan , and beats the other NN engines, the programmers will maybe think different.

Very interesting point, but I'm not sure I am fully in agreement with it. I think that picking the best move based on the data trumps making plans in chess.

"No plan survives contact with the enemy" - Helmuth von Moltke (link)

Completely removing classical evaluation from Stockfish (for me) isn't a good move . I would prefer to still have Hybrid evaluation. Now I hope that they will add strategic patterns to apply them according to the position when out of the book.