Equating Go Skill with Intelligence

    Keywords: Opening, MiddleGame

A bad habit I got into at one time was to consider my rank or skill as a sign of intelligence.

It really isn't. Sadly, one of the first books I had read had done so, and I have messed myself up thinking I am unintelligent if I am bad at Go.

-- Tim Brent

40 years ago, it was thought that ability at games like chess and Go was correlated with general intelligence, but AI findings so far suggest that if anything there is an inverse correlation!

Go embodies a microcosm of the real world; of concepts of growth and inter-object (ie group) competition for limited resources (territory) which, like the real world, can be addressed and navigated by the use of abstract models of what is going on and what is likely to happen. So from this point of view, it does indeed seem to be related to general intelligence - which, for a biological organism, can be characterised as the ability to behave appropriately, ie to make decisions that are in it's own long-term interests.

Unlike arcade games, Go requires patience, a willingness to accept deferred gratification. It is like Sudoko or Rubik's cube or a jigsaw puzzle in that each move has an apparently small effect, but a wrong turn can lead you off the cliff. Because of this, reading ability is key to beating others, so it is a game suited to semi-autistic savants who have prodigious rote memories. If i am right, DCNNs may yet prove one day to be superior to MCTS, which has an inherent flaw it cannot overcome, despite having carried Zen pretty high up Go mountain.

-- djh brown


Bill: I don't think that you can say that a professional 9-dan is more intelligent than an amateur 10-kyu, in general. However, one aspect or kind of intelligence is the ability to visualize spatial relationships. In that regard most pros are probably better than most amateurs. Also, research suggests that pros utilize areas of the brain not typically devoted to memory to store go information. Whether you want to call that intelligence or not is another question. :-)

I used to play tournament bridge, and my impression is that good bridge players are quite good at non-monotonic reasoning, and many are good at probabilistic reasoning.

As for go players, my impression is that we (good and not so good) have a high tolerance for ambiguity. That is also an aspect of intelligence.

Finally, as a group I find go players to be highly cultured, aside from any skill at the game. I cannot say that for other groups of intellectual game players.


Rafael Caetano: Bill, surely your last comment refers to Western go players, right? ;-)

Bill: No. I find Oriental go players to be highly cultured, as well.

Bob Myers: A few hours in a Japanese go parlor should disabuse you of this notion.

Rafael Caetano: That's what I meant... I thought Bill would agree, since he has lived in Japan.

Bill: Well, I guess Bob and I made different go-playing buddies in Japan. ;-)


Dieter: Q: Can we equate Go skill with intelligence ?

A: Define intelligence. Define Go skill. Consider the impact of intelligence on go skill and the other way round, logically where possible and statistically where necessary.

I will not undertake the endeavour here. From my experience with the Go world, which is limited to parts of Western Europe and the Internet Go community, I can safely conclude that Go players mostly have a high level of education. This relation can be used to attempt to "prove" a couple of things.

  1. Highly educated people are more likely to find out about the game. (That's what I think it says)
  2. Highly educated people are better at Go. (Possibly, but that's not what it proves)
  3. Go players are more likely to get a good education. (False, I think)
  4. Playing Go improves school results. (Equally false)

RichLancashire: could you also "prove" that highly educated people are more likely to stick at Go, say because of increased enjoyment?

NickJ? Actually, I understand that a number of studies were done whose results suggest that, yes, in fact, playing go DOES improve school results for elementary schoolers.

Phelan: Feel free to improve this article by posting links to them, then.

damien: While these studies are often cited by the Go community, I found it difficult to find any real published articles. The only well written study I found was on Go and children at risk for internet addiction. [ext] http://www.cinp2013.com/media/files/abstracts/0061%20Son.pdf (That said, I do believe more children should play go, and feel children can only benefit from the game.)

I have not discovered a relationship between Go rank and education. I do see a relationship between rank and the amount of time devoted to the game.

Maybe we could define intelligence as the ability to (rapidly) acquire new abilities . Go then merely represents a field where intelligence can be measured.

Charles Something to add to this: Go skill itself isn't really helped by the aspects of education that focus on control of inessential detail (rules, microscopic points about the endgame, the outer fringes of joseki) - at least for most amateur levels. It's a game, after all, and children may acquire the key abilities faster than some adults with university education who aren't lacking either in motivation or intelligence.

Malweth Though I'll agree with 3 of your four statements, I would have to say that in terms of refining thought process Go could have a very strong relationship to ones test taking or other scholastic abilities. The only caveats would be:

  • Go is more likely to improve analytical intelligences over social ones.
  • Go can certainly improve memory retention abilities.
  • Go can also have a detrimental effect on school... depending on how much time you spend playing Go instead of doing homework ;)

TDerz The beneficial requirements for being good at something (not limited to Go) are

  • Talent (something obscure from nature: strong legs, good brain etc. (German: "Begabung")
  • Environment (w.r.t. Go: Asia or not? Teachers, multiplicators, stimulating friends, available free time ...)
  • Motivation (Self- or external; call it longing for (a game), enjoyment (while playing one; even if losing against stronger opponents), lust, interest, fun))
  • Vision (= a goal/plan ("in a year I want to be better than X" etc.), this overlaps only partly with motivation)

Set realistic goals within a good learning strategy.

  • Discipline/Perseverance/concentration (to achieve the goal, keep the motivation, find a good environment ...).
  • positive feedback e.g. the joy coming from success (or compliments from teachers etc.)

- the more the better, you must have something from the list and they seem to be synergistic -

To return to the statement/question of this page, relation between skill & intelligence I would guess that the usual single dimensional tests referred to, focus (usually) heavily on mathematical stuff, which also is learned/already acquired but not intelligence in the sense of " deal appropriately with a new, unknown situation".

IQ-test in schools show furthermore - consequently in line with above -that there is a dependency upon where you where raised on the results (which children's songs and fairy tales you listened to - which animals & plants or machines & cities you saw alters your common knowledge for associative questions). Hence your background can influence your measured intelligence in a particular test.

I could try and equate the obscure talent with that specific intelligence. This would be wrong. Here comes a definition which I encountered recently:
"'Talent is (present), if you can do more than you know (think)'" ("' Talent ist, wenn Du mehr kannst als Du weißt'" (in German with the intended 2nd meaning of the pun ...than you could imagine; "Wissen" here within the meaning of true cognitive understanding/fact knowing/believing to know).

Hence, I think only it is safe to say: if all other factors are the same of 2 learning people (ceteris paribus), then a higher speed of improvement (e.g. coping with new concepts in Go)) could be a sign of of a very specific Go (Tennis/Guitar/making friends ...)-Intelligence (this was pointed out above by others). This specific intelligence had little to do with school results, success in professional life, being popular etc. (if there is no relationship, resp. correlation between this intelligence and the profession; It is however arbitrarily to assume that the correlation will be zero, as people would be more qualified for jobs/professions according to their likings & talents).

To relate it even more, next to the very-difficult-to-measure talent, the motivation affecting the result of improvement over time (=above measure for intelligence) is also difficult to establish accurately and objectively.

My conclusion: forget about the relationship with intelligence. If you do not want to leave the subject, read at least Daniel Goleman's "Emotional Intelligence", it showed me how important it is to be a complete person, hence being intelligent in many aspects (emotional, cognitive, motorically etc.).

I estimate that the motivation is the most important factor:

  • 1. some (obscure) talent will have caused/maintained it,
  • 2. to some extent, the right motivation can influence & set the other factors environment (look for a teacher, go to Asia), vision (take realistic goals), discipline (take study time) etc.

Talent in point 1 has been equated above with the specific intelligence, hence anyone would be allowed to induce from there, that after all talent/intelligence is the ultimate source of all skill. I believe however, that we are born, better procreated, relatively equal w.r.t. cognitive skills (this whole thing here ends up in the old dogmatic nature/nurture and chicken&egg discussions).

I respect anyone for special skills in some area, but would not dream of assuming beforehand something of this person in other areas.

We will all know similar examples (in Go): EUR 6dan's unsuccessful in real life and e.g. hopping in real life from underpaid job to dole (addiction/motivation + too much time); successful doctors, professors, business people (again 6dan EUR) being lonely or divorced (too much time?, not all-round personality? EQ?) ; ambitious 5- to 12-kyus which learned the game when they were 30-40, have a family with kids and few time/opportunity to improve; the prodigy neighbouring kids who reach simultaneously 2-3dan within 18 months (motivation) - and then quit! (lost goal,environment)! 2-dans in Go, Chess, Twixt, table tennis their business and many more unrelated hobbies & activities who are satisfied with their level and move on (motivation). 5 kyus (Go) who are best friends, talkative, sociable and know much about literature, history, events, travelling, countries ... (all-round). People who cheat (too early) on Go problem solutions (discipline).

ilan: For many years, I thought that IQ tests were fairly ridiculous and that it took a lot of chutzpah to try to measure human intelligence on a linear scale, as if it could be characterised by a single number. Then I realised that human performance in games like Go and Chess can be characterised by a single number... Oh well, I still think that IQ is meaningless, but I'm less sure about my reasons.

Migeru: I disagree that performance in Go and Chess can be characterized by a single number, because "A can beat B" is not a transitive relation. Apparently Rui Naiwei can beat Lee Chang-Ho, and he can beat everyone else, but Rui can't beat everyone else. It is a matter of style. Threfore there must be more than one characteristic of Go strength. If you want to do the same as psychologists do and posit that there exists one "g factor" that quantifies all there is to quantify about go skill, go ahead, but it will be as meaningless as IQ.

This is not to say that Chess and Go ranking systems are meaningless, but their meaning is nothing more than what can be derived from the algorithm by which they are computed.

ilan: But, au contraire, Go and Chess ability can be characterised by a single number, up to first order, that is. Transitivity is a second order phenomenon in the sense that it considers performance with respect to three players, as opposed to the simple (first order) A beats B. Clearly what I just said can be formalised. This fits well with the statistical correctness of ratings, since probability theory is essentially the study of first order approximations.

Migeru: IQ also does correlate with people's intuitive concept of intelligence, does it not? Anyway, I was just reacting to your use of the word "characterize".

I am also appalled at your lack of ambition regarding the scope of probability theory ;-)

ilan: Sorry, I was imprecise again. I meant that a probability is a first order approximation, as in: "the probability it will rain tomorrow is 1/2", which itself has many possible interpretations. Second order phenomena can also be analysed by probability theory, in particular, the variance. Speaking of which, it seems to me that your above example is actually third order, since the number of games you mention is about 10, which is significantly less than the square root of all professional games played in that period, that number being representative of the variance, i.e., second order.

Agilis: In response to the above, that seems to be confusing a debate about what IQ tests measure (I believe they measure a specific set of faculties) with what scores/rankings mean in themselves. Regardless of what IQ tests measure, it's a fact that X% of a population complete a given test with Y% accuracy, and IQs are defined off those statistics. Ranks are much the same way, people with similar rank are expected to perform roughly equally at a specific set of tasks, playing Go.

ilan: Right, I was about to edit my previous statement anyway. IQ tests are meaningful and do measure something, namely the ability to perform IQ tests, but this has no bearing on intelligence. You can take that as a good model of why I think Go playing has no bearing on intelligence either.

tderz: Ilan, I agree with you on "IQ tests are meaningful and do measure something, namely the ability to perform IQ tests". On Intelligence I introduce/quote Multiple Intelligence?.
Hence, measuring e.g. one's music-IQ tells us s.th. about his/her music intelligence. Only in so far Go playing skill has s.th. to do with intelligence, then of course with the Go intelligence (how to cope well with new situations/games).

This resembles the same chicken & egg question "What's there first? The talent, the intelligence or the motivation?".

surunveri?: The easiest way to approach this question is to ask: Is achievement in Go (rank) correlated with high intelligence? Or measured high IQ? And to do that you can ask the following question: What is the probability/IQ range of a person randomly selected from the street? What is the probability/IQ range of a 9-dan professional Go player? The answers to the questions seem so obvious that it would require actual evidence to contradict the expected result. Without actual testing this still providence reasonably high confidence in favor of IQ / achievement correlation. After which it boils down to a question of magnitude. IQ is very likely correlated to achievement in Go, but it is not the only factor. So the question remains, how much are they correlated?

tapir: The question is are they positively or negatively correlated, I recall a study where young promising go players achieved the amazing IQ score of 93. That should cure you from the notion that IQ-tests have any relation to intelligence or Go skill :)

Hyperpapeterie Not quite. One off the cuff explanation of those results is that the young professionals suffered on the IQ test because of lack of formal education (I believe they were Korean, and I think that many Korean professionals leave school at early ages). You can only get so far looking at correlation anyway, but I think a more suggestive study would concern strong amateurs in the CJKT countries.


Hicham: It seems to me that most here seem to be scared of making generalisations. I think one can say that a 9dan pro is generally more intelligent then a 30kyu. Or that go players are more intelligent then the whole population.

Not because they play go, but because intelligent people are more inclined to come in to contact and keep playing (in general ofcourse). They will be (in the west that is) often highly educated, and a high education has a statiscal link to intelligence.

Now do I think that you can say that a 10k is smarter theb a 30k? No not at all. Do I think that a 3d is (in general) smarter then a 30k? I think you could say that. Go is a very frustrating game if you get stuck and I think a dumber person would not persevere to make it to dan level. They might just be able to get there, but they would have to wok really hard.

You just should remember that these are only true in general,and that the differnce isn't linear probably. But I really think if one would seriously test this, that there would be some kind of pattern.

tderz: Quote Hicham: I think one can say that a 9dan pro is generally more intelligent then a 30kyu.
I think Einstein was 30 kyu (Go) around 1905 when he developed his theories. So you can say it, but showing you several counter examples would prove your hypothesis wrong. ... Go players are more intelligent then the whole population. comprises a logical flaw, because Go players are a part of the population. (How can you be smarter than yourself or slower than your shadow?) Comparing any Go player (e.g. 9dan, ELO nnnn) ceteris paribus (meaning "all other things remaining equal." A term used in research when isolating or changing one factor while keeping all the other factors unchanged.) with any other Go player (e.g. 7dan, ELO nnnn) would make sense. Think how difficult it is, two players with the same environment, training, teachers, opponents, age, health ... motivation, effort, other interests/distractions, boy/girlfriends ...
... do you get it? It is a very theorethical approach (ceteris paribus) which would work, but there are many more comparisions which make not much sense. In real life there are so many influences which seem not much have to do with intelligence.

Hicham:I think I have to repeat again that I don't think that Go is the causal factor here. (Go might just help intellectual growth but the difference will be minimal I think.)

If you would take the go players as a group and compare them to the population as a whole, there would be a difference in intelligence. But you would get the same when you took doctors or programmers.

Another example would be to compare the intelligence of people who did a ph.d, to people who have a master or bachelor in the same field(again as a group, not individual comparisons!). I predict there would be a (minor) difference. The same would count for goplayers.

Ofcourse the biggest difference between a 9dan pro and a 1dan amateur would not be their intelligence, but the upbringing and perseverence(not to say obsession) of the 9d. Intelligence is only one of the factors, but not one you can just ignore I think.


See also:

Intelligence


Equating Go Skill with Intelligence last edited by 58.165.169.148 on December 11, 2015 - 02:35
RecentChanges · StartingPoints · About
Edit page ·Search · Related · Page info · Latest diff
[Welcome to Sensei's Library!]
RecentChanges
StartingPoints
About
RandomPage
Search position
Page history
Latest page diff
Partner sites:
Go Teaching Ladder
Goproblems.com
Login / Prefs
Tools
Sensei's Library