Skip to main content
Chess

Return to Answer

added 1 character in body
Source Link
user1583209
user1583209
  • 20.9k
  • 3
  • 42
  • 99

Several points regarding your question:

Cannot compare computer and human rating

In your question you compare a computer with "players that would be similarly ranked."

However, the rating systems used in chess only describe the relative strength among players from the same pool. So a FIDE rating only tells you how strong you are relative to other FIDE rated players. An online rating from say lichess only tells you how strong you are relative to other lichess players, a computer rating only tells you how strong the computer is relative to other computers. In order to get comparability, you would need lots of computer-human rated games, which is not done.

So the computer may in fact play at lower absolute strength than you think based on its (computer) rating.

Many factors determine absolute strength

Chess is complicated enough that many factors determine "absolute strength". One (human) player might be blundering pieces now and then, another might be very weak in positional play, etc. Still they could end up having the same rating. So it might be very well that one player lets you build a very nice pawn structure and still beat you now or then.

There are many ways to cripple an engine

If you are a programmer, aiming for the strongest engine is a well defined task. However in your case you obviously played against a weakened engine.

If you are asked to reduce the strength of an engine, there are many ways to do this. You could just limit the search depth, which would make it vulnerable to tactics; or you could make the engine blunder pieces occasionally; or you could remove (or lower the weight) of one of the factors in the evaluation function. For instance if you remove the part which is relevant for "king safety", the engine might neglect the safety of its king.

It seems the programmer of your engine chose to reduce the relevance of "piece activity" and/or "space control". However another programmer might have chosen another way to reduce the strength. That's why I am not sure that your findings should be generalized to all computer engines.

Regarding the sample game

I would not describe this as "allowing to create a much better pawn structure". In fact at some point around move 12, you have a rather weak pawn structure with lots of weaknesses on the light squares, because your pawns are all on dark squares. Typically, if you have only one bishop left, you put the pawns on the squares of opposite color to that on which the bishop operates. This way the pawns orare not in the way of your bishop and also you cover both, dark and light, squares with bishop+pawns.

As mentioned above, to me the white side's play looked very passive. At some point the computer was moving his pieces only on the first and second rank, waiting for something to happen.

Several points regarding your question:

Cannot compare computer and human rating

In your question you compare a computer with "players that would be similarly ranked."

However, the rating systems used in chess only describe the relative strength among players from the same pool. So a FIDE rating only tells you how strong you are relative to other FIDE rated players. An online rating from say lichess only tells you how strong you are relative to other lichess players, a computer rating only tells you how strong the computer is relative to other computers. In order to get comparability, you would need lots of computer-human rated games, which is not done.

So the computer may in fact play at lower absolute strength than you think based on its (computer) rating.

Many factors determine absolute strength

Chess is complicated enough that many factors determine "absolute strength". One (human) player might be blundering pieces now and then, another might be very weak in positional play, etc. Still they could end up having the same rating. So it might be very well that one player lets you build a very nice pawn structure and still beat you now or then.

There are many ways to cripple an engine

If you are a programmer, aiming for the strongest engine is a well defined task. However in your case you obviously played against a weakened engine.

If you are asked to reduce the strength of an engine, there are many ways to do this. You could just limit the search depth, which would make it vulnerable to tactics; or you could make the engine blunder pieces occasionally; or you could remove (or lower the weight) of one of the factors in the evaluation function. For instance if you remove the part which is relevant for "king safety", the engine might neglect the safety of its king.

It seems the programmer of your engine chose to reduce the relevance of "piece activity" and/or "space control". However another programmer might have chosen another way to reduce the strength. That's why I am not sure that your findings should be generalized to all computer engines.

Regarding the sample game

I would not describe this as "allowing to create a much better pawn structure". In fact at some point around move 12, you have a rather weak pawn structure with lots of weaknesses on the light squares, because your pawns are all on dark squares. Typically, if you have only one bishop left, you put the pawns on the squares of opposite color to that on which the bishop operates. This way the pawns or not in the way of your bishop and also you cover both, dark and light, squares with bishop+pawns.

As mentioned above, to me the white side's play looked very passive. At some point the computer was moving his pieces only on the first and second rank, waiting for something to happen.

Several points regarding your question:

Cannot compare computer and human rating

In your question you compare a computer with "players that would be similarly ranked."

However, the rating systems used in chess only describe the relative strength among players from the same pool. So a FIDE rating only tells you how strong you are relative to other FIDE rated players. An online rating from say lichess only tells you how strong you are relative to other lichess players, a computer rating only tells you how strong the computer is relative to other computers. In order to get comparability, you would need lots of computer-human rated games, which is not done.

So the computer may in fact play at lower absolute strength than you think based on its (computer) rating.

Many factors determine absolute strength

Chess is complicated enough that many factors determine "absolute strength". One (human) player might be blundering pieces now and then, another might be very weak in positional play, etc. Still they could end up having the same rating. So it might be very well that one player lets you build a very nice pawn structure and still beat you now or then.

There are many ways to cripple an engine

If you are a programmer, aiming for the strongest engine is a well defined task. However in your case you obviously played against a weakened engine.

If you are asked to reduce the strength of an engine, there are many ways to do this. You could just limit the search depth, which would make it vulnerable to tactics; or you could make the engine blunder pieces occasionally; or you could remove (or lower the weight) of one of the factors in the evaluation function. For instance if you remove the part which is relevant for "king safety", the engine might neglect the safety of its king.

It seems the programmer of your engine chose to reduce the relevance of "piece activity" and/or "space control". However another programmer might have chosen another way to reduce the strength. That's why I am not sure that your findings should be generalized to all computer engines.

Regarding the sample game

I would not describe this as "allowing to create a much better pawn structure". In fact at some point around move 12, you have a rather weak pawn structure with lots of weaknesses on the light squares, because your pawns are all on dark squares. Typically, if you have only one bishop left, you put the pawns on the squares of opposite color to that on which the bishop operates. This way the pawns are not in the way of your bishop and also you cover both, dark and light, squares with bishop+pawns.

As mentioned above, to me the white side's play looked very passive. At some point the computer was moving his pieces only on the first and second rank, waiting for something to happen.

Source Link
user1583209
user1583209
  • 20.9k
  • 3
  • 42
  • 99

Several points regarding your question:

Cannot compare computer and human rating

In your question you compare a computer with "players that would be similarly ranked."

However, the rating systems used in chess only describe the relative strength among players from the same pool. So a FIDE rating only tells you how strong you are relative to other FIDE rated players. An online rating from say lichess only tells you how strong you are relative to other lichess players, a computer rating only tells you how strong the computer is relative to other computers. In order to get comparability, you would need lots of computer-human rated games, which is not done.

So the computer may in fact play at lower absolute strength than you think based on its (computer) rating.

Many factors determine absolute strength

Chess is complicated enough that many factors determine "absolute strength". One (human) player might be blundering pieces now and then, another might be very weak in positional play, etc. Still they could end up having the same rating. So it might be very well that one player lets you build a very nice pawn structure and still beat you now or then.

There are many ways to cripple an engine

If you are a programmer, aiming for the strongest engine is a well defined task. However in your case you obviously played against a weakened engine.

If you are asked to reduce the strength of an engine, there are many ways to do this. You could just limit the search depth, which would make it vulnerable to tactics; or you could make the engine blunder pieces occasionally; or you could remove (or lower the weight) of one of the factors in the evaluation function. For instance if you remove the part which is relevant for "king safety", the engine might neglect the safety of its king.

It seems the programmer of your engine chose to reduce the relevance of "piece activity" and/or "space control". However another programmer might have chosen another way to reduce the strength. That's why I am not sure that your findings should be generalized to all computer engines.

Regarding the sample game

I would not describe this as "allowing to create a much better pawn structure". In fact at some point around move 12, you have a rather weak pawn structure with lots of weaknesses on the light squares, because your pawns are all on dark squares. Typically, if you have only one bishop left, you put the pawns on the squares of opposite color to that on which the bishop operates. This way the pawns or not in the way of your bishop and also you cover both, dark and light, squares with bishop+pawns.

As mentioned above, to me the white side's play looked very passive. At some point the computer was moving his pieces only on the first and second rank, waiting for something to happen.