Why are our instincts so bad?

[Calvin Clark]

I think it's a good question.

I suppose instinct is the move that you'd play without thinking. Or at least it's one of the moves that pops to mind, the "first feeling" that comes from the subconscious. One of my problems is that often I don't trust this move and grant it less credibility than something I've tried to crunch consciously out of principles. If I play the other move and it turns out badly, and later in in review my instinct move is pointed out to be correct, I have mixed feelings. On one hand, it's great that at least my instinct served me up a good move. But who is the demon that rejected it and can I have a talk with him?

It's not as simple as that, though. The ghost in my machine is not exactly Sai, and sometimes does serve up idiotic moves, which is why blitzing through ever game is not a great idea. Then the analyzing demon comes in handy, but it has its own problems and one may as well ask "why is our calculation so bad?"

One piece of advice I was given was to play more concretely. The vague move that looks so pretty in your imagination has to have some value you can justify. (But don't hate it just because it is pretty, I guess. )

[RobertJasiek]

I do not believe in instincts. I'd rather say that some of my first considered move candidates are bad and the reasons can be:

a) insufficient knowledge,

b) the correct / better move is non-obvious and requires effort of decision-making before it becomes apparent.

[John Fairbairn]

I have thought a bit more about this question, and I think that overlearning is important. That is, we need to practice eliciting the impulse to make a certain type of bad play, then inhibiting that impulse and making a better play. And do that again and again. If we can do that while playing, in the heat of battle, that's fine. But if we could do that it wouldn't be a problem, would it? One way, perhaps, is to review our games more than once, so that we face the problematic situation again, and reinforce choosing the better play. Another is to play over pro games and note where we have the impulse to make the bad play and see where the pro played.

I agree with this. Anki is a good way to handle it. As far as I can see, use of Anki in go seems limited to tsumego and joseki, but I think the most powerful use would be for whole-board positions. The ideal position to input would be e.g. of the type where you would have played an invasion but the pro played a reducing move, but positions from next move collections might do almost as well. Apart from memorising the right move, I think it is safe to assume that your subconscious will be noting all the salient aspects of the position on your behalf, though it would surely do no harm to help it along with some "deliberate effort" type thought.

I suppose it's this kind of knowledge that people aspire to when they decide to memorise pro games. But gluttonously memorising a whole game is boring, hard and not cost effective. Picking out the juicy bits is surely the mark of the epicure.

[John Fairbairn]

One piece of advice I was given was to play more concretely. The vague move that looks so pretty in your imagination has to have some value you can justify.

This seems to be a good formulation of something chess players will be familiar with. When computer chess first began, engines with one-ply search were abysmal. Even when some attempt was made to add an evaluation function, the result was little better than a random-move generator. But as soon as two-, or better three-ply trawls were added, the improvement was remarkable. Presumably his was because so many obvious captures were scooped into the net. Moving up to five-play was enough of a further improvement to make programs commercially viable.

Even though captures are not such a big factor in go as in chess, my experience is that forcing yourself to do a three-ply search on your candidate moves can offer at least two-stones instant improvement for most amateurs, and a five-ply search will get you winning tournaments.

There is, though, one very common trap that people fall into (and that includes me): virtuously doing a five-ply search on a half-dozen candidates and not liking the results for any of them. We then tend to let the eye alight on some other move and play that instantly. New candidates have to undergo the n-ply search as well. That is the commonest case of bad instincts that I know about.

[oren]

I do not believe in instincts.

This may be why your instincts are wrong.

[RobertJasiek]

I see instinct as a bad word for "very fast initial knowledge application" and not as what the word often seems to be associated with: "mixture of emotion / subconscious thinking without explicit knowledge application". Even my emotion about moves is driven by my knowledge.

[Fedya]

Even though captures are not such a big factor in go as in chess, my experience is that forcing yourself to do a three-ply search on your candidate moves can offer at least two-stones instant improvement for most amateurs, and a five-ply search will get you winning tournaments.

The plies have to be the correct ones, however. I don't know how many times I've read out something that works, only to find out when some of the stones in the sequence I read out have been played that there was an atari I missed. Or, my opponent plays a move I hadn't considered that scuppers my plans.

[oren]

I see instinct as a bad word for "very fast initial knowledge application" and not as what the word often seems to be associated with: "mixture of emotion / subconscious thinking without explicit knowledge application". Even my emotion about moves is driven by my knowledge.

So you're misunderstanding what people mean by instinct. Without some instinct, I don't think you could play go very much.

[RobertJasiek]

What then IYO do people think that instinct was? Do they have the same instinct as an absolute beginner without any knowledge of go theory and as a player with some established rank and some (or even much) knowledge? I.e., is IYO instinct independent of go theory knowledge? Depending on what it is that you call instinct, I might or might not have some.

[Tami]

What then IYO do people think that instinct was? Do they have the same instinct as an absolute beginner without any knowledge of go theory and as a player with some established rank and some (or even much) knowledge? I.e., is IYO instinct independent of go theory knowledge? Depending on what it is that you call instinct, I might or might not have some.

If it's of any help to you, I think most people mean "System One" responses by "instinct", meaning the kind of ideas that come to mind automatically as a result of past experience, study, and so on. I don't think we're talking about instinct in the sense of how animals know what to do in various wildlife scenarios. See https://en.wikipedia.org/wiki/Thinking,_Fast_and_Slow for an overview.

I think daal wants to learn how to overcome his faulty System One responses, and as I said before it may well be a matter of making deliberate attempts to apply different ideas, even when they go against such "instincts".

[Abyssinica]

I don't think we're talking about instinct in the sense of how animals know what to do in various wildlife scenarios. See https://en.wikipedia.org/wiki/Thinking,_Fast_and_Slow for an overview..

I thought we were talking about this and how it translates to most people's playing in Go.

[EdLee]

Hi Robert,

I see instinct as a bad word for "very fast initial knowledge application"

Is there such a word in German for "very fast initial knowledge application" ?
( I'm only curious; I speak zero German. )

What then IYO do people think that instinct was?

Again, I'm only curious what you think about this:
Suppose we label "the best move you come up with under some time period t" as a "t-instinct" move.

Examples: 1-second instinct move; 3-second instinct move; 0.5-second instinct move, etc.

Then, when you say:

I don't believe in instincts

(in terms of Go moves),
maybe we can rephrase it as you "don't trust your t-instinct moves,"
for some given range of t
-- Would this be a fair description ?

[Abyssinica]

Does playing a tesuji count as instinct if you can correctly identify it in a given shape in 1 second rather than playing what would be considered the "normal" response?

[oren]

I had a recent review which was interesting. I played a fairly slack move. My teacher said when he was back in Korea, his teacher would force him to just play the correct move on the board 100 times before moving on in order to build his instinct on it. When bad moves were made, play the right one a lot just to build up your instinct and get rid of the bad move.

[daal]

What then IYO do people think that instinct was? Do they have the same instinct as an absolute beginner without any knowledge of go theory and as a player with some established rank and some (or even much) knowledge? I.e., is IYO instinct independent of go theory knowledge? Depending on what it is that you call instinct, I might or might not have some.

I think you are right to question the term "instinct." I would say that it would be valid to use the word if we are referring to reactions that one might have in a similar situation that is not a go game. For example, if I am being surrounded by unfriendly people, I want to get out. This is an instinct. A similar situation sometimes occurs in a go game, and if running out seems to be an attractive option, it might be in part due to one's human instinct to get to safety. If this is a factor in one's decision-making process, then there is a problem, because while go situations may resemble real-life situations, the correct plan of action has nothing to do with real life, and everything to do with the particularities of a go game.

Other types of moves that could be influenced by one's instinct are: jealous moves, greedy moves, timid moves etc.

Clearly, there are other reasons for making mistakes that have more to do with misconceptions about the game or misjudgements of the situation, and although instinct may not be the right word, the problem is similar. There is something in our minds influencing us to make a bad decision. In these cases, we may be following principles that do not apply, basing our judgements on bad assessments etc.

Either way, what we are left with is that in go, things are often not what they may seem to be.