Visualization

[vier]

But who made the change in the GoGoD file?

Could it be that you changed it yourself and later forgot?
I have a gobase source dated 1998-07-28 that says
US[Jan van der Steen] C[These two players were the only 9 dan players at the time.]
and a source (from GoGoD?) dated 1999-05-31 that says
US[GoGoD95] GC[1st game between 9-dans].

[oren]

Go players have a handicap when it comes to visualization when compared to chess players, due to the lack of notation. When you read a book of chess games, variations are specified textually ("1.e4 e5 2.Nf3 Nc6 3.d4 ed 4.Nxd4" etc.). It is a standard skill one picks up to read the text of the variation and execute the corresponding moves in your head. On the other hand, go variations are generally specified by actually displaying the moves in the diagram, which means that the visualization has already been done for you!

So I think you have to do a little work to build up your visualization muscles. Lay out the current position on a board or in a go program. Look at the variation. Now look back at your board and imagine the variation being played out. Hold the final position in your head. Now play it out again with stones, from memory. See how the board is different from what you had in your head (it is possible, for instance, that you "saw" all the stones but didn't see some implication of the position until you could see it with your eyes). Repeat!

I wonder if this is completely true. It took a while to build the skill of looking at a 100 move diagram and visualizing the board state as the game progressed. It might be "easier" than chess, but it's not so easy as you make it out I think.

[dfan]

Yes, there is definitely a different skill that you pick up, of looking at the final diagram and playing out the numbered stones in your mind. I feel that this is pretty different from the process of looking at a position and imagining the position at the end of a variation, though, which of course is what happens when you perform reading during actual games.

[Uberdude]

When people ask "Why is Go hard for computers?", particularly in comparison to chess, the bigger position / game tree complexity is often given as a reason (plus harder to make evaluation function, thanks neural networks). But I think perhaps the more pertinent question is "Why is Go so easy for humans?" given you might expect it to be so much harder with so many more orders of magnitudes of possibilities. I think a big part of that is that Go is easier to visualise because the pieces don't move so the evolution of a position over time is (mostly) additive and amenable to chunking whereas with chess the pieces move around. That makes reading in chess much harder for me, but is that just because I'm a weak chess player? Is the kind of additive visualisation of Go inherently easier for the average human as I suppose, or is it just easier for me because I've trained it on my way to becoming a relatively strong Go player? Or perhaps I was better than average at it before I started Go, and that helped me make decent progress.

[dfan]

I generally find chess visualization easier than go visualization, but that is probably just because I have more experience with chess than go, especially when I was young. You learn to chunk in chess just as with go, and in fact I often find it easier because chess positions have more "texture" (different sorts of pieces that interact in particular geometric ways) as opposed to being more abstract (at first glance) collections of homogeneous stones.

In fact my general overall visualization abilities are extremely poor (I don't actually see anything in my head the way I can hear things in my head) and perhaps this makes it harder for me to imagine ten additional stones being placed on a board than to imagine the pieces in front of me being moved around ten times. Despite my supposed expert level in chess, I'm completely hopeless if I don't have a board in front of me as a reference point.

[John Fairbairn]

I'm sceptical about the sort of visualisation that seems to be the one talked about here - seeing something concrete in your mind's eye.

Surely the goal is to get chunks into your subconscious, that is your neural network. It seems highly unlikely that the brain stores these chunks as images. I have no idea how they are stored but I strongly suspect they are in something like a JPEG's compressed pixel format which can be unpacked into an image (rather slowly, I expect) by some decompression algorithm.

Of course, trying to visualise a chunk may be a good way for some people to send strong signals to the subconscious that this is something you do want to compress and store (others may use words, sounds or associations, etc). But that doesn't mean there is necessarily any correlation of efficiency between forming an image in the mind and storing it. It may work for you but other ways may work better if you could learn them.

In fact, it may hardly work at all. There is a famous psychology experiment by de Groot cited endlessly in chess books which showed that chess masters did much better than amateurs in reconstructing positions shown to them and then hidden. The explanation was that the masters saw chunks, which resided in their subconscious. I think it's often said they could visualise the positions, but I suspect in reality they were just unpacking coded chunks. The amateurs had no or few chunks in memory so had to stare intently at the board and try to force themselves to remember it, and I suspect that is the only place where visualisation came in. It's a sort of default option if you're up a gum tree. But it didn't work very well for the amateurs.

[dfan]

I'm sceptical about the sort of visualisation that seems to be the one talked about here - seeing something concrete in your mind's eye.

I do not see things in my mind's eye at all, and yet am able to "visualize" in both chess and go, so I certainly agree with you that visualization as it applies to lookahead in games is a different beast. I think that people have been using the term "visualization" here mostly to mean "accurate lookahead" in general - at least I have.

In fact, it may hardly work at all. There is a famous psychology experiment by de Groot cited endlessly in chess books which showed that chess masters did much better than amateurs in reconstructing positions shown to them and then hidden. The explanation was that the masters saw chunks, which resided in their subconscious. I think it's often said they could visualise the positions, but I suspect in reality they were just unpacking coded chunks. The amateurs had no or few chunks in memory so had to stare intently at the board and try to force themselves to remember it, and I suspect that is the only place where visualisation came in. It's a sort of default option if you're up a gum tree. But it didn't work very well for the amateurs.

Indeed, studies have shown that masters are much better than amateurs at reconstructing real chess positions but are hardly, if at all, better at reconstructing random configurations of pieces.

[dfan]

A video of GM Patrick Wolff doing a couple of these chess reconstruction exercises and talking about his thought process is here.

[Gomoto]

@OP:

playing lots of games, reviewing every game I play, looking at many pro games, doing tsumegu regulary, playing around on the go board aka studying go every day, ...

I for one do not "visualize" the stones, the stones are not placed on the empty places of the go board in my brain while I ponder variations, it more feels like: I play here, he plays there ... nonetheless I ponder about whole sequences of plays but there are no stones before my eyes in my brain, just (invisible) tagged places

[EdLee]

I do not see things in my mind's eye at all,
and yet am able to "visualize" in both chess and go,

Hi dfan,
Could you elaborate what you mean by the first part ?
Does it mean, when you close your eyes --

• You cannot visualize anything, any shapes or images, at all ? Say, a circle;
• Then how can you recognize any shapes ? Like faces, animals, letters, words, etc. ?!
• In your dreams, you have no visual components at all ? What about sounds, speeds, gravity, touch, smells, sense of direction, etc. (i.e. all physics) in your dreams ?

Thanks.

I for one do not "visualize" the stones, the stones are not placed on the empty places of the go board in my brain while I ponder variations, it more feels like: I play here, he plays there ...

Hi Gomoto, same questions as above! Could you elaborate how you read in Go ?!

Fascinating.

Some years ago I was chatting with someone online, and they said, one day, they asked their parents, "Do you hear music in your head ?" And the answer was No !
I was so surprised, like I am now, with dfan's and Gomoto's descriptions.

[dfan]

I do not see things in my mind's eye at all,
and yet am able to "visualize" in both chess and go,

Hi dfan,
Could you elaborate what you mean by the first part ?
Does it mean, when you close your eyes --

You cannot visualize anything, any shapes or images, at all ? Say, a circle;

Correct. Sometimes I can sort of begin to see things when I drift off to sleep.

Then how can you recognize any shapes ? Like faces, animals, letters, words, etc. ?!

Generating images is different from recognizing images.

As another example, I can draw quite well when there is a source in front of me but I'm completely hopeless when told to draw something from scratch without a visual reference.

In your dreams, you have no visual components at all ? What about sounds, speeds, gravity, touch, smells, sense of direction, etc. (i.e. all physics) in your dreams ?

I see in my dreams.

I wrote about my experiences a long time ago here. Things have not changed except for the fact that more is understood about it (back in 1999 I don't think there was even a name for it - now it is called aphantasia). If you're interested in other experiences there are 175 responses to this later blog post of mine.

I do just fine without visualization and in fact I like to think that one of the reasons I am very good at a lot of math and physics, including geometrical thinking, is that my body has developed really great coping mechanisms for it. Brains are great!

[Gomoto]

I can see the stones in black and white in my brain if I try to do it. But it is much more demanding and there is no benefit for my reading in go. I think it is much more effective for me when I think about the sequences, without actually building the images of the physical stones in my brain. I think I just imagine the shapes of the stones (a row of three, an empty triangle, a tigermouth, ... but also only as a "invisible" representation of this shapes not the actual physical shapes)

[kasai]

I'm still unclear on how you're supposed to read if you can't or have trouble seeing images at all in your brain (I don't quite have aphantasia, but the first time I was able to make an image in my head was when I was twelve and I thought that was what hallucination was ). If you're not reading by placing stones in your head, what are you actually doing? I understand reading out general stuff with "hane, then jump, then" etc etc, but how do you hold the previous moves in your head while you're reading further ahead, especially in something like a complicated fight? Is it like having a list in front of you that you read from? But if you can't see at all, do you have to repeat the list over and over to "see" the previous moves?

[dfan]

Mostly I try to imagine things in non-visual ways, like knowing the shapes of strings of stones, counting liberties, etc.

There are definitely chunks where you can group a bunch of moves together in a flow and pretty much know what the outcome is going to be, then come back and check move by move if you have to (e.g., squeezes).

I definitely have trouble imagining a variation and then backing up a bit and taking another branch (I generally have to start again from the root of the tree every time), but I think this is difficult for lots of people.

I used to think that my lack of visualization was really holding me back in games like chess and go, but if that were true I would be much better than my peers in correspondence games where it is not an issue, and I'm not.

[Gomoto]

I read out a chunk of the problem and usually only store the consquences for the next chunk not the whole chunk. This is quite error prone (shortage of liberties, etc ...). If I want to recheck a chunk I more or less have to go through the partial sequence step by step again (I will do this very fast, but mostly still step by step).