Life In 19x19

Baduk TV often has the opinion of 4 AI's running simultaneously, which is also how I think it should be used. Just treat them like another pro with their own biases, not an oracle. And now I'm mixing up the plurality of my pronouns like someone who just got up from bed, even though I haven't

John Fairbairn wrote:I'd be interested to know what you and anyone else think W+5.7 means.

Why would it be difficult to understand? Ishida Yoshio had a series of articles in Go World called "Bad Moves: How many points do they lose?" where he wasn't shy about analyzing certain moves that were considered bad and putting points values on how much they lost. This is the same thing, except the player is probably stronger than Ishida and you can ask about any position you like.

Why would it be difficult to understand? Ishida Yoshio had a series of articles in Go World called "Bad Moves: How many points do they lose?" where he wasn't shy about analyzing certain moves that were considered bad and putting points values on how much they lost. This is the same thing, except the player is probably stronger than Ishida and you can ask about any position you like.

Well, lightvector seemed to agree it's not straightforward, but in any case I don't see that the comparison with Ishida is really valid. He takes local positions not whole games, admits other pros have different opinions, and uses numbers as a journalistic gimmick, simply in lieu of a table he also kindly provides. Minor bad moves are 2 or 3 point losses, bad moves are 5 point losses and very bad moves are 10-point losses. He does not explain his counts, beyond using the criteria of this table. So apples and pears come to mind.

Furthermore, Ishida claims moves that lose ten points are few and far between (which suggests to me he may have been at the cider ), and his aim is not to teach an understanding of counting, but to make the amateur realise that his main fault is not making bad moves per se, but making an accumulation of minor bad moves and bad moves. This is valuable information that AI is not able to verbalise.

W+5.7

What does it mean?
it depends

John, I'll have to disagree with you on this.

"...He takes local positions not whole games..." AI does this also, and better than humans do

"...admits other pros have different opinions..." Just as different AIs may have different opinions. Though this seems to be less true as AI gets better.

"...He does not explain his counts..." To some extent AI doesn't either, but AI numbers can be traced back to something meaningful.

"...his aim is not to teach an understanding of counting..." This is similar to AI.

"...but to make the amateur realise that his main fault is not making bad moves per se, but making an accumulation of minor bad moves and bad moves." This is something that AI point loss and win rate makes clear at least to me. Some moves are worse than others, but the accumulation of even small errors over the course of a game leads to a loss against a much better opponent.

"This is valuable information that AI is not able to verbalise." While AI won't verbalize it, I think AI is uniquely able to show us that it is true.

John, I'll have to disagree with you on this.

Fine, but you don't seem to be disagreeing with me really. My main point was that I didn't think the comparison between AI and Ishida was valid - apples and pears. You haven't mentioned that, so pro tem I'll assume you agree.

I could debate the other points, but instead I'll try to take a different approach that may let a bit more light in. I'll ask you to try to view things from my point of view.

My go activity can be split into three unequal parts. The smallest by far is reading SL and, very occasionally, other internet material. I get very little out of this in go terms, but I have an abiding interest in the way other people think and for that SL does provide material.

I read Japanese go magazines, newspapers and books, and sometimes Chinese material. Nowadays I rarely look at Korean material. This is the bulk of my go activity.

I spend a lot of time writing books. Overall, I probably spend more time on this than anything else, but a high proportion of this time is actually devoted to reading books and magazines. For simplicity, therefore, let us say I spend my time between reading SL and reading Japanese material. I rarely look at AI programs and don't watch videos (can't hear them, but prefer books anyway).

What I see when I contrast these two sources (SL vs Japanese) is a chasm. SL seems dominated by people who have mathematical interests or careers of some kind (I include programming in that) in the real world. Most of these people love AI and spend a lot of time on it, but given the nature of the SL readership (as I see it), they are preaching to the already converted.

When I look at Japanese material, I see very, very little reference to AI or numbers or rules or maths of any kind. None of these elements is run down in any way, and any of them can be respectfully highlighted at times (though in a quite different way from SL). The people who write this Japanese stuff (professional journalists) are preaching to the generality of people.

If I look a little more deeply, I see SL as a group of amateurs who are not themselves improving as a result of AI and who are not helping to grow the game. I see the Japanese press as well supported by professionals (some of whom actually are improving with AI!) who are not really preaching but explaining, and who above all are successfully growing the game.

Now, I have to stress that there is no question in anyone's mind that AI is stronger than any professional, probably by a large margin, by the criterion of winning games. But humans make mistakes for all sorts of extraneous reasons (e.g. time trouble) and so lose games they might win in perfect circumstances. By that measure, they are not so distant from AI as many people make out. That aspect tends to get lost on SL.

There is also no question that AI is fascinating in many ways. But the fascination is strongest for those already inclined that way and who have a relevant background. The fascination for everyone else is much less, can be minimal for many, and actually off-putting for some (personally, I'm somewhere in the middle). That gets lost on SL.

In the professionally produced Japanese material, I think nothing gets lost. It is well balanced. Apart from study material, which seems to be what SL denizens are mainly concerned with, the journalists provide news, history, interviews, crosswords, etc ,but there is also balance within commentaries - the occasional reference to AI evaluation, occasional AI charts and so on. I try quite often to bring some of these other aspects of go to SL (and books), but in the main it gets ignored. My fault, or is it the make-up of the SL readership, or the western readership in general?

Now if we go back to the original title of this thread (Is this opening really so bad?), I see that as a topic that could easily appear in a Japanese magazine. But when I envision how that question would be answered there, I see something that almost belongs to a different universe from SL. On SL, we get rival AI evaluations, decimal points and general disregard for professional expertise. I try to red flag that, but I'm the one who gets red flagged.

But I don't really mind. I've got my Japanese material, so instead of gruel for every meal, I still have what I regard a well-balanced diet.

Does that help you understand how I see things? For me, it's not really an argument between AI and pros. It's about the SL treatment versus the more balanced treatment.

John Fairbairn wrote:He does not explain his counts, beyond using the criteria of this table.

One could say an AI doesn't explain either, at least without effort to experiment with the position. But in truth, Ishida actually does try to explain how he arrives at his counts, with explanations along the lines of "This territory is worth 15 points, but the two stones by themselves would have been worth about 10, so Black 1 here has only gained 5 points. I assess the loss of strengthening White on the outside as 15 points."

I imagine the focus on one side of the board is mostly a question of limited space in print. They are opening/early middlegame problems so the reader seems to be invited to imagine an otherwise mostly empty board. Seems like a reach to call this an important distinction.

Furthermore, Ishida claims moves that lose ten points are few and far between (which suggests to me he may have been at the cider ), and his aim is not to teach an understanding of counting, but to make the amateur realise that his main fault is not making bad moves per se, but making an accumulation of minor bad moves and bad moves. This is valuable information that AI is not able to verbalise.

It's easily seen when playing the AI and watching your estimated score go down over time, with all kinds of step sizes.

But again, my point was: why overcomplicate this and not just say "it's a strong player's opinion about the position, for whatever that's worth", in either case.

John Fairbairn wrote:If I look a little more deeply, I see SL as a group of amateurs who are not themselves improving as a result of AI and who are not helping to grow the game.

Growing the game would be nice but it is out of scope for a thread where someone simply asks whether an opening is bad. (My own attempts at growing the game include writing a piece of software that includes beginner tutorials, and, pre-COVID, going to my city's annual Japan Festival and teaching Go to beginners). And I wouldn't be able to tell who here is improving or not, and for what reason, so I think it's not worthwhile to engage in such speculation.

I see the Japanese press as well supported by professionals (some of whom actually are improving with AI!) who are not really preaching but explaining, and who above all are successfully growing the game.

Are they? For a long time the general consensus seemed to be that Go in Japan was in decline, and if anything it was Hikaru no Go that made a dent in it. But what you are overlooking is that people are using the tools and resources available to them. Japanese press or Chinese literature is not readable to the vast majority of people here. AI evaluations and game reviews are available easily and essentially for free, unlike pro teaching.
The one exception is Michael Redmond, who makes video content for USGO and his own channel - the man is a treasure and as far as I can tell he seems to be well loved by the community. But he also references AI evaluations so maybe you wouldn't be a fan.

Now if we go back to the original title of this thread (Is this opening really so bad?), I see that as a topic that could easily appear in a Japanese magazine. But when I envision how that question would be answered there, I see something that almost belongs to a different universe from SL. On SL, we get rival AI evaluations, decimal points and general disregard for professional expertise.

I don't think anyone disregards professionals - it's a matter of accessibility. But most people here do not worship them either, that is true. (Ishida, by the way, also has decimal points, when he says a move loses "perhaps a half or two thirds of a point").

But what you are overlooking is that people are using the tools and resources available to them. Japanese press or Chinese literature is not readable to the vast majority of people here.

This really belongs in a different thread, but a HUGE amount of JCK material is available in English, in books and Go World. It is still mostly available, but western go players have chosen to put western go publishers out of business.

AI evaluations and game reviews are available easily and essentially for free, unlike pro teaching.

Yes, but you get what you pay for. Water for most of us is essentially free, but most people prefer champagne, beer, coffee, tea, etc. (There are even people who like Prosecco!!!!!!) There is water in them thar drinks, of course, and I don't object to AI being used similarly for commentaries as an ingredient. But I prefer to pay for extra taste.

This is in line with my perception that the reason we have a plethora of go apps in the west despite the smaller western go population, to the point of where the cutting edge of go AI started in the west and even today Asian pros use western-made go AI, is that the go population in the west is dominated by computer types.

But I adopt the Carlson philosophy that large size is not inherently good. Growth is good only if it's healthy, and paves the way for future growth. However, if people in the west have the perception that only computer, then a large go population of computer types is not as good as a population half the size with a more diverse mix of people, since in the former, there is an absolute limit to the potential of possible people who could 'fit in'. Computer nerds. I find it ironic go players in the west seem to only think taking short-term losses for long-term gains applies only over the go board. I say this with the greatest admiration for them, as I think my brain is allergic to the concept of computer programming, so see computer programmer and espicially teen hackers as divine beings far beyond what I could possibly reach, more impressive that the teen Asian go pros. But still.

What I don't agree with John Fairbairn--well, perhaps what I agree with him more than he does himself--is waiting until AlphaGo to raise this observation more diligently. It's unfair to suddenly expect 'numbers guys' to suddenly change and be who they're not. The solution to the overdependence on 'numbers thinking' is not downing there valuable contribution but rather balancing the 'numbers people' with 'poetry people', 'philosophy people', 'new agey guru people', people who won't be too hung up on whether AI thinks an opening is +0.2 points or not, people.

I only have about 1 year 6 months left so I'm going to start being more candid about my history.

When I had just learned go at around 12 years of age, I use to pay a lot more attention to women's, go for two embarrassing reasons. One is being able to identify as a demographic that is traditionally not as strong at go as a go player in Europe, but more majorly is that children identify their future selves with men and their current selves with women, for obvious biological reasons.

In this case, there should be a concerted effort to promote go among people in artistic or scientific fields that are underrepresented in. The people who claim to be awake all the time and never sleep—they think they're Nasir Jones—focus on genetic diversity, but I'm more into diversity of thought in this particular case, despite ethnology being my main interest. However fixing the go delusion is the west is more important to me now in the time I have left.

I'm currently less intelligent than I was when I was a child, but apart from that haven't really changed much. But a few perceptions have changed. I think that that majority of activity in the west to improve western go is a complete waste of time and is even discriminatory. I would spend all resources on only:

1-People who are neuro non-atypical or with special needs
2-Women and youth go, they can't realistically be separated
3-Apsects of go that don't require you to be a 'go player' to appreciate. For one I hate the wrong and ridiculous assumption that when a beginner learns go, the first thing you should do is try to make them stronger, since being stronger is how you can appreciate the game. These same people will then complain about when they think online go players had misbehaved in some way!?!? That's what what happens when your go culture is dumb enough to obsess over 'getting stronger' to the point where it's assumed for everyone. Or maybe it's a consequence of computer types disregarding poetic and artistic elements of go.

Now I've taken things off topic enough, if we want to evaluate what the computer program actually means when it says +5.7, it could mean 'There are a huge range of possible endings with possible scores including were both white and black wins by huge margins, but if you average it out, the difference if +5.7 for white,' or it could mean 'I'm fairly confident that there are only a few good lines in which white wins by 5 or 6 points', but the latter is normally only during the endgame. By definition of how AI trains, it is the former. This is also probably what Cho chikun means by being able to tell a game is within half a point even though the middlegame has barely ended, the Japanese pros expect us to make moderate mental effort and use common sense if we claim to want to get stronger in a mindsport. This is What Ishida Yoshia means when he gives points to opening patterns, blahdy blahdy blah . . .

I for one do think it would probably be far more valuable to learn how Chinese and Korean pros talk and think about go rather than Japanese pros, for obvious reasons. However we must also diffrentiate and discern how pros talk about go between themselves and how they present their thoughts with a general audience in mind--we shouldn't assume it's the same thing