Cute gimmick, but fails at the main point of a book review. Instead of taking a complicated, long or unreadable piece and making it accessible, it just makes an equally-inaccessible variant of it by keeping it in rhyme.
Haven't read this one yet but wasn't there another review that took this tack? I thought it was very impressive. Interesting two people attempted it. But my memory is poor so if no one else remembers that other one, I'll have to go look at Scott's initial list of lists, which I found painful.
There was also the review of "Alphabetical Diaries." As a formal experiment, I preferred that one to the Don Juan review here, since it engaged in more extensive reflection on the book rather than focusing so much on summary. But I realize my preferences are not universally shared: lots of people here seem to prefer book reviews that are essentially summaries of the book reviewed.
I agree. If this was a review that tried to explain to me what this is and why it is great I think I could've gotten something out of it at least.
But this really seems like preaching to the choir: If you already like English poetry you might be impressed and enjoy it, if you don't then there's just nothing here.
Honestly, can't blame the reviewer. I think trying something like this is a cool idea and if people voted it a finalist it deserves to be here, my own dissatisfaction notwithstanding
I can't ever seem to pull meaning out of poetry. Probably due to lack of erudition but exacerbated by my brain reminding me that I won't understand in place of actually trying to understand (is there a name for this?).
I really enjoyed this book review. So much so that Iight try to read Don Juan for myself. Great job!
Generally I think poetry is a worth while literary genre. Really smart people tend to enjoy it which for me is kind of a heuristic on why I should care (they could also all be full of shit, so I'm still skeptical). If I read enough, and read enough critically, maybe I'll start to get it.
As for the review, it provided enough to make me interested in the story. Why I should care? Because good stories are fun to read.
I actually read Byron's Morte d'Arthur as part of a book club and really enjoyed it.
(I don't want to be too critical here: I appreciate the attempt to do something different and the execution isn't bad. But the bar for long poetry to be readable is very, very high)
Seconded! I found the language to be sufficiently strained that I gave up on reading it. Perhaps the reviewer had an interesting summary of the book. Perhaps the reviewer had some interesting ideas of their own. As Viliam said:
>I got tired the way I don't get tired reading long prose.
and as Shaked Koplewitz said:
>But the bar for long poetry to be readable is very, very high
Replying since there's no upvote button. I was most excited to read this particular review but couldn't finish for the exact reasons above. Makes me feel better I'm not the only one.
It's a novel written in verse. Short as a novel, long as a poem but at least when I read it many years ago, it seemed quite good, and I usually can't make it through long poems (like this one for instance)
From the poems Scott has posted before, I feel like he's a bit better at making the words fit the meter and rhyme scheme without being tortured than this reviewer. I don't think it's impossible it's him, but I do doubt it.
> Blake? That’s been done. Keats? There’s no doubt he’s charming,
> But none too popular around this joint.
Contain link to Unsong and a mid-obscure Sequence. This points to someone deeply embedded in the ratsphere, I have read Unsong and may have read that lesswrong article (or have heard it discussed on the Bayesian Conspiracy), but even if I did know my English poets (which I don't), I don't think that I would have noticed a link from Unsong to Blake or remembered that Eliezer quotes Keats. (I also don't expect that a current LLM would come up with that.)
As you point out, Scott has experimented with poetry under harsh constraints before, for example, his Trump talks for a whole paragraph over the Ukraine without the vowels A, E, I in heroic hexameter. [1] "Write a synopsis of Don Juan in ottava rima" sounds exactly like the challenge he might enjoy.
Starting from the prior that any finalist entry might be written by by Scott with a probability of 5-10%, updating for
* the topic (previous entries by Scott have covered fiction where most contestant entries stick to non-fiction),
* the unconventional format (like re-imagining a scene of Njal’s Saga as a Phoenix Wright episode)
*the deep links to the rational culture
and
* for Scott recently showing an interest in writing poems in esoteric formats,
my gut feeling is that p(Scott) should be 80-90%.
(Where is rootclaims when you need them?)
Most of the remaining probability is on "someone who tries to convince the readers that they are Scott rather than winning the contest by conventional means (and also doing an excellent job of the former, as far as I am concerned)".
The substance doesn't feel quite Scott-y (Alexandrian?). The metre's competent, but a little forced in places, and there's a lot of meandering to fit it. There's a bit too much enjambement, to the point where it clunks. It's missing the bit at the end where he says he has mixed feelings about Byron. In my view, probably not Scott.
Absolutely not for me. I imagine I have a basic grasp on English, but I could hardly get through any of the verses without looking up a word, and most of the references to other literary works, even when linked, were wasted on me.
Matthew Scully, the author of the previous review's subject Dominion, argued that we could understand Wittgenstein's Lion. After this review, I'm not so sure about that.
I'm very surprised you say you had to look up words in most of the verses. there's nothing there extraordinarily recondite.
Maybe I'm just older and from the days of "you *will* hack your way through this dense 19th century prose and you *won't* be reading any 'relevant to the youth' modern nonsense" schooldays, though 😀
As for the review - I can see why people suspect it's Scott. I'm not entirely sure of that, but it certainly could be.
As for the poem by Byron - in my youth I hacked my way through the poem (voluntarily) and came away disappointed. As the review says, it ends abruptly, but up until then, nothing really happens. I know that sounds odd, given the catalogue of adventures that befall Juan, but he's such a blank centre that it really is taking him up and plopping him down in an Exotic Setting so Byron can show off his versifying skills.
And they are real skills, I'm not denying that. But there is the strong suspicion that Juan is an authorial self-insert, and having poor hapless Juan be pursued by all these women through no fault of his own, falling in love and then being whisked away by circumstances so if their lives end up being ruined it's not *his* fault - yeah, protesting too much, George.
Juan never develops as a character so he's very tiresome in the end. You can't even say that he's looking for true love, because things happen to him and he reacts, not acts. Maybe Byron is being cynical in intimating that Juan is fooling himself with the idea that "oh my poor dear [last love], had we only been able to stay together, how happy our lives would have been!" and that in fact the latest romance was always doomed to fail so it was lucky that Juan got whisked away to a new scene and a new love - but that kind of mocking, knowing, deconstruction is hollow in the end. He's not old-fashioned morality enough to condemn Juan, nor modern optimistic 'love is love and love wins' to build him up as a hero. It's Romanticism turned sour, but he still likes the big action set pieces and can describe with bravura the new scenes that Juan ends up in, but ultimately Juan is too wishy-washy to be the hero of anything.
I suppose when it's a literary work, a poem, my confidence to assign each word its intended meaning drops, even when I've read the word before. Soliloquize, staid, "stray dents", philanderer, homily, rake, and that's just the first chapter. Needless to say, it destroys any semblance of reading flow.
Ah, I see what you mean, but I think that's over analysing it - the result of all of us being put through English class in school to hunt meanings, dig out 'between the lines' readings like pigs rooting for truffles, and put through our paces with "if the author says this is a red wheelbarrow, what are they *really* trying to say by this? what does it symbolise?"
Sometimes a wheelbarrow in the rain is just a wet wheelbarrow 😀
By your competence writing comments, I'm surprised you wanted to look those up also — the meaning seems perfectly clear in-context. The only one that confused me was "stray dents", which is one of the lines that convinces me this probably isn't Scott...
I guess we'll see if I am worse at prediction...-ary than... regular dictionary. uh, so to speak.
It's unfortunate that now, when I see verse posted by anyone I don't know, I automatically suspect it to be ChatGPT output. -.- (I'm not making any claim about this particular piece. It's just a knee-jerk reaction I've recognized in myself.)
For whatever it's worth, I've written enough poetry by myself, and prompted enough poetry out of GPT-4o, to say that the latter is not yet at the point of being able to write most of this. You *can* get it to output Ottava Rima, if you're strict about it, but you can't get it to do verses like verse 3, which has joking references to a number of other poets and also a reference to a specific stanza in the original, talking about Latin poets. A few stanzas here and there could plausibly have been GPT-4o, but most of them couldn't have been.
Yeah, one of the tell-tales of tuned models, I find, is they amplify the base models' sampling problems in terms of concreteness and specificity. LLMs can vary a lot in whether they will do something like verse 3. For example, here is another LLM, where I simply prompt it with OP up to 'Blake' to prompt engineer some references, and you can see that it is reasonably specific in the humorous references (BO=8, temp=1, 0.01 frequency+presence penalty), even if the ottava rima rhyming is highly imperfect:
"**...And one to tell the tale, plus two disarming**
**Poor commoners who— well, you get the point.**
**Blake? That’s been done**. Dickens is too alarming;
Thackeray, Elliot, Shelly, all disjoint;
It seems too much to ask you find someone less charming;
It would not do to come too straight to the point—
Besides, it’s called poetic license, dear
To take some shortcuts now and then, from here
\
To there. We need a sort of epic hero
Whose heroic deeds we’ll all admire.
See, Don Quixote’s far too out of vogue, though
He’s epic in the most heroic manner.
Who else? Don Giovanni lacks panache (and ego);
Don King lacks a degree from any manner
Of schools. That leaves just Don Diego, Donner,
And, Don — oh damn, forgot — ah: Juan.
\
II.
\
If you want work, you come to California,
(That’s just ‘Los Altos’, but for poor and swarthy);
If you want cocaine, Miami’s at your fingertips;
If you want love, you’ll come to snowy Fargo;
And if you want some lobster, then it’s Maine;
But if you want damn everything, you go to Spain.
\
“Oh, Barcelona! Show me through your streets —
“Your beaches, cafés, discos, mountain views!
“Oh, Seville! Feed your paella to my soul —
“And your Andalusian horses too!
“Oh, Granada! (Yes, each place gets its own exclaim
I just tried the same idea with GPT-4o, prompting it with the OP up to "Blake? That's been done." as you did, but rigging the dice with specifics (I explained what Ottava Rima is, explained that I'm looking for a list of poets to make fun of, and included Stanza 42 from the original poem for parody context).
This is the best result I got from n=5 samples; the rhyme decays into consonance by the second stanza, and the meter suffers too (the third line would have been way better as "And Keats is all too often all too literal"), but it's actually *much* closer to OP's quality than I was expecting. I still don't think it could come up with the "Pope, Alexander" joke without being explicitly prompted to, and still don't think most of these stanzas could be LLM-written, but...wow.
\
And one to tell the tale, plus two disarming
Poor commoners who— well, you get the point.
Blake? That’s been done. His verses, though they're charming,
Are far too serious for our new joint.
Coleridge on opium is most alarming,
With albatross and nights in icy point;
And Wordsworth’s daffodils all set aside,
He still can't make a hero if he tried.
\
Shelley dreams big but then gets too ethereal,
With clouds and West Winds more than human deeds;
And Keats, though haunting, is too often all too literal,
Mildly surprising. I would've expected GPT-4o to maintain rhyming, given the usual ChatGPT mode collapse onto rhyming, that it is a rhyming format to begin with, and that other poetry tasks I've tried in ChatGPT-4o tend to veer into rhyming the further they go. So I wonder how this manages to error in the *opposite* way?
But yes, you can see that given some scaffolding, LLMs do substantially better than one might expect. See also my recent Claude-3.5-sonnet exercise https://gwern.net/rubiks-cube where the big prompt setting up 'the scaffolding' is where most of the coherence and cleverness comes from.
Re: rhyming, I find that the default mode collapse is rhyming, but rhyming with a simpler ABCB quatrain and with a much looser meter than iambic pentameter; YMMV, of course.
But yes, scaffolding can do amazing things. I'd never read your Rubik's Cube essay, but the details like the footnotes are very clever, and would have led me to assume that this was organic if I wasn't reading closely. Your overall point at the end about how you could have done better writing it by hand, but wouldn't have had time to write an essay as long, is well-taken. I could have come up with a better continuation from "Blake?" than GPT-4o, but judging by past poetry-writing, it'd take me ~10 minutes per stanza, as opposed to (amortized across all attempts) ~20 seconds. My poems would be 'better', but I doubt they'd be 30x 'better', for any meaningful use of the term.
This is really pretty good. I'm just surprised that some lines don't scan to the extent they don't (in particular "It seems too much to ask you find someone less charming").
Regarding the book review itself - I'm very impressed with the Ottava Rima; it feels as natural as Byron's, and if it turns out this is Scott Alexander, I wouldn't be terribly surprised.
But I think it's appropriate to repost a comment I made here a few years ago, about Don Juan.
In Book 4 of his Odes, Horace wrote:
Vixere fortes ante Agamemnona multi; sed omnes inlacrimabiles urgentur ignotique longa nocte, carent quia vate sacro.
meaning approximately:
Many brave men lived before Agamemnon; but all are unwept and unknown, thrust into an endless night, lacking a sacred bard.
Byron quoted this at the beginning of Don Juan, but (as this review points out in its second stanza) reversed the meaning. Byron himself had the skills to be a sacred bard, and knew it - but where was his Agamemnon? Who could he write about that he wouldn't have some reservations over? Earlier in life, his writings made it clear that he wanted to write an epic sincerely...but between the tragedies in his life that weren't his fault and the ones which were entirely his fault, he'd lost his youthful optimism, and now was stuck with skills he could no longer use the way he'd originally intended. And so, in the first five stanzas of Canto I, he became convinced that *nobody* could write a sincere epic anybody, that these modern times (the early 1800s) were just too progressed for epics to work anymore, or for people to take them seriously.
Instead, he took a classic villain, Don Juan, and made Juan into a satiric and misunderstood hero, so that he would have something and someone to sing about.
A century later, GK Chesterton wrote a completely sincere epic - Lepanto - about a different man named Don Juan (Don John of Austria). I would love to know if this was done as a deliberate reply.
Yes, it feels very natural in imitation, but the voice is, of course, his own. I am very glad to see such writing. It far surpasses the formless tripe passed off as "poetry" nowadays.
The vast majority of poetry these days is written as song lyrics, to the extent that poetry in other contexts should probably be regarded as an extremely niche subgenre.
That's fair - even so, I'd argue that seventy percent sincere is significantly more sincere than Byron's estimated upper bound on modern-day high-quality epic sincerity, based on his anti-Southey introduction.
On the one hand, it's pretty hard to keep track of exactly what's happening or how long it's taking, and how much of this is directly from the story versus how much is flourishing for the sake of the rhythm.
On the other hand, I don't care anymore, this is excellent on its own.
I appreciate the effort, truly I do, but the rhyming verse makes reading this review a chore. I think I would've preferred unstylish yet concise prose.
(That is, great review, but its use of the same form makes it somewhat harder to understand than most reviews. If you can easily understand it, go read Don Juan!)
I realize that some here wish more modern poetry rhymed, but personally I loathe this sort of rhyme scheme and can't bear to read more than a few stanzas of it. I have no opinion of the review since I couldn't get through it, but I can't really vote for it given that.
Much prefer prose that reads like poetry. Give me Virginia Woolf.
OK. I forced myself to read it slowly and aloud as suggested below, and my takeaway is that I never, ever want to read any Lord Byron. So at least I learned that. I can appreciate Robert Lowell or T.S. Eliot, but I can't find the aesthetic appeal to all that rhyming, and the lines aren't particularly evocative. It all seems so corny. But the fault is probably in me, I'm no judge of this type of poesy.
Is Lowell your example of rhyming poetry you *do* like?
If so, I agree.
I also wish more modern poetry rhymed, but not because I like rhyming poetry so much. I just want people to have some reason to work and rework their lines. When it comes to the older stuff, I often find the blank verse more appealing than the rhymed.
I have the Collected Poems of Robert Lowell and sometimes at night open to a random page and read the poem aloud and usually like it, but I'm no Lowell expert. I mostly think of his poems that don't rhyme but I guess some do.
It is weird that I, like most everyone, appreciate good song lyrics, which have to always almost rhyme whereas I can't appreciate much poetry that does. Obviously, poetry came from song.
I think the difference is that when it comes to reading my brain is tuned to prose. I mostly read literary fiction and mostly care about the quality of the prose, and good modern prose is about evocative language, the same as modern, non-rhyming poetry.
I mean, I like some rhyming poetry. I like Shakespeare's rhyming work, although I think a famous Rationalist proved by calculus that Shakespeare was his own father's ghost and not a great writer.
I feel like maybe I've been too negative on this review already, but just maybe, maybe it is bad poetry and that is the problem? Obviously more people like it than don't. I'm certainly not capable of writing rhyming poetry anywhere close to as good as that.
Out of curiosity, have you read Pale Fire by Nabokov? I read it recently, and had the (perhaps unusual) reaction of ultimately liking the poem better than the prose narrative surrounding it.
Although more people seem to like rhyming poetry, it does strike me that more famous *epic poems* are unrhymed than rhymed. It could be that it's harder to sustain good poetic writing over a long space when you're also focusing on trying to rhyme. It could also be that you're not the only one who gets tired by one rhyme after another, even if people tend to enjoy it in smaller doses.
I'm sure most commenters are here after reading, but if not, I would encourage you to read this one out loud. I found it much more tractable, and half the challenge for poem amateurs like myself is finding the right cadence and breaks to make it work. At first, I saw the verse and thought I shouldn't dedicate time to it, but admittedly the review's format succeeded in not just interesting me in the review but the book itself. It's a success in that respect, though I dread this format being anything more than a novelty.
So it's like prose, except you have to do completely unnecessary extra work to understand it? I found this incomprehensible. After a few verses I had a headache. Opposite of what a book review should try to accomplish.
Probably the weakest review so far (of three, so not really saying much). The rhyming scheme is a cute idea, but it doesn't work very well and a lot of the lines don't quite rhyme or scan right making it jarring. And ultimately it delivers little commentary on the work. It's just a summary in verse
I applaud the book selection. Also, this is an heroic effort and "courageous". But no. Byron was a genius and "Don Juan", as another Greer (Germaine) wrote, is the finest comic poem in English. It's unwise (& immodest) to try to review by imitation. For one thing -- as noted above -- ottava rima is terribly hard in English (easier in Italian). When it fails, as here when lines don't scan, it's terrible: the faults protrude. Then Byron's "longeurs" serve the plot, usually by teasing out a local climax (see e.g. the brilliant stanzas 118-134 of Canto I here: https://petergallagher.net/byrons-don-juan-cantos-1-2-annotated.) They are clever, often politically incorrect and, with one or two exceptions, more entertaining than the story of his harmless, feckless hero. Here the longeurs are, alas, just long.
I have just started your annotated edition and appreciate the neat format of it, but am a little confused about this: "A play bill advertising a performance
(about 1832) at the Pavilion Theatre, Stepney, in the East
End of London. There were several operatic versions of
the Don Juan pantomime given in pre-Napoleonic London
that Byron may have attended."
Wouldn't Byron have been a little young to attend the theater in the "pre-Napoleonic" period?
You are absolutely right: I slipped up with the dates here. Byron would have been unlikely to be attending plays in London before (say) 1805 when he started at Trinity College in Cambridge. The "Pre-Napoleonic" period, whenever that is, must be before Napoleon started his European campaigns in 1803. So I should have omitted any reference to Napoleon and simply speculated (there's no proof) that Byron saw one of these performances. Thanks for that and please let me know of other errors you might find. Best, P
I want to suggest that this is not a good thing to do. If we believe in anything here, it's that everything shouldn't be about personalities, gossip, and who did what. These book reviews are about writing great reviews, us as readers gaining some exposure to books we'd never normally read, and judging writing based purely on the quality of the writing. The anonymity thing matters.
It's probably a dumb complaint to say that the reviews here cater too much to a Rationalist audience, but I'm going to stupidly lodge it anyway. The typical reader here probably doesn't want to be lectured about vegetarianism or read a book review in verse. Those who reviewed these books (And I am one) probably aren't as typical as the readers. Full disclosure: I also submitted a review that wasn't chosen to be a finalist, so likely I am not objective about the subject.
I don't think most Rationalists want to read a book review in verse either. But Scott reserved 25% of the finalist slots for nontraditional books, so this is one of them.
Wait, isn't this just self-insert isekai fiction by the famous author of "I am very sad because this Greek couple I just met wouldn't sell me their twelve-year old daughter as a concubine: the poem" and "maybe incest with my blood-related onee-chan was not such a good idea: an apology video"?
Of course Lord B, the man so extra that Victorian England found him too ostentatious, would see himself in Don J. Britain owes us one for ridding them of him... although we're partially responsible for BoJo too, so let's call it even.
The penultimate stanza in section II is one line shorter than the rest. I wonder if that's deliberate, as the author's way of finding out how many commenters read the review out loud.
So now, having read OG Don Juan - wow, what a terrible poem. Part of it amounts to a "slam note" about his wife.
I conclude that the fame of the poem lies with Byron's looks and style; and his poetic fame is owing to other works. So to fairly evaluate this review calls for evaluating a picture of the reviewer.
The verse is impressive. Only a few lapses of meter and form, and the rhymes can get up to three syllables deep. However, since I don't normally deal in verse, I kept getting distracted from the content by the format. I wish there were more analysis, because it seems to boil down to "the character didn't have much agency, but Byron himself did, and that seems better".
I didn't get much of a sense of why the book would be interesting, although I suppose the verse serves to let me have a taste of what it would be like. The content seemed like a lot of erotic stories from an older age, which might be entertaining were I in the mood, but those moods are rare.
Ladies and Gentlemen, I think we have a winner.
Not if my vote has anything to do with it!
I just skipped over the whole thing without reading it, waiting for the content to start.
Me too, I'm afraid. I'm sorry to admit it, because this seems like something *I* might do — but I don't have the patience.
Cute gimmick, but fails at the main point of a book review. Instead of taking a complicated, long or unreadable piece and making it accessible, it just makes an equally-inaccessible variant of it by keeping it in rhyme.
Haven't read this one yet but wasn't there another review that took this tack? I thought it was very impressive. Interesting two people attempted it. But my memory is poor so if no one else remembers that other one, I'll have to go look at Scott's initial list of lists, which I found painful.
The Lady of Shallot, which was a much shorter one, though in my view more charming.
Thank you, yes - that was it, I really liked that one, but then I loved "The Lady of Shalott" so it will get some reflected shine from that as well.
I feel like I should be able to remember things like that, that I've just read a few weeks ago.
There was also the review of "Alphabetical Diaries." As a formal experiment, I preferred that one to the Don Juan review here, since it engaged in more extensive reflection on the book rather than focusing so much on summary. But I realize my preferences are not universally shared: lots of people here seem to prefer book reviews that are essentially summaries of the book reviewed.
I imagine that this reviewer knows "Don Juan" about as well as anyone alive, from the exercise, which is kind of cool.
I agree. If this was a review that tried to explain to me what this is and why it is great I think I could've gotten something out of it at least.
But this really seems like preaching to the choir: If you already like English poetry you might be impressed and enjoy it, if you don't then there's just nothing here.
Honestly, can't blame the reviewer. I think trying something like this is a cool idea and if people voted it a finalist it deserves to be here, my own dissatisfaction notwithstanding
I can't ever seem to pull meaning out of poetry. Probably due to lack of erudition but exacerbated by my brain reminding me that I won't understand in place of actually trying to understand (is there a name for this?).
I really enjoyed this book review. So much so that Iight try to read Don Juan for myself. Great job!
Interesting. If you don't mind me asking, what was your general attitude towards poetry, at least before reading this review?
For me it is disinterest, which is why I wished the review would've expanded on why I should care. But it seems for you it is different?
Generally I think poetry is a worth while literary genre. Really smart people tend to enjoy it which for me is kind of a heuristic on why I should care (they could also all be full of shit, so I'm still skeptical). If I read enough, and read enough critically, maybe I'll start to get it.
As for the review, it provided enough to make me interested in the story. Why I should care? Because good stories are fun to read.
I actually read Byron's Morte d'Arthur as part of a book club and really enjoyed it.
Gotcha, thanks :)
Actually Morte d'Arthur is Tennyson not Byron. My poetic prowess is revealed!
Not just really smart people. Poetry set to music is enjoyed by most people around the world. Eg as pop songs etc.
(I don't want to be too critical here: I appreciate the attempt to do something different and the execution isn't bad. But the bar for long poetry to be readable is very, very high)
Yeah, I enjoyed the first 20% but then... sorry, too long; I got tired the way I don't get tired reading long prose.
Perhaps this would be more enjoyable as audio.
Seconded! I found the language to be sufficiently strained that I gave up on reading it. Perhaps the reviewer had an interesting summary of the book. Perhaps the reviewer had some interesting ideas of their own. As Viliam said:
>I got tired the way I don't get tired reading long prose.
and as Shaked Koplewitz said:
>But the bar for long poetry to be readable is very, very high
In reading "Don Juan" itself, I stumbled right out the gate at this:
And therefore I shall open with a line
(Although it cost me half an hour in spinning)
Narrating somewhat of Don Juan’s father,
And also of his mother, if you’d rather.
Which if not strained, being the original of the mock-style (I guess) seems pretty cheesy and lazy.
Ouch! Agreed. Many Thanks!
Replying since there's no upvote button. I was most excited to read this particular review but couldn't finish for the exact reasons above. Makes me feel better I'm not the only one.
Many Thanks!
Have you read the golden gate by Vikram Seth?
haven't heard of it, is it good?
It's a novel written in verse. Short as a novel, long as a poem but at least when I read it many years ago, it seemed quite good, and I usually can't make it through long poems (like this one for instance)
Yeah, my first thought was that the secret author has got to be either Vikram Seth himself or one of his stans.
Hats off.
Ottava Rima is really hard, and this feels Byronic in tone.
I'm calling this is the mandatory Secret Scott Review, he's been experimenting with poetry lately, IIRC.
I do agree that while it's very very high effort and entertaining, it went a bit too long.
From the poems Scott has posted before, I feel like he's a bit better at making the words fit the meter and rhyme scheme without being tortured than this reviewer. I don't think it's impossible it's him, but I do doubt it.
The lines:
> Blake? That’s been done. Keats? There’s no doubt he’s charming,
> But none too popular around this joint.
Contain link to Unsong and a mid-obscure Sequence. This points to someone deeply embedded in the ratsphere, I have read Unsong and may have read that lesswrong article (or have heard it discussed on the Bayesian Conspiracy), but even if I did know my English poets (which I don't), I don't think that I would have noticed a link from Unsong to Blake or remembered that Eliezer quotes Keats. (I also don't expect that a current LLM would come up with that.)
As you point out, Scott has experimented with poetry under harsh constraints before, for example, his Trump talks for a whole paragraph over the Ukraine without the vowels A, E, I in heroic hexameter. [1] "Write a synopsis of Don Juan in ottava rima" sounds exactly like the challenge he might enjoy.
Starting from the prior that any finalist entry might be written by by Scott with a probability of 5-10%, updating for
* the topic (previous entries by Scott have covered fiction where most contestant entries stick to non-fiction),
* the unconventional format (like re-imagining a scene of Njal’s Saga as a Phoenix Wright episode)
*the deep links to the rational culture
and
* for Scott recently showing an interest in writing poems in esoteric formats,
my gut feeling is that p(Scott) should be 80-90%.
(Where is rootclaims when you need them?)
Most of the remaining probability is on "someone who tries to convince the readers that they are Scott rather than winning the contest by conventional means (and also doing an excellent job of the former, as far as I am concerned)".
[1] https://www.astralcodexten.com/p/hardball-questions-for-the-next-debate
The substance doesn't feel quite Scott-y (Alexandrian?). The metre's competent, but a little forced in places, and there's a lot of meandering to fit it. There's a bit too much enjambement, to the point where it clunks. It's missing the bit at the end where he says he has mixed feelings about Byron. In my view, probably not Scott.
I'll go ahead and bet not-Scott also.
0023e5196f0fc8c27d64f79343e68693d01ddc818bbfc34b7f019764560f3f45
Don’t be wasting GUIDs like this!
Yep. We are now one GUID closer to a collision. We can't afford this wanton waste!
Absolutely not for me. I imagine I have a basic grasp on English, but I could hardly get through any of the verses without looking up a word, and most of the references to other literary works, even when linked, were wasted on me.
Matthew Scully, the author of the previous review's subject Dominion, argued that we could understand Wittgenstein's Lion. After this review, I'm not so sure about that.
I'm very surprised you say you had to look up words in most of the verses. there's nothing there extraordinarily recondite.
Maybe I'm just older and from the days of "you *will* hack your way through this dense 19th century prose and you *won't* be reading any 'relevant to the youth' modern nonsense" schooldays, though 😀
As for the review - I can see why people suspect it's Scott. I'm not entirely sure of that, but it certainly could be.
As for the poem by Byron - in my youth I hacked my way through the poem (voluntarily) and came away disappointed. As the review says, it ends abruptly, but up until then, nothing really happens. I know that sounds odd, given the catalogue of adventures that befall Juan, but he's such a blank centre that it really is taking him up and plopping him down in an Exotic Setting so Byron can show off his versifying skills.
And they are real skills, I'm not denying that. But there is the strong suspicion that Juan is an authorial self-insert, and having poor hapless Juan be pursued by all these women through no fault of his own, falling in love and then being whisked away by circumstances so if their lives end up being ruined it's not *his* fault - yeah, protesting too much, George.
Juan never develops as a character so he's very tiresome in the end. You can't even say that he's looking for true love, because things happen to him and he reacts, not acts. Maybe Byron is being cynical in intimating that Juan is fooling himself with the idea that "oh my poor dear [last love], had we only been able to stay together, how happy our lives would have been!" and that in fact the latest romance was always doomed to fail so it was lucky that Juan got whisked away to a new scene and a new love - but that kind of mocking, knowing, deconstruction is hollow in the end. He's not old-fashioned morality enough to condemn Juan, nor modern optimistic 'love is love and love wins' to build him up as a hero. It's Romanticism turned sour, but he still likes the big action set pieces and can describe with bravura the new scenes that Juan ends up in, but ultimately Juan is too wishy-washy to be the hero of anything.
I suppose when it's a literary work, a poem, my confidence to assign each word its intended meaning drops, even when I've read the word before. Soliloquize, staid, "stray dents", philanderer, homily, rake, and that's just the first chapter. Needless to say, it destroys any semblance of reading flow.
Ah, I see what you mean, but I think that's over analysing it - the result of all of us being put through English class in school to hunt meanings, dig out 'between the lines' readings like pigs rooting for truffles, and put through our paces with "if the author says this is a red wheelbarrow, what are they *really* trying to say by this? what does it symbolise?"
Sometimes a wheelbarrow in the rain is just a wet wheelbarrow 😀
Well said!
By your competence writing comments, I'm surprised you wanted to look those up also — the meaning seems perfectly clear in-context. The only one that confused me was "stray dents", which is one of the lines that convinces me this probably isn't Scott...
I guess we'll see if I am worse at prediction...-ary than... regular dictionary. uh, so to speak.
It's unfortunate that now, when I see verse posted by anyone I don't know, I automatically suspect it to be ChatGPT output. -.- (I'm not making any claim about this particular piece. It's just a knee-jerk reaction I've recognized in myself.)
For whatever it's worth, I've written enough poetry by myself, and prompted enough poetry out of GPT-4o, to say that the latter is not yet at the point of being able to write most of this. You *can* get it to output Ottava Rima, if you're strict about it, but you can't get it to do verses like verse 3, which has joking references to a number of other poets and also a reference to a specific stanza in the original, talking about Latin poets. A few stanzas here and there could plausibly have been GPT-4o, but most of them couldn't have been.
Yeah, one of the tell-tales of tuned models, I find, is they amplify the base models' sampling problems in terms of concreteness and specificity. LLMs can vary a lot in whether they will do something like verse 3. For example, here is another LLM, where I simply prompt it with OP up to 'Blake' to prompt engineer some references, and you can see that it is reasonably specific in the humorous references (BO=8, temp=1, 0.01 frequency+presence penalty), even if the ottava rima rhyming is highly imperfect:
"**...And one to tell the tale, plus two disarming**
**Poor commoners who— well, you get the point.**
**Blake? That’s been done**. Dickens is too alarming;
Thackeray, Elliot, Shelly, all disjoint;
It seems too much to ask you find someone less charming;
It would not do to come too straight to the point—
Besides, it’s called poetic license, dear
To take some shortcuts now and then, from here
\
To there. We need a sort of epic hero
Whose heroic deeds we’ll all admire.
See, Don Quixote’s far too out of vogue, though
He’s epic in the most heroic manner.
Who else? Don Giovanni lacks panache (and ego);
Don King lacks a degree from any manner
Of schools. That leaves just Don Diego, Donner,
And, Don — oh damn, forgot — ah: Juan.
\
II.
\
If you want work, you come to California,
(That’s just ‘Los Altos’, but for poor and swarthy);
If you want cocaine, Miami’s at your fingertips;
If you want love, you’ll come to snowy Fargo;
And if you want some lobster, then it’s Maine;
But if you want damn everything, you go to Spain.
\
“Oh, Barcelona! Show me through your streets —
“Your beaches, cafés, discos, mountain views!
“Oh, Seville! Feed your paella to my soul —
“And your Andalusian horses too!
“Oh, Granada! (Yes, each place gets its own exclaim
“Before its name — don’t make me say it twice — )
“Oh, Granada! How I delight in your refrain
“Of Flamenco guitar and toasted rice!
“Oh, Spain — I’ll always come to you again!”"
I just tried the same idea with GPT-4o, prompting it with the OP up to "Blake? That's been done." as you did, but rigging the dice with specifics (I explained what Ottava Rima is, explained that I'm looking for a list of poets to make fun of, and included Stanza 42 from the original poem for parody context).
This is the best result I got from n=5 samples; the rhyme decays into consonance by the second stanza, and the meter suffers too (the third line would have been way better as "And Keats is all too often all too literal"), but it's actually *much* closer to OP's quality than I was expecting. I still don't think it could come up with the "Pope, Alexander" joke without being explicitly prompted to, and still don't think most of these stanzas could be LLM-written, but...wow.
\
And one to tell the tale, plus two disarming
Poor commoners who— well, you get the point.
Blake? That’s been done. His verses, though they're charming,
Are far too serious for our new joint.
Coleridge on opium is most alarming,
With albatross and nights in icy point;
And Wordsworth’s daffodils all set aside,
He still can't make a hero if he tried.
\
Shelley dreams big but then gets too ethereal,
With clouds and West Winds more than human deeds;
And Keats, though haunting, is too often all too literal,
Chasing nightingales through autumn leaves.
Milton’s epic range is too imperial,
With Heaven’s wars and Godly reigns and creeds.
But Byron’s Juan, in its scandalous streak,
Amuses still; the rest seem less unique.
Mildly surprising. I would've expected GPT-4o to maintain rhyming, given the usual ChatGPT mode collapse onto rhyming, that it is a rhyming format to begin with, and that other poetry tasks I've tried in ChatGPT-4o tend to veer into rhyming the further they go. So I wonder how this manages to error in the *opposite* way?
But yes, you can see that given some scaffolding, LLMs do substantially better than one might expect. See also my recent Claude-3.5-sonnet exercise https://gwern.net/rubiks-cube where the big prompt setting up 'the scaffolding' is where most of the coherence and cleverness comes from.
Re: rhyming, I find that the default mode collapse is rhyming, but rhyming with a simpler ABCB quatrain and with a much looser meter than iambic pentameter; YMMV, of course.
But yes, scaffolding can do amazing things. I'd never read your Rubik's Cube essay, but the details like the footnotes are very clever, and would have led me to assume that this was organic if I wasn't reading closely. Your overall point at the end about how you could have done better writing it by hand, but wouldn't have had time to write an essay as long, is well-taken. I could have come up with a better continuation from "Blake?" than GPT-4o, but judging by past poetry-writing, it'd take me ~10 minutes per stanza, as opposed to (amortized across all attempts) ~20 seconds. My poems would be 'better', but I doubt they'd be 30x 'better', for any meaningful use of the term.
This is really pretty good. I'm just surprised that some lines don't scan to the extent they don't (in particular "It seems too much to ask you find someone less charming").
gwern help I have been trying to figure this out for weeks:
• how out-of-distribution can new data be for these big neural net AIs before their output is worthless? is there some measurement for this?
someone referenced "perplexity" and "V*" and "label rationalization" in a related thing but I'm not sure how much those apply or what V* even is
Agree wholeheartedly.
Regarding the book review itself - I'm very impressed with the Ottava Rima; it feels as natural as Byron's, and if it turns out this is Scott Alexander, I wouldn't be terribly surprised.
But I think it's appropriate to repost a comment I made here a few years ago, about Don Juan.
In Book 4 of his Odes, Horace wrote:
Vixere fortes ante Agamemnona multi; sed omnes inlacrimabiles urgentur ignotique longa nocte, carent quia vate sacro.
meaning approximately:
Many brave men lived before Agamemnon; but all are unwept and unknown, thrust into an endless night, lacking a sacred bard.
Byron quoted this at the beginning of Don Juan, but (as this review points out in its second stanza) reversed the meaning. Byron himself had the skills to be a sacred bard, and knew it - but where was his Agamemnon? Who could he write about that he wouldn't have some reservations over? Earlier in life, his writings made it clear that he wanted to write an epic sincerely...but between the tragedies in his life that weren't his fault and the ones which were entirely his fault, he'd lost his youthful optimism, and now was stuck with skills he could no longer use the way he'd originally intended. And so, in the first five stanzas of Canto I, he became convinced that *nobody* could write a sincere epic anybody, that these modern times (the early 1800s) were just too progressed for epics to work anymore, or for people to take them seriously.
Instead, he took a classic villain, Don Juan, and made Juan into a satiric and misunderstood hero, so that he would have something and someone to sing about.
A century later, GK Chesterton wrote a completely sincere epic - Lepanto - about a different man named Don Juan (Don John of Austria). I would love to know if this was done as a deliberate reply.
Yes, it feels very natural in imitation, but the voice is, of course, his own. I am very glad to see such writing. It far surpasses the formless tripe passed off as "poetry" nowadays.
The vast majority of poetry these days is written as song lyrics, to the extent that poetry in other contexts should probably be regarded as an extremely niche subgenre.
I agree
It's almost a platitude to say that nobody likes poetry anymore, and yet rap music is everywhere
Well, a seventy percent sincere epic.
> Cervantes on his galley sets the sword back in the sheath
> (Don John of Austria rides homeward with a wreath.)
> And he sees across a weary land a straggling road in Spain,
> Up which a lean and foolish knight forever rides in vain,
> And he smiles, but not as Sultans smile, and settles back the blade....
> (But Don John of Austria rides home from the Crusade.)
That's fair - even so, I'd argue that seventy percent sincere is significantly more sincere than Byron's estimated upper bound on modern-day high-quality epic sincerity, based on his anti-Southey introduction.
That's fair.
On the one hand, it's pretty hard to keep track of exactly what's happening or how long it's taking, and how much of this is directly from the story versus how much is flourishing for the sake of the rhythm.
On the other hand, I don't care anymore, this is excellent on its own.
>Its lavish scenes can make you smell the spices;
>Its wit can leave you rolling in the aisle.
Oh I get it. Byron invented the Bond movie. Exotic locations, endless affairs, captures, escapes , puns, sudden death...
Through poetry and verbage quite impressive
Our humble guide whisks us through this lay
But though your diction’s really quite expressive
It’s summary, at the end of the day
But it would seem an amature detective
in many a comment thread will now oft say
“I have to say that is, quite the review
But really I must ask: Scott is that you?”
I appreciate the effort, truly I do, but the rhyming verse makes reading this review a chore. I think I would've preferred unstylish yet concise prose.
While this review has an impressive gimmick
Ottava rima being used throughout
In order the original to mimic
Done skillfully, but still I have a doubt
Of the style's point, go on, call me a cynic
But the complexity is still about
The style is less of use and more just comical
The verses all excessively Byronical
(That is, great review, but its use of the same form makes it somewhat harder to understand than most reviews. If you can easily understand it, go read Don Juan!)
I realize that some here wish more modern poetry rhymed, but personally I loathe this sort of rhyme scheme and can't bear to read more than a few stanzas of it. I have no opinion of the review since I couldn't get through it, but I can't really vote for it given that.
Much prefer prose that reads like poetry. Give me Virginia Woolf.
Unless I'm reading out loud, I tend to read poetry as prose. So I definitely agree with you.
OK. I forced myself to read it slowly and aloud as suggested below, and my takeaway is that I never, ever want to read any Lord Byron. So at least I learned that. I can appreciate Robert Lowell or T.S. Eliot, but I can't find the aesthetic appeal to all that rhyming, and the lines aren't particularly evocative. It all seems so corny. But the fault is probably in me, I'm no judge of this type of poesy.
Is Lowell your example of rhyming poetry you *do* like?
If so, I agree.
I also wish more modern poetry rhymed, but not because I like rhyming poetry so much. I just want people to have some reason to work and rework their lines. When it comes to the older stuff, I often find the blank verse more appealing than the rhymed.
I have the Collected Poems of Robert Lowell and sometimes at night open to a random page and read the poem aloud and usually like it, but I'm no Lowell expert. I mostly think of his poems that don't rhyme but I guess some do.
It is weird that I, like most everyone, appreciate good song lyrics, which have to always almost rhyme whereas I can't appreciate much poetry that does. Obviously, poetry came from song.
I think the difference is that when it comes to reading my brain is tuned to prose. I mostly read literary fiction and mostly care about the quality of the prose, and good modern prose is about evocative language, the same as modern, non-rhyming poetry.
I mean, I like some rhyming poetry. I like Shakespeare's rhyming work, although I think a famous Rationalist proved by calculus that Shakespeare was his own father's ghost and not a great writer.
I feel like maybe I've been too negative on this review already, but just maybe, maybe it is bad poetry and that is the problem? Obviously more people like it than don't. I'm certainly not capable of writing rhyming poetry anywhere close to as good as that.
That's all fair.
Out of curiosity, have you read Pale Fire by Nabokov? I read it recently, and had the (perhaps unusual) reaction of ultimately liking the poem better than the prose narrative surrounding it.
Although more people seem to like rhyming poetry, it does strike me that more famous *epic poems* are unrhymed than rhymed. It could be that it's harder to sustain good poetic writing over a long space when you're also focusing on trying to rhyme. It could also be that you're not the only one who gets tired by one rhyme after another, even if people tend to enjoy it in smaller doses.
Yeah, I like Pale Fire a lot and agree that the poem in it is very good.
Another Pale Fire fan here.
Boy you'd hate Arabic poetry. You think *English* poems rhyme excessively, well...:
https://youtube.com/watch?v=VozG9qDtho0?&t=11m6s
I'm sure most commenters are here after reading, but if not, I would encourage you to read this one out loud. I found it much more tractable, and half the challenge for poem amateurs like myself is finding the right cadence and breaks to make it work. At first, I saw the verse and thought I shouldn't dedicate time to it, but admittedly the review's format succeeded in not just interesting me in the review but the book itself. It's a success in that respect, though I dread this format being anything more than a novelty.
I’m my experience poetry is always more enjoyable and more comprehensible when read aloud.
So it's like prose, except you have to do completely unnecessary extra work to understand it? I found this incomprehensible. After a few verses I had a headache. Opposite of what a book review should try to accomplish.
To the contrary, you learned in just a few verses you wouldn’t like this book. Super efficient.
Fair enough!
My first guess was that this was Scott, but I think Scott's a better poet than this, so I'm guessing it's someone else?
Probably the weakest review so far (of three, so not really saying much). The rhyming scheme is a cute idea, but it doesn't work very well and a lot of the lines don't quite rhyme or scan right making it jarring. And ultimately it delivers little commentary on the work. It's just a summary in verse
As a poem? Not bad. As a review? The worst Book Review Finalist ever. Hands down. Stunningly superfluous.
I applaud the book selection. Also, this is an heroic effort and "courageous". But no. Byron was a genius and "Don Juan", as another Greer (Germaine) wrote, is the finest comic poem in English. It's unwise (& immodest) to try to review by imitation. For one thing -- as noted above -- ottava rima is terribly hard in English (easier in Italian). When it fails, as here when lines don't scan, it's terrible: the faults protrude. Then Byron's "longeurs" serve the plot, usually by teasing out a local climax (see e.g. the brilliant stanzas 118-134 of Canto I here: https://petergallagher.net/byrons-don-juan-cantos-1-2-annotated.) They are clever, often politically incorrect and, with one or two exceptions, more entertaining than the story of his harmless, feckless hero. Here the longeurs are, alas, just long.
I have just started your annotated edition and appreciate the neat format of it, but am a little confused about this: "A play bill advertising a performance
(about 1832) at the Pavilion Theatre, Stepney, in the East
End of London. There were several operatic versions of
the Don Juan pantomime given in pre-Napoleonic London
that Byron may have attended."
Wouldn't Byron have been a little young to attend the theater in the "pre-Napoleonic" period?
You are absolutely right: I slipped up with the dates here. Byron would have been unlikely to be attending plays in London before (say) 1805 when he started at Trinity College in Cambridge. The "Pre-Napoleonic" period, whenever that is, must be before Napoleon started his European campaigns in 1803. So I should have omitted any reference to Napoleon and simply speculated (there's no proof) that Byron saw one of these performances. Thanks for that and please let me know of other errors you might find. Best, P
I'm really enjoying the photos and other illustrations; maybe all epics should be illustrated.
I've created a market on Manifold to bet on whether each of the finalists was written by Scott: https://manifold.markets/TimothyJohnson5c16/were-any-of-the-2024-acx-book-revie?r=VGltb3RoeUpvaG5zb241YzE2
I want to suggest that this is not a good thing to do. If we believe in anything here, it's that everything shouldn't be about personalities, gossip, and who did what. These book reviews are about writing great reviews, us as readers gaining some exposure to books we'd never normally read, and judging writing based purely on the quality of the writing. The anonymity thing matters.
It's probably a dumb complaint to say that the reviews here cater too much to a Rationalist audience, but I'm going to stupidly lodge it anyway. The typical reader here probably doesn't want to be lectured about vegetarianism or read a book review in verse. Those who reviewed these books (And I am one) probably aren't as typical as the readers. Full disclosure: I also submitted a review that wasn't chosen to be a finalist, so likely I am not objective about the subject.
I don't think most Rationalists want to read a book review in verse either. But Scott reserved 25% of the finalist slots for nontraditional books, so this is one of them.
Wait, isn't this just self-insert isekai fiction by the famous author of "I am very sad because this Greek couple I just met wouldn't sell me their twelve-year old daughter as a concubine: the poem" and "maybe incest with my blood-related onee-chan was not such a good idea: an apology video"?
Of course Lord B, the man so extra that Victorian England found him too ostentatious, would see himself in Don J. Britain owes us one for ridding them of him... although we're partially responsible for BoJo too, so let's call it even.
This definitely gets an A for effort but I really can't read it, even though I'm interested in the book it discusses.
Absolutely brilliant tribute to Byron's own genius.
A clean limerick on Don Juan,
Seems almost entirely wrong,
But given a touch of the 'tism,
And stewed in feminism
It's all I can do-so move on!
...
There once was a blogger from the Bay
Who wrote anon rhymed reviews one day
We couldn't follow his thoughts,
So we all lost the plot,
And hope he returns to his forté.
The penultimate stanza in section II is one line shorter than the rest. I wonder if that's deliberate, as the author's way of finding out how many commenters read the review out loud.
So now, having read OG Don Juan - wow, what a terrible poem. Part of it amounts to a "slam note" about his wife.
I conclude that the fame of the poem lies with Byron's looks and style; and his poetic fame is owing to other works. So to fairly evaluate this review calls for evaluating a picture of the reviewer.
Tried this a few times and stopped at various points after I lost track of what was being said. Not for me personally I'm afraid.
The verse is impressive. Only a few lapses of meter and form, and the rhymes can get up to three syllables deep. However, since I don't normally deal in verse, I kept getting distracted from the content by the format. I wish there were more analysis, because it seems to boil down to "the character didn't have much agency, but Byron himself did, and that seems better".
I didn't get much of a sense of why the book would be interesting, although I suppose the verse serves to let me have a taste of what it would be like. The content seemed like a lot of erotic stories from an older age, which might be entertaining were I in the mood, but those moods are rare.
Random bits:
It reminds me of this Let's Play of "Chrono Trigger" at the lowest possible level, which lapsed into couplets during the Black Omen: https://lparchive.org/Chrono-Trigger/Update%2018/
Is that you moldbug