Rational choice by amateurs

[Bill Spight]

The other day I ran across this talk by Malcolm Gladwell ( https://www.youtube.com/watch?v=PwWq1K- ... gs=pl%2Cwn ), on, of all things, marketing. He reports research that may possibly bear on amateur choice of moves at go. This is all speculative, OC, but I thought I would share.

The research is based upon expert ranking by taste of 44 different brands of jam. The researchers picked the top and bottom rated brands and two intermediate brands and had amateur subjects (college students, OC ) rank the four jams. The college students agreed pretty well with the experts. Their ranking had a correlation of 0.55 with expert ranking. Then the researchers set a different group of college students a slightly different task, to rank the jams and also to explain their ranking. (I think it is important that they were told to come up with explanations before making their choices.) These students' ranking had a correlation of only 0.11 with expert ranking. That's a big effect. Gladwell also mentions a similar effect with amateurs' preferences in art. When asked to choose between two things and to say why, amateurs do worse (at least for certain kinds of choices) than if they are simply asked to choose. Thinking about their choice is counterproductive.

Gladwell goes on to offer a possible explanation for these results. How much of this is his own speculation and how much comes from the literature I couldn't tell. Experts, he points out, have a vocabulary to explain their choices, amateurs do not. So when asked to explain their choices, to make rational choices, amateurs oversimplify. This oversimplification leads to worse choices, in many cases. (Social science research in always iffy. )

Edit: This may relate to our recent discussion about style. Consciously sticking to a particular style of play can be an oversimplification.

All of this is in line with my thinking about amateur go. IMO, rationality is overrated. (Now, I have made myself an expert in certain rational aspects of go, but that's me. I can lay no claim to expertise at go itself.) That's why I emphasize seeing (and looking and feeling) over reading. Beginner's mind is rather good.

This also underscores why imitation of expert play works so well. It's never too early to play over pro games. Imitation allows us to pick up good play without having to say why. Scoffers say that you should stick to plays that you can understand, that is, that you can explain rationally. But for amateurs, is that a recipe for inferior play based upon oversimplification? Speaking for myself, if I wanted to play a game that I understood, I would stick to tic-tac-toe.

Let me mention my heuristic for choosing between two gote that are relatively independent. Treat them as miai, so that if you play one your opponent plays the other (or thereabouts). Compare the two different whole board positions. Which do you prefer? Note that I am not asking for a reason why. But if you have one, why not?

Am I against reading? No, of course not, at least once you have reached SDK level. Takemiya says to play what you want to play. OC, if that's all you do, that is also an oversimplification. Here is my suggestion. Use your intuition and judgement to discover your preference of plays, perhaps utilizing my heuristic. Then use reading, not to justify your preference, but to try to refute it. If you can't refute the play you want to make, go ahead and make it. Don't worry if it is a leap into the unknown. You can learn from that. (Note that this is a possible antidote to wishful reading, as you are not trying to find a line of play that works for yourself, but one that works for your opponent. )

What about the vocabulary of go? Go has a rich vocabulary, as well as actual proverbs, which it is good to learn. (But, IMO, beware of amateur proverbs, as they will tend to oversimplify.) Many important go terms are not well defined. It is not by logic that you can understand them, but by judgement. As your judgment improves, you become more expert. IMX, attempting to understand go terminology does help you to play better.

[Knotwilg]

When you say "rationalizing" do you mean "backed up by variations" or "backed op by verbal argument"? In the beginning of your post I thought you meant the latter, however in the second part you seem to point to the former.

In a previous discussion it was Shorin German I believe who pointed to "variations" as the language used by experts and "words" by amateurs, meaning that amateurs should rather spend time acquiring a "vocabulary" of sequences than one of terms.

[Bill Spight]

In a previous discussion it was Shorin German I believe who pointed to "variations" as the language used by experts and "words" by amateurs, meaning that amateurs should rather spend time acquiring a "vocabulary" of sequences than one of terms.

I pretty much agree. I think that well defined go concepts consist of pairings of positions and plays or sequences of play. OC, many of these concepts have names.

Help! I accidentally edited this note instead of quoting it. Is there any way to restore the original?

[Gomoto]

How about pros who have to explain why the play their moves? Probably their play will also deteriote.

Is there any data on experts that have to do a task and explain their choices at the same time versus only doing the task?

[Bill Spight]

How about pros who have to explain why the play their moves? Probably their play will also deteriote.

I can't speak for pros, but I doubt it.

Is there any data on experts that have to do a task and explain their choices at the same time versus only doing the task?

Well, in the research Gladwell cites it was the experts' choices that provided the standard against which to judge. However, he stressed that the experts can explain their choices in ways that make sense to other experts. OTOH, there is a fair amount of research that indicates that human experts are not as good as everybody assumes. I actually took the agreement between experts and amateurs as providing validation to the experts. OC, in go, the experts' judgement is tested every time they play.

[gowan]

I learned how to play go mostly by playing and observing what happened and by playing through pro games. I've always felt inadequate, when discussing games, in explaining why I prefer one move to another. In other words I play at a much higher level than my ability at explaining. When a kyu player asks me why I preferred a certain move I might say because it's a better move. Not much of an explanation. Another thing, we notice many pros showing us moves and their "explanation" is that it's a better move.

[Knotwilg]

I learned how to play go mostly by playing and observing what happened and by playing through pro games. I've always felt inadequate, when discussing games, in explaining why I prefer one move to another. In other words I play at a much higher level than my ability at explaining. When a kyu player asks me why I preferred a certain move I might say because it's a better move. Not much of an explanation. Another thing, we notice many pros showing us moves and their "explanation" is that it's a better move.

I'm frequently reminded of that when reading the long study journals by people like myself. We probably become better go writers but hardly better go players. It adds up to the frustration that "we are so good at reasoning about the game, why can't our results follow?"

[John Fairbairn]

Very interesting thread, Bill!

Gladwell also mentions a similar effect with amateurs' preferences in art. When asked to choose between two things and to say why, amateurs do worse (at least for certain kinds of choices) than if they are simply asked to choose. Thinking about their choice is counterproductive.

I'm not yet convinced by this. It all seems to fuzzy (as you say, it's a social "science").

Speaking to one's closely matched elite peers (i.e. pro to pro) where you can assume you all know most things about the subject (and there is little variation and there is no higher oracle) is a doddle compared to what amateurs face. The range of people amateurs have to explain things to is enormous. They can be called on to explain their moves/reasons to a pro, in which case they probably get tongue-tied or suffer other psychological blockages. They can be called on to explain to beginners, all without knowing what the beginners know already. Even with another amateur of ostensibly the same level there is so much neither of you knows and there may be little overlap of the portions you do know.

There is also explanation for instruction and explanation for entertainment. Pros tend not to instruct each other, at least directly (in my experience, in go, almost all of what pros teach other pros is about things off the board: attitude and mental discipline). They explain to amateurs or tv audiences for entertainment. Amateurs explain mostly to teach other amateurs, or to share experiences as fans. Also (in most sports and activities), most amateurs don't want to be pros and don't even necessarily seek to become stronger in actual play or performance. They just want to understand more about the subject so they can enjoy it better as fans.

So it seems to me that asking amateurs "to choose between two things and to say why" is not really very close to anything that happens in their real lives. In that case, interference (typically trying too hard to give the answer they think is expected) is almost bound to happen.

[daal]

It makes sense to me that amateurs may do worse when trying to consciously justify their moves with simplifications and half-truths than when they just play what their intuition (backed by some reading) tells them. I know I've spent a lot of time trying to find some principles to guide my play, and these "principles" are invariably not the whole story. There are hundreds of such principles, and hardly any way to prioritize them. Anyone basing their move choices on some proverb is invariably ignoring another.

[Calvin Clark]

Link to timestamp where the strawberry jam story begins. The whole talk is good, but a lot of the earlier stuff is not as relevant.

Suspending my disbelief for a while that there might be such a thing as strawberry jam 'experts' at any level that is useful to compare to go professionals...

What is mainly shown here is that when amateurs explain things it disrupts their access to their intuitive preferences. I am not really convinced it follows that their intuitive preferences are better, because as another anecdote in the talk shows (the Sprite one where customers were convinced the formula of Sprite had changed when in fact it was just the size of the lemon on the can), the intuition can be fooled by irrelevant information.

Numerous reviews over many years have shown me that my first feeling of a move is often better than the move I talk myself into playing. What usually happens is that I see the move instantly, and then think about it a bit, try to read some, etc. Then I see some drawbacks so I look for other choices (or I just force myself to look for other choices just to have choices). At some point either the clock or a timer in my brain says, "time's up, better move now" and then I'm more likely to play the last thing I thought about rather than the first thing. I think Kotov talks about this problem. Of course, my first feeling can be junk, too, otherwise I'd be better at blitz than anything else. But maybe I should make a mental note of my first feeling and revisit that right before playing, if only to reflect on whether that still may be the best play.

[sorin]

How about pros who have to explain why the play their moves? Probably their play will also deteriote.

Is there any data on experts that have to do a task and explain their choices at the same time versus only doing the task?

One good example is Hajin Lee explaining her thoughts while playing online, for her video series.
Not sure how much of her power is lost this way, we would need to ask her I guess.

[Gomoto]

(I am not talking about explaining while playing to an audience. But for example to give a reason for every move in a written form.)

[RobertJasiek]

Bill, please list the important terms you think should be defined and I have not defined yet.

daal, discard bad principles that cannot be prioritised. Good principles specify (or have given by context of presentation) their scope of application, work together with dynamic input (reading, analysis) as necessary and can be prioritised. Priority can allow parallel processing of not directly interacting principles.

[sorin]

(I am not talking about explaining while playing to an audience. But for example to give a reason for every move in a written form.)

I am not sure why the distinction is important; I think what she is doing matches Bill's example, basically verbalizing the reasons behind moves in real time. I don't think writing the thoughts down makes a difference.

[Gomoto]

There are probably two different possible negative impacts of explaining your play (decisions)

-distraction (I do not want to talk about this, this is trivial)

-formulating explicit grounds for your play may lead to different move selection by your brain, you may select lesser moves (I understand this is true for non experts in the study. I claim this is probably also true for experts, but this is not covered by the study.)


(It is also possible these two aspects are not easy separable or are indeed the same aspect)