Great idea for a post. I’ve really enjoyed reading the comments and discussion they generated.
GreenRoot
Karma: 408
Does it make sense to speak of probabilities only when you have numerous enough trials?
No, probability theory also has non-frequency applications.
Can we speak of probabilities for singular, non-repeating events?
Yes. This is the core of a Bayesian approach to decision making. The usual interpretation is that the probabilities reflect your state of knowledge about events rather than frequencies of actual event outcomes. Try starting with the LW wiki article on Baesian probability and the blog posts linked therefrom.
It sort of seems like a critique of the terminology if dark arts tend not to be deployed by the dark side.
I agree. I think the dark side terminology is based on the “dark side of the force” from Star Wars, which has connotations of a personal fall into temptation, and the dark arts refers to magic of evil intent or effect, perhaps from Harry Potter, where it is used by evil but not self-deceiving villains. This could explain the inconsistency.
I think you have really helped to clarify the go side of this analogy, and I’m grateful for your description of sabaki play and what makes it different from trick moves. I think the connection you draw to rationality and debate are pretty good.
I’m not sure about this, but I think there’s another sense in which the term “dark arts” is used on LessWrong: using one’s knowledge of common cognitive biases and other rationality mistakes to get people to do or believe something. That is, fooling others, not fooling yourself. For the go analogy, I think this is most closely related to trick (non-obviously suboptimal) moves. Or perhaps the technically unsound but necessary aggressive moves used by white in handicap games to which black often responds with too much humility.
Small advantages escalate
Actually, one thing I enjoy about go is that small advantages don’t escalate, at least not nearly as much as they do in chess. In go, if you make a mistake early that puts you behind by, say, 30-40 points, the place where you made that mistake usually interacts with the rest of the board little enough that you’re not hugely disadvantaged elsewhere, and if you play better in the time and space that is left, you can catch up. But as you say about chess, I’m not sure if this is a very generalizable idea, at least when it comes to rationality.
To become good at poker it’s crucial to be able to distinguish between bad luck and play mistakes. You have to keep your cool when your opponent makes bad moves and wins anyway....In life, we are very often faced with situations where we have to analyze to what extent something is the result of our own actions and to what extent it is the result of factors outside our control.
I think this sounds like a valuable lesson to learn, and as you say, the kind of thing you couldn’t get from a deterministic game. And as with go, I suspect that some lessons from poker sink in better when you experience them in play than when you just read them. I would be interested to read more about it, if you (or any other poker players out there) have the time and interest to write a post on rationality in poker or other games with a chance component. I have a feeling that there are lessons related to probability and quantifying your beliefs that could be drawn, or perhaps stories from games that can be used as illustrations of probabilistic or Bayesian reasoning.
There’s an interesting essay by William Pinckard that contrasts the philosophical perspectives of the gameplay of three ancient games; backgammon, chess, and go, which says in summary: backgammon is man-vs-fate, chess is man-vs-man, and go is man-vs-self.
What’s your playing strength in Go? The article reads a bit like it’s either too much targeted at people without understanding of go or is written without by someone with a playing strength >5 kyu?
I’m about 12k on KGS. I definitely aimed the article at people who knew nothing about go, but I think it’s also interesting that you could tell that I’m not a very strong player myself. I would be interested to know if you have found generalizable lessons which only came after you achieved a deeper understanding of the game.
But if the two players disagree, the solution is simply to resume play. That true for some Go rule sets but it isn’t true for Japanese style rules.
According the wikipedia article on rule sets’ treatment of the end, all the sets actually say that you should play things out, capturing dead stones. I guess I’ve only ever played with the more convenient practice of mutual agreement about dead stones. It happens this way in every club and internet server I’ve ever played at, even when using Japanese rules. So in this sense, the actual experience of playing go does reinforce the idea that new evidence is the arbiter of conflicting beliefs.
I think you’re right that most goal-directed activity, especially formalized pursuits like abstract board games, encourages rational thinking. Nevertheless, I have gotten the feeling that go is particularly good in this regard, at least in my experience. I played chess for a long time, and have tried many other types of formal table and online games, and of them all, go seems to have the strongest tendency to show me how bad habits of thinking work against me.
I would love to see more articles like this one explicitly illustrating how other activities can be be approached as a means of rationality practice.
(Perhaps you have had experience gambling in the paper clip casino to increase your hoard, which has given you valuable practice in understanding probability?)
OK, I see, thanks for explaining it. I’d never really heard of this difference before. I’d have to say that the more subtle moves that risk only a little and feel out your opponent sound more akin to the “dark arts” in rationality.
Again, thank you. I’ve made another fix. As you can see, life and death problems are not my strength!
The “dark side” has an analogy in go. It is tempting to play moves that you know don’t work because you think your opponent won’t be able to figure out the correct response. It is usually not obvious to beginners that doing this is really holding them back.
Good point! I thought about including this connection between trick moves and the dark arts, actually. They don’t seem quite parallel to me, but there are definitely similarities. If anybody is interested, you can read more about trick moves here.
it contains approximately 10,000 times the maximum safe dosage of in principle.
Great quote.
One aspect of go which is present on LW but not true about rationality in general (and so not part of the article) is a culture of welcoming and mentoring. Good players are honored by teaching beginners, and the handicap system facilitates interesting teaching games. You should not worry about bothering (go playing) humans with any lack of sophistication. Not all players have this attitude, of course, but surprisingly many do. The place on the internet I’ve found best reflects this welcoming culture of go is the Kiseido Go Server.
Also, I should note that I’ve been advised by strong players on a few occasions not to play against computer opponents much, especially those set to easier difficulty levels, because it can build bad habits.
Thanks for the feedback. You’re right: for players with more than beginning skill, I agree that Fig 3 is alive (and Peter de Blanc is right that Fig 2 is not “unconditionally alive”) in the original versions of the figures. I’ve revised Figures 2 and 3 accordingly. (So the rest of you shouldn’t worry if this comment thread seems confusing! If you’re interested, the original versions are here and here.)
In choosing examples, I was aiming for arrangements that visually conveyed the three states of close surrounding, surrounding with internal structure, and something intermediate. The goal is to be able to talk about “life” and “death” as alternative states the game might be in, like alternative hypotheses of reality, to serve the go/rationality analogy, without having to explain the rules. I hope the revised versions still do this, while making their labels more correct.
This is a nice observation, and I think it’s true about both go and rationality. Wish I’d thought of it for the post!
Yes, this is true, but it’s also true that some kinds of imitation can take you far even if you don’t understand them. Personally, I try to play with good shape, I have seen it pay off, but I don’t understand most of ways that this helps me. A good parallel in rationality might be learning self doubt. This can help, even if one doesn’t know the myriad ways people have of fooling themselves which it is intended to thwart.
Rationality Lessons in the Game of Go
A couple thoughts on places to look for ideas, places where people have probably been thinking about similar challenges:
Interstellar Travel There’s a lot of speculation about feasibility here, and I think people generally assume the need for some sort of long-term, low-power cryogenic preservation. They do assume access to interstellar vacuum, though.
DNA “arks” and similar biodiversity libraries. I haven’t heard of anything in this space looking at zero- or low-maintenance preservation, but maybe there’s a paranoid fringe?
Yes, or else posted very soon. In any case, if the content ends up separate, please link each post to the other.