Could people defeat chess engines?

Philosophy of Science
1

For many years already chess engines beat human beings. As the calculating power of computers is larger than the calculating power of human beings, the only possibly prospect people have at winning is by emotion.

Emotions are agency to decisions. So then potentially emotion can deal with a zillion possible chess moves in an instant, by having all the chess moves as possibilities in a decision on them. Which is an advantage over the chesscomputer calculating options sequentially.

Then also, because the chesscomputer operates in a forced way by cause and effect, it should be possible to know what move the chesscomputer makes in advance. While a decision on the other hand is inherently unpredictable because of it being spontaneous. Which is another advantage for people.

A chess-yogi would be focused on the happiness of winning, rather than on the technical event by which a match is won.

That’s all I’ve got. Please, your thoughts on developing this idea further.

1 Like
2

Computers actually haven’t yet been able to beat humans yet.

Also, I think the prospect of using emotions to beat computers ridiculous. If your emotion influences your decisions, it typically blinds you to potential problems.

You mentioned emotions having a zillion possible solutions, and I don’t understand how this makes sense at all.

Do you know something ironic about what you’re saying? You’re giving logical advantages in which emotions are better than logic. Any logical difficulty could be solved using logic, e.g. the problem of cause and effect and emotions’ randomness could be solved by adding a factor of randomness to the computer’s choices, and it would be able to do it better than the emotions ever could.

Your choice being spontaneous = higher chance of blundering. To say that emotion-riddled spontaneity is better the cold, calculated, slow decisions and risks in a situation where you have the time to do so is to say that bare hands are better than spears.

3

Chess is a royal game. Goes with the role as a ruler. A human has not a snowball’s chance in hell of winning against the latest the computing world has to offer. Kasparov was dispatched without a sweat by Deep Blue in the 1980s. It’s 40 years now and I can see nothing but improvement in the game of computers.

Nevertheless …

Cuius rei demonstrationem mirabilem sane detexi. Hanc marginis exiguitas non caparet

4

Oh yes they have. Long since. There was a famous tournament, Gary Kasparov v Deep Blue, programmed by IBM, I think in the 1980’s (noted above). Since then, of course, the algorithms have become much stronger.

5

I think that a human can theoretically defeat a computer under certain conditions.

First, there’s the time per move. A computer has multiple computing cores, a high calculation frequency, and a large memory for storing variations.
A human player isn’t as well-equipped in this regard. Therefore, a human player should be given more time per move.

Second, there’s the competition. A computer, when in a single box, represents many computing machines simultaneously. A human player, on the other hand, is alone and can only hold one line at a time. Therefore, we need to find the right number of chess players on the human side. Not 100 or 20, but the necessary number of top specialists with a pre-agreed strategy for deciding which moves to play (people will also have to negotiate moves).

Third, although chess is a game in which the rules don’t change during the game (unlike human life), there is a premise that all variations have possible outcomes. But computers haven’t reached that level yet. When they do, the best we can hope for is a draw.

Fourth, there’s the opening base. How large a database of openings will the computer be given to process? Or should it learn on its own, like a neural network?

What you called emotions (and what everyone immediately criticizes as a mistake) I would call intuition. Human intuition is truly capable of many things and is still present in chess. But if several players are playing on the human side, the group won’t risk making a move based solely on intuition. Intuition, or the “magic” of chess, disappears in a collective decision.

In the end, we came to the conclusion that a single player lacks the power, and a group of players lacks intuition. Perhaps, if we approach this problem with these assumptions, the games will be interesting. I’ve never encountered anything like this in real life: before, they simply put a human player against a machine, now engines compete with each other. In the current reality, a single human can’t beat a neural network.

The most important thing in these arguments, in my opinion, is not to extrapolate them to everyday life. Yes, a computer may be better than a human at algorithmic tasks. But the world is algorithmic only to a certain extent (and even that is debatable).

1 Like
6

I don’t think of computers as “beating” humans, they are just more efficient at a particular function. But for the computer. the game doesn’t sit in the context of a comprehensive life experience and feelings of enjoyment, so the human has already “won” even if they “lose”.

1 Like
7

I misspoke. I meant that computers still haven’t beaten humanity at chess. It’s pretty good and can beat very strong players, but there are still humans who can beat computers at chess.

That said, I just looked it up, and the last time a human beat a computer was in 2006, so I concede the altered point.

1 Like
8

I think there’s a need to correct a bit of the context.

On calculating ability / brute force:
It’s actually a bit of a misconception that brute force is the reason that computers beat humans at chess. Computers could see millions of moves long before even deep blue, but struggled to beat grandmasters because grandmasters don’t calculate a broad decision tree but have all kinds of intuitions that allow them to hone in on solutions.
Even when computers were developed that could beat a grandmaster, they could only do so for a time, because they still need algorithms for deciding whether a given position at the end of a decision tree is good or bad, and chess players would ultimately notice these algorithms, evident in patterns of play. Once the patterns of play were noticed, the algorithm could be beat.

Which brings us to…

Kasparov / Deep Blue II
We take for granted that this is the point where computers eclipsed humans but this was a controversial match, at least within the chess community.
Because, for example, while the computer was trained on Kasparov’s entire game history, Kasparov got to see zero deep blue II games. And humans were assisting the machine; able to discard some proposed moves, making it extra hard to see any chinks in its armor.

After the competition, and all the positive press for IBM, the computer was disassembled so we’ll never know if humans would have found flaws.

Modern chess computers
It’s a different story in terms of modern chess computers though.
Modern chess computers are based on deep learning and don’t (only) brute force solutions; they genuinely have something of a chess intuition.
For example, stockfish might sacrifice a knight without having calculated a path to winning the material back, or mate. It just likes the position more with, say, a more constrained opponent position, even without its knight.

And no-one thinks humans have a chance against computers now. Chess computers are now around 800 elo better than world champion humans.
To put this in perspective: it means the chance of the best human beating the best computer in an individual game is 1/2,500. And a world championship is a best of 12 games.

The OP question: can emotion help us defeat computers at chess?
No. This is as likely to be effective as thinking about the happiness of winning taking my fat, 47 year old ass to olympic gold in the 100m dash.

2 Likes
9

You have to emphasize what computers do not have, but what human beings do have, which is emotion, or actually decision. So that is why I think the word emotion is better than intuition, because some algorithm might still be called intuitive algorithms, but no computer algorithm may be called emotional.

And if someone builds a “computer” with free will, then that would make the human vs computer distinction meaningless, so then the question becomes mute.

So you have to go with emotion, otherwise it’s meaningless.

There are some people who have a talent for calculating numbers just by seeing the result in their mind. Also people with talents like seeing prime numbers instantly (which people are strangely called “idiot” savant in the literature). There was also some reference to this in the big bang theory tv show, which associated prime numbers to color and smell.

But only if this works in combination with decision, would it have any interest for humans winning, because if it doesn’t work by decision, then the computers would also be designed to use it, so then it is no advantage for humans anymore.

Another thing is, there was a 2006 experiment where a database was searched without running the search algorithm, which experiment was based on decision. And although it is unknown if human beings played a fundamental role in this experiment, because of the socalled observer problem, as said, it would not be allowed for the computer to make any decision, because then it is no longer acting as a computer. So this technique may only be used by the human being.

So I imagine a psychologically talented and trained chess yogi, who like a martial artist, uses the computing power of the chess engine against itself.

Because the procedure of searching the database, this procedure works by cause and effect, same as the chess computer works by cause and effect. And as the experiment could in principle know what the result of running the search algorithm would be, without running the search algorithm, by a way of exploiting the possibility that it could run, so too it could be known what move the chess computer would make in response to a move, without actually making the move.

10

I should clarify one detail. I didn’t use the word “intuition” in my original post. I did use the word “gut feeling,” and I don’t know what the translated word should be to convey the meaning I intended. (These are questions of the untranslatable remainder, and they shouldn’t interfere with us.)

Regarding the decision, I’d like to say the following.

Through my personal research, I’ve (so far) found sufficient grounds to assert that every human decision has a “non-operational part.” This is the part of any decision that defies rational description and can be anything (faith, a sign, an opinion, confidence, etc.; it depends on the individual (and the person will never admit it).

A computer doesn’t have this. Any decision will be determined by certain weights and can be examined after the game.

When I talk about “feeling,” I’m talking about that very subjective feeling that arises at the board during a game. You simply see a certain structure and just know, “Damn, there’s something here,” even without knowing for sure.

When you play a human, they can also see it in their own structure while defending, and they too can be subjectively poisoned by a “negative feeling.”

With a computer, it’s more complicated. A computer doesn’t have feelings and analyzes every position using its own (known) scales.

I think that despite all the difficulties of beating a computer at chess, a human and a computer are still capable of beating a computer alone. So perhaps you should look here.

11

The Elo ranking system places Magnus Carlson as the highest rated player at 2882 and Stockfish as the highest rated computer at 3650, a 770 point gap.

These point spreads can be used to statistically predict the liklihood of winning. A 700 point spread equates to a 99% chance Stockfish will win.

So, given enough games, Magnus could theoretically win. The gap between human and computer grows wider over time, so the human chance diminishes. My guess is that the 99% liklihood would prove too conservative with empirical testing.

So, no, John Henry will never beat the steam shovel. Of course that’s no slight to humans, given we are their creator.

1 Like
12

Technically that would be unfair, but it’s not like either side stops thinking when it’s not their turn. Taking more time just gives the computer more time as well.

I’m pretty sure that an army of chess experts could not raise their rating by 700 points. It’s not like you can assign each a different move to consider.

It’s true that every position is either win, lose, or draw, and that in theory any move that isn’t a blunder (moves from draw to lose say) is a good move. In reality that doesn’t work. You don’t win by preventing a forced mate. You win by making it real hard for the opponent to not blunder.

It’s not like humans (not good at chess) can teach computers an opening book. All the modern ones figure out their own opening table. The best ones don’t even have chess built into them. They’re programmed only to play any game, and to learn. Then they’re explained the rules of the game, after which they’re given time to get proficient at it. A few days is plenty to beat any human.

Thanks for this history. +1 for the post.
I very much knew the rigged nature of that match where Kasparov was not allowed any access to the machine’s playing style. But I was unaware of the human intervention, which is outright cheating. The whole thing was a wildly successful PR stunt on the part of IBM, and it’s obvious why the machine was dismantled after the match. It’s job was done, and it wasn’t yet quite better than Kasparov in a fair match. One that was better would soon follow, but IBM wanted this particular PR stunt.

One can arbitrarily define that a chess move made by a machine is not an ‘actual’ decision, but it doesn’t make what the machine is doing any different that what a human does. It’s true that the machine does not mimic the way a human expert would approach the problem, but I could then equally argue that the human move is not an ‘actual’ decision since it doesn’t get there the way the machine does.

As for ‘emotion’, almost all game playing is improved by setting emotion aside. A great deal of sports plays on this, trying to get to the emotions of the opposing team in effort to draw penalties and eliminate key players. The strategy works, and significant practice is needed to resist it. Same goes for say police training. Emotions make generally for irrational decisions. Evolution selected for that. It has less use for rationality.

Interesting. I’ve never seen it called that, but there’s a part of everyone that is fast, efficient, and not part of conscious thought. It’s incredibly operational. For instance, my subconscious can solve 6th level differential equations in a few seconds, but it does so in analog, not in any way that the school teaches, which would equivalently take hours, or even near an hour with a calculator.
Likewise, good chess playing is a subconscious task. It’s learned just like learning to walk.

13

@syamsu, I wonder if you are enquiring into our ultimate computational capacity, if we’re open to the possibility of what we might call ‘supra-material’ action in humans?

One controversial and provocative line that’s somewhere around the intersection of philosophy of mind and philosophy of mathematics, is that humans may be capable of some kind of non-algorithmic computation. The infamous version of this line is the ‘Penrose-Lucas argument’. To my mind, the less problematic sub-part in this is Lucas’ argument that humans are capable of non-algorithmic computation, an argument that’s based around Godel’s incompleteness theorems. (The more problematic part is Penrose’s ‘non-algorithmic must mean quantum’. Even the Lucas part is not clear, IMO; I personally don’t buy it as a proof of non-algorithmicity. But the question remains open, as an empirical fact.)

What does any of that have to do with ‘chess and emotion’?

Well imagine a human who had this conjectural/putative non-algorithmic capacity intentionally online at all times, and who could deploy it in chess. Could such a person beat even the strongest algorithmic (computer-based) system? Perhaps.

And emotion? Others (I think you, too!) have reframed your original expression in terms of emotion, to ‘intuition’. Of course, as I think some have said, intuition might just mean unconscious-algorithmic-processing in the brain, which likely won’t outperform expert computer systems. But intuition as non-algorithmic insight might. (Very speculatively!)

Would the best use of a such supra-materially enabled human be to beat chess computers though? I’d guess not: there are plenty of other pressing problems for humankind that such a One might be interested in!

14

Chess ratings, despite their apparent objectivity, aren’t all that objective. These figures are arbitrary. For example, the gap between a rating of 800 and 1000 isn’t as large as the gap between 3000 and 2800.

As I wrote in a previous thread, my current rating (I thought) was 1800, and my peak was once 2000. I decided to log in and play a little. Within a couple of days, I’d dropped to 1600. Now, I’ve started to improve slightly and think I’ll get back to 1800 in a couple of weeks, but to get back to 2000 form, I’ll need three or four months of continuous practice, playing, theory, and solving problems. I might even be able to go beyond 2000 by no more than 20-30 points. But to reach 2200, I’d have to spend at least 1-2 years of nonstop practice, quit everything, and live solely on chess. The gap between 2000 and 2200 is practically unattainable for me (considering I’m not 12 years old and have other things to do besides chess). 2200 and above is the level where you’ve been hugging a chessboard since childhood: sleeping with it, eating with it, analyzing blindfolded positions while walking down the street.

So if your rating is 1600 or 1800 or 2000 or 2200, you don’t have even a 0.5% chance of beating stocfish. Although, there’s still a 0.5% chance of beating Magnus: for example, if he’s stressed, drinking heavily for a month straight, and not sleeping for a couple of days before your game.

Another point about rating. Sometimes, when playing a specific game, you elevate your combination to grandmaster level. Sometimes, when making a move, you overlook something that even a beginner should see (there are millions of examples of blunders at the top level). Rating reflects that you statistically play well more often.

Let’s say Magnus and Fabiano are playing each other: one blunders in one place, the other in another. One wins, the other loses. That’s why people still play tournaments against each other, even though a computer would beat them all. We value the element of unpredictability and chance here, and this element—the “great randomness” that operates everywhere—is ultimately what keeps a philosopher employed =))

15

For the 1st move you have 8 \times 2 + 4 = 20 possibilities. Say your move is E3. Now you’ve opened up the Queen. The next number of possible moves = 7 \times 2 + 1 + 4 + 5 + 5 = 29. Say your move is D4. You can now make a total of 7 \times 2 + 1 + 4 + 5 + 5 + 5 = 34. For this opening we have 20 \to 29 \to 34. I was expected a faster growth in the number of possible moves. :thinking:

16

It’s obviously easier to lose than win. The Elo system establishes a rating floor for established ratings, so it’s not possible to go from a 1500 to an 1100 assuming a 200 point floor from peak.

That you can’t get beyond a high point only means the Elo system accurately provides your peak strength and real improvement is difficult beyond that.

Simplify this to golf. If I average a 95 per round, it will be less difficult to move that average to 105 than 85, but should I play a 105 while my average is 85, we could compute my odds of winning as very high.

We can also likely say I’ll never pull the average to a 75 and likely never shoot a 75 even once.

My average is not arbitrary but fully objective. I could still lose though to a 105 average player, owing to an injury. Regardless, it still holds that I have a statistically specific chance of winning that can be computed. But the long shot can always win, just unlikely.

17

I’ve played casually since childhood and love the game but I never been able to comprehend how blindfold chess can be played. It must require such a feat of memory - eidetic memory even. Let alone simultaneous games blindfold! :scream: