dfan

Ha, I didn't realize that I was replying to the person who wrote the post!

The audience I had in mind when writing it was people who were already quite experienced in playing interactive fiction and could then be challenged in a new way while bringing their old skills to bear. So it's sort of a second-level game in that respect (so is 9:05, in different ways, as someone else mentioned).

As the author of For A Change, I am astonished that anyone would think it was a good testbed for an LLM text adventure solver. It's fun that they tried, though.

The paragraph on Go AI looked accurate to me. Go AI research spent decades trying to incorporate human-written rules about tactics and strategy. None of that is used any more, although human knowledge is leveraged a bit in the strongest programs when choosing useful features to feed into the neural nets. (Strong) Go AIs are not trained on human games anymore. Indeed they don't search the entire sample space when they perform MCTS, but I don't see Sutton claiming that they do.

Even when Google Translate got pretty good I was not really able to effectively translate Chinese or Japanese text about Go (the game). I had similar issues to the ones mentioned in this post. Many Chinese and Japanese words (e.g., "ko") have a very specific meaning in the context of Go, but they also have regular meanings (e.g., "robbery") in more normal contexts, so Google Translate would translate text in a generic way, which made everything unintelligible. With modern LLMs, I can now preface my translation requests with instructions such as "I am going to ask you to translate some Chinese text accompanying weiqi diagrams. Your translations should be idiomatic and not shy away from Go jargon. For example, 拆 = extension, 夹 = pincer, 刺 and 觑 = peep.", and it does a fantastic job, enough for me to basically read anything I want. It was lucky for me that evidently enough Go material already existed in the training set that I didn't have to do anything more special.

(Some chess corrections, in case the author is reading: the moves at the start of chess games are called openings in English, not openers; there are not distinct white-piece and black-piece openings, although of course an individual player will probably study a given opening from the point of view of one side or the other; their study is considered fundamental all the way up to the highest level, in fact more so as you increase in skill; and the Sicilian variation in question is the Najdorf, not Najdork.)

I agree with all of the above, including the hypothesis that the OP likely had absolute pitch to some degree as a child and not thought about it. Everyone I personally know with perfect pitch (including myself) started engaging with music seriously very early on in life. (I don't know quickly I acquired it because it's always been as easy as identifying colors and I didn't know it was unusual until my piano teacher noticed.)

For me piano is definitely the easiest instrument to identify, I'm sure largely because it's what I've played all my life. Pipe organs are the worst. I assume that in general the purity of the tone correlates negatively with ease of identification.

The Letterman bit must have been parodying Mummenschanz, an avant-garde Swiss theater group that was pretty popular (as those things go) at the time. They do feel a bit in the same corner of the arts universe as the Residents...

The thing that finally made me very confident that I had aphantasia (back in 1998, before it was A Thing) is that I realized that my ability to "hallucinate" sounds is excellent. I can re-hear songs in my head, I can compose music and hear it as I think about it, I can hear my friends and family talking with their particular cadences and accents. I can't do anything remotely like that with visual images. Before I had that realization, I thought it was pretty possible that I was "just describing the same experience differently".

Same here! I think whatever compensatory mechanisms I've come up with turn out to be real advantages in some ways.

One interesting thing that I've found is that my approach to physics and math problems is often extremely geometric. Even if I don't visually look at things, I'm constantly constructing objects in my head (e.g., graphs of functions) and playing with them, although it's in more of a tactile way. I'll immediately start thinking "what does this function look like?" when my peers are more likely to start by pushing symbols around.

As an addendum, when I calculate variations in chess or Go I sometimes close my eyes because my "board database access" can be easier to operate when everything is purely in my head, as opposed to performing mental diffs on the physical board in front of me, which requires me to keep track of both real and virtual pieces.

IM David Pruess has aphantasia and can play multiple simultaneous blindfold games.

Pretty much everyone at my level (2000 USCF) can play blindfold. I always assumed that I was completely unable to because of my aphantasia, but when I heard about Pruess's story, I decided to work on it, and I now can, although with difficulty and very slowly.

Basically I still keep around all the information about where all the pieces are; it's just not on a virtual board that I "look" at, it's stored more abstractly. I keep track of clusters of pieces and relations between them. The fact that I have an excellent sense of the board itself (I know how all the squares relate instinctively) helps. But I still have to stop all the time and confirm where all the pieces are (or, conversely, what's on every square).

One issue with this is that it encourages collusion. If you're a top GM playing someone of equal skill, it's +EV to agree to flip a coin beforehand to determine who will win (and then play a fake game) rather than playing it for real.

Some chess tournaments have experimented with giving 1/3 point for draws instead of 1/2 and it didn't really change much. Mostly it acted as a tiebreaker, which you could have done by just using "most wins" as a tiebreaker anyway.

My favorite idea (not mine) for creating decisive results in chess is that when a draw is agreed, you switch sides and start a new game, but don't reset the clocks.

It kind of predicted LLMs too! According to the framing story, the text of the novel was supposedly created by feeding in lots of source material and then having the computer WESCAC generate a plausible first-person account of the protagonist's life.

"Weakly solving" a game is a technical term. If you have weakly solved a game, you can play perfectly (achieve the optimal result) when the game starts from its initial position. If you have strongly solved it, you can play perfectly starting from any position.