(12/30/2022)

Douglas Hofstadter's Gödel, Escher, Bach calls itself "a metaphorical fugue on minds and machines". The preface, to make a point about how difficult the book is to categorize, says bookstores have put it under math, science, philosophy, religion, and even the occult (I found it in a Cognitive Science section). One of the reviews on the back describes it as "a philosophy book disguised as a book of entertainment disguised as a book of instruction". I like this description, but I think it misses some of the essential flavor. I would revise it as "a book of instruction, hiding a book of entertainment, hiding a philosophy book, hiding an instruction book, hiding an entertainment book, hiding a..."

In formal education, media on STEM-type academic subjects tend to fall into two strict categories. There's "pop science", as performed by Neil deGrasse Tyson and Hank Green and etc, which presents a kind of grab-bag of random Cool Nerd Fact's for pure entertainment. Teachers put these on to keep students from falling asleep. And then there's "textbooks", which are very boring, but pack as much subject-relevant information as possible into the space of one book. Teachers give these to students so they can be memorized before test days, at which point everything in them is immediately forgotten.

GEB sits at an odd kind of halfway-point between the two extremes. It's WRITTEN like a popular science book. The tone is light, and there are many interesting-but-not-directly-relevant tangents in each chapter. But the CONTENT follows the pace of a textbook: very dense and sometimes hard to follow. GEB even has problems, the way a math class has problems, and it hides its answer keys to make it harder to cheat. The first one - the "MU puzzle", which for some reason has its own Wikipedia page - took me 40 minutes. But I skipped most of the puzzles in the second half of the book, though, after I felt like I'd gotten the point (or maybe I just got lazy).

Ok, fine, but a textbook on what? Rather than try to capture the thesis of GEB in one snappy sentence - which I don't think I can do - I'd say there are three or four of them, any of which you could call 'the point'. GEB is about, at times: how complicated systems can arise from simple parts; the consequences of something becoming "self-aware"; an argument against the existence of souls; and an elaborate series of analogies connecting the previous three. That one about souls, incidentally, is where the pseudo-religious aspects come in. We'll come back to that later.

Flip to a random page, and the most likely subject you'll see is "Gödel's Incompleteness Theorem", a proof of how any complex formal system cannot be both "complete" and "consistent" (with corresponding nitpicky definitions). The first half of the book is spent explaining and proving this in elaborate detail, in the hopes of getting people without math backgrounds (like me!) to understand it. The book succeeded at this for me; I'm now successfully schoolfed on the subject of Incompleteness. This does mean that the title is a little bit misleading. It'd be more accurate to write it as "Gödel, (escher, bach)"; Bach is used for reasons of analogy and Escher does the illustrations, but neither of them take center stage the way big G does.

Formal logic and classical music aren't the only pieces of the analogy, either. Computer science, neurology, Zen Buddhism, and DNA all take turns on the analogical carousel. He could have easily called the book "Gödel, Turing, Crick", or maybe "Gödel, Mumon, Zeno", if he wanted to. There's a sense, though, that artificial intelligence is the "true" main subject of the book. It's only brought in for the last four chapters, but that's when it all really started to 'click' for me. Many of the digressions, seemingly meaningless at the time, make sense in light of the question: how can you teach a computer to think like a person? How do you teach a computer to become self-aware, make analogies, or become capable of thinking 'outside itself'?

---

Another way to frame this is, "how can you teach a computer to understand GEB?", and this is where the book gets bizarre. Since this is a book about consciousnesses, and you're one of them, it tries to make you poke at your own thinking as you're reading it. The concept he uses to bring this across is something referred to in computer science as an "isomorphism", or in regular English as a "metaphor". The two words are synonyms; to put it another way, "metaphor" is isomorphic to "isomorphism", and "isomorphism" is a metaphor for "metaphor". That sentence you just read is isomorphic to a lot of sentences in GEB - by which I mean, if you found reading it annoying, GEB probably isn't the book for you.

GEB's structure works like this: after each chapter, there is a short fictional dialogue that sets up the concepts of the next one. The two main characters of the Dialogues are Achilles and the Tortoise, like in that one paradox about how no one can actually move. Achilles is the straight-man, and the Tortoise runs verbal circles around him with clever tricks and wordplay. There are a few other recurring characters (a Crab, a Sloth, GOD, etc.), and they collectively have wacky GEB-themed hijinks. These are usually a relief after the difficult abstract discussions of the main text. I really like them; part of me wants to reread GEB and skip all the heavy stuff, just burning through all the Dialogues back-to-back, though another part of me feels like this would be like pouring out all the cereal to just eat the marshmallows.

There is a dark side to the Dialogues, though. In the light, playful land of Achilles' nonsense world, there lies an opening for a great and terrible evil to sink in its gnarled teeth. That's right: PUNS. The Dialogues are a nonstop barrage of pun after pun after pun, and most of them are really, really bad. To be honest, I actually like some of the more structural wordplay - many of them are made to fit some kind of theme, such as the "Crab Canon". The form of the Crab Canon is a kind of zoomed-out palindrome; inside, Achilles and the Tortoise find a (real!) song called "Crab Canon", which is a musical palindrome, and a print of an Escher painting made up of literal crabs, which is also a VISUAL palindrome. I thought this was neat, and the ridiculous sentence-knots he ties to make it work are entertaining.

Because there are so many structural details tied together in the Dialogues, the actual spoken words get very stilted. Achilles and the Tortoise talk in this exaggerated, verbose, overly-polite faux-British, which gives them leeway to say weird things that suit the wordplay. It becomes its own kind of puzzle: why is that sentence phrased in such a strange way? What's it hiding? How does this connect to the rest of the book? At its best, these are excellent brain candy. Sometimes, though, I feel like he's forcing it. Hofstadter knows he's writing a book against people having souls, and he doesn't want to look like the kind of stuffy person who's just too boring to understand the magic of the human condition. The Dialogues occasionally strike me as overcompensating for this. "Look how fun I am!" they seem to scream. "Just because I don't have a soul, doesn't mean I can't have a heart!" Yeah, dude, I get it. You're very clever.

I don't want to give you the impression I think the Dialogues are just a bit of sugar to help the medicine go down. By tying your brain in knots, GEB prompts a kind of self-reflection that's almost like meditation. Its questions about how self-awareness is created implicitly ask you: how are you thinking? How are you thinking about these words right now? How EXACTLY? Working through the puzzles is as much about watching yourself think as it is about actually solving them. Similarly, the experience of READING the Dialogues is as important to the book as their actual contents. Here's where you might think the Zen analogy fits in, though this connection is one of GEB's weaknesses, in my opinion. Hofstadter admits in his introduction that he thinks Zen is dumb and doesn't take it very seriously, and the analogy is correspondingly flimsy. I think he included a bunch of Zen Koans 'cause he thought they were funny, then just lets them sit there without seriously trying to fit them in with everything else.

---

So, what about GEB as a kind of "artificial intelligence Bible"? As a religious text, GEB is... unsurprising. In a way, this speaks well of it. The foundational ideas of AI expressed here, 45 years ago, were successful enough that my reaction to a number of them was "well, duh". It's also a little disappointing, though. If you can already imagine why a person like Douglas Hofstadter thinks souls don't exist, actually reading the book isn't going to give you any earth-shatteringly original arguments to rock your worldview. On the other hand, that could be because it WAS earth-shatteringly original, but has since been absorbed into general discourse through cultural osmosis.

As for GEB's take on the meaning of life... it kind of skips this step? It seems like a pretty important piece of any religious text to me, but Hofstadter pretty much takes it for granted. There is a chapter early on dedicated to "meaning", but more in a definitional sense than a purposeful one. GEB seems to say: look, here are all these wild and complicated things like six-part fugues, and Zen koans, and typographical number theory. Obviously these things are meaningful and should exist. Anyone who pretends to disagree isn't worth taking seriously. A thinking machine would be another one of these things, so of course it should exist if possible. (This might sound negative, but I don't see it as a flaw. GEB is satisfying as a collection of ideas, and I'm fine with meaning being left as an exercise for the reader. If it had told me explicitly what I was "supposed" to think the point of it all was, I would have found that annoying.)

By the way, none of this is at all relevant to superintelligences, or singularity, or being turned into paperclips. Hofstadter mentions offhand that he thinks 'superintelligence' isn't real, and the subject never comes up again. 1979 is before the time when people thought about paperclip stuff, I think. And anyway, Hofstadter isn't really thinking on the wavelength of productivity. My impression of him - and, if this is what he thinks, I agree - is that he's not really interested in having computers supplant humans in the workforce. He wants computers to be alive so they can appreciate the joy of living, and because it would be cool and weird, and they could expand our understanding of the world and maybe be our friends.

I wasn't alive in 1979, but I suspect this way of thinking reflects a different time. It's like with psychedelics and "microdosing"; something that was cool and weird, which could be appreciated for its own sake, but was then sublimated into the grand designs of Work and Productivity, to be discarded if its promises do not prove Economically Useful. This makes me sad, but there's nothing I can do about it. I wonder what he thinks of current developments in AI - I know he doesn't think they're conscious yet, but I haven't been able to find anything more detailed than that. Does he think they're a dead end? What does he think they're missing? When he talks to ChatGPT, what does he ask it?

Do I recommend GEB? Uhhh, maybe. I can't recommend it lightly, that's for sure. This is a long, hard, sometimes very boring book. If you do choose to read it, I recommend skipping Chapters 11 and 16. These cover neuroscience and typogenetics, respectively, and I found the two of them collectively so dull I almost gave up on the whole thing. They're not really necessary, either; you can guess what he's trying to get at in them with a quick skim-read. The flipside is that I've spent the entire last couple weeks thinking in GEB-mode, and I can't deny it's been pretty fun. You start wondering how everyday things could fit into the great big analogy soup, or how they might be understood by an artificial intelligence. AIs aren't going anywhere, and if you want to understand them on a very basic, primordial level, this is the place to look.

7.76/10

(1/2/2023)

So, I listened to it 12 times, once for each month of 2022. I'd heard enough people sounded good, actually, and you just needed to get used to its style, that I needed to see if it was true for myself. I'll be damned if I'll let something cool and weird pass me by just 'cause it takes a little patience. On the other hand, if Captain Beefheart is really just a dumbass and a bunch of people tricked themselves into liking him for status reasons, that seems like important information too. So, does Trout Mask Replica ACTUALLY have anything real to it? One year later, my final answer is "mostly".

I don't know much music theory vocab, so it's going to be difficult for me to describe exactly what happened in my ears. Probably there are better terms for all the concepts I'm about to fumblingly attempt to present. I'm sorry in advance if this ends up being slightly incoherent. Nevertheless, I can think of two things that I believe contribute to the Trout Mask Replica experience:

One: an effect where different instruments, doing very different things, "split apart" so you hear both of them at once. This might be a result of what is referred to as "polyrhythms", but I'm not sure. The instruments in TMR frequently fly in wildly different directions; the guitar is playing one riff while the bass is popping off its own little solo, and there's a saxophone wailing in the corner, all at the same time as Beefheart's crazed ramblings. On a first listen, most of these effects would melt together into one unfathomable cacophany. But on later listens, I was able to "separate them out", and hear each instrument without contamination from the others. A (slightly inaccurate) visualization: imagine hearing one melody in one ear, and another melody in the other ear, but being able to 'hear' both of them clearly. This isn't literally what happened - I'm pretty sure the whole thing was coming through both ears - but it was like that on the mind-level. I was "multitasking", handling both at once. (Told you it'd be incoherent.)

Two: the drums. Percussion is enormously important to me when listening to music. Probably 40% of whether I like or don't like any given song comes down to whether or not I like whatever the drums are doing. Plus, I remember hearing once that they're supposed to be how the mind keeps time in a song. Pretty much any normal song holds your hand in the time-keeping process; simple, repetitive patterns, with big, obvious sounds at clean intervals. TMR is not like this. The drums go WILD here; not only are the patterns complex and sometimes hard to follow, they switch up practically every bar. The speed of the songs keeps changing, too, so even if you "catch on" to the pace of the music for a little bit, it'll be gone before you can catch your breath. It's not that there's NO head-bangable tempo to the songs, it's just a lot harder to find it.

So, that's two things that made listening to TMR unbearable in January, but fun in December. Am I sure they're "real"? Another way to put this: it's not unreasonable to expect that you could enjoy literally anything, if you actively try to. Did Captain Beefheart build these responses into his music "on purpose", or is it just the random flailings of a brain trying to find something to like in meaningless noise?

My guess is that #2, at least, is "real". I think there's a sense where finding tempo is a skill the same way finding pitch is a skill. Some people can have "perfect pitch", and just know the pitch of any sound instantly; below that, though, there's a spectrum of ability to find pitch, and you can improve at it with effort (this is what I've been told, anyway; I can't find pitch at all). I don't know if there's such a thing as "perfect tempo" in this analogy, but I do think my ability to catch the pace of a piece of music can be better or worse, and it's now better than it was. I bet I can now more easily find the tempo of other songs, too, after the Year of Beefheart.

For #1, I'm less sure. The thing is, this "multitasking" effect doesn't JUST happen with Trout Mask Replica. I've noticed it happen with ANY piece of music after listening to it enough times. For example, I can't listen to any big song off the Undertale soundtrack anymore without hearing the disparate pieces 'seperately'. And (1) I don't think Toby Fox was actively trying to be some kind of weird avant-garde genius, (2) those songs still sounded good BEFORE that happened. Anyway, isn't the beauty of music supposed to be in its power to make a coherent whole? Isn't it missing the point to make an album that you need to mentally break into its separate compenents in order to enjoy?

Honestly, I don't think it matters. Suppose Beefheart's genius is totally fake, and the real power was inside me all along. Who cares? It doesn't make me stop enjoying it. (Well, it's easy for me to say this from my cozy position here in December. Maybe January-uugr would have some different thoughts about that.) To put it another way: maybe I did trick my brain into liking Trout Mask Replica. But that doesn't imply it was valueless. It's not true, for example, that anything would sound good if you listened to it once a month for a year. There's lots of music that I enjoy on a first listen, but get sick of when I listen to it too many times. My experience is that lots of songs trend downward over time. TMR trends up; it gets better the more time you invest in it. Even if this is totally accidental, it's a pretty happy accident. Sometimes it's fun to play weird tricks on your brain.

Fun, but not super profound. Even if I grant that the weird brain stuff is at least a little bit intentional, I don't think that singlehandedly makes Trout Mask Replica a masterpiece. I love gimmicky novelties, but that would be a bridge too far even for me. It's a neat effect - definitely not like anything else I've heard before - but it's not necessarily BETTER than listening to normal-person music. It's maybe a little bit too psychedelic; it reminds me uncomfortably of trying to watch a movie on acid. You can spend hours observing the subtle beauty of a single leaf, but when you try to move to something intrinsically complex like a movie, with a plot and characters and stuff, it can be exhausting. When TMR ends, my ears feel sleepy, like they've just run a marathon. It's still a cool effect, but it's a lot to take in for an hour and twenty minutes. I might have enjoyed it more if I'd LARPed like I had a real vinyl record, and listened to the A- and B-sides separately.

I should move away from the weird brain stuff for at least a little bit to talk about how Trout Mask Replica is also, like, an album. It's kind of oddly-arranged. It seems like the more bizarre songs are concentrated near the beginning, while a lot of the simpler ones are near the end. At the extremes, "Frownland" - the opener - is TERRIBLE. For the first few listens, I thought this was because I needed to "acclimate" on each listen, so the first track would sound bad no matter what it was. But I tried listening back to "Frownland" at the end of one of my listens, and it still sounded like total shit. Now I wonder if Captain Beefheart intended it as some kind of ridiculous entry fee, like an "ABANDON ALL HOPE" sign for non-weirdos. Either way, I wish it was taken out. For the other extreme, the last track ("Veteran's Day Poppy") is one of his best, so at least the album ends on a high note.

28 tracks is pretty big, but Trout Mask Replica is varied enough that I don't think it's a problem. There are vocal interludes where Beefheart's absurd word salad is given free reign, and bizarro tracks like "The Blimp" make for entertaining interludes. "China Pig" is a cool bluesy piece, a thousand miles from all the others, and another of my favorites. Beefheart's vocals themselves are funny in a way that doesn't get old after repeated listens. He covers the full spectrum of crazy old hobos: in "Moonlight On Vermont", you can almost hear the bottle of whiskey in his hand as he shouts at you from a street corner. In "Well" he sounds beaten down and wizened by hard years on the street; in "Old Fart at Play" he almost sounds hobo-intellectual; in "Pena" he's utterly psychotic. No two Beefheart hobos are quite alike. The lyrics themselves are suitably nonsensical - I still smile at "NEON MEATE DREAM OF A OCTAFISH".

I won't go so far as to agree with Rolling Stone or whoever when they say this album is an unequivocal masterpiece that will never be matched by the likes of normal-person-music. I saw it as a funny novelty album in 2018, and I still basically see it that way. All that changed is that it's now a funny novelty album I can actually enjoy listening to. I don't think the thing I'm enjoying is really just a series of random non-harmonious sounds with no serious talent behind them, but even if it is, I don't mind. It gave me several dopamines, and it will likely give me several more now that some of the tracks I like are in my regular rotation. GEB is still better, though.

7.74/10