“may someday help in a more objective assignment of books...”

January 28, 2016 4:29 PM

Scientists find evidence of mathematical structures in classic books. [The Guardian] James Joyce’s Finnegans Wake has been described as many things, from a masterpiece to unreadable nonsense. But it is also, according to scientists at the Institute of Nuclear Physics in Poland, almost indistinguishable in its structure from a purely mathematical multifractal.

“The absolute record in terms of multifractality turned out to be Finnegans Wake by James Joyce. The results of our analysis of this text are virtually indistinguishable from ideal, purely mathematical multifractals,” said Professor Stanisław Drożdż, another author of the paper, which has just been published in the computer science journal Information Sciences.

posted by Fizz (28 comments total)

Is just another incarnation of power law dynamics. It's an attractor in the renormalization group space. Basically, if it's symbolic in nature, it will have that multifractal structure. Linguistic networks have been studied before. Money. City size. Societies. Can't do anything with it, although a group tried.
posted by hleehowon at 4:44 PM on January 28, 2016 [13 favorites]

(connection between linguistic networks and money to the multifractal spectra is the generalization of complex network laws by the multifractal hypothesis: basically, the adjacency matrices are multifractals for a solid number of complex networks, whatever the subject).

Sorry for the PDF linking.
posted by hleehowon at 4:46 PM on January 28, 2016 [1 favorite]

Obligatory DFW fractal reference.
posted by brappi at 4:49 PM on January 28, 2016 [3 favorites]

I am having trouble understanding how this is saying anything but that Joyce wrote long sentences while others wrote shorter sentences.
posted by mittens at 5:17 PM on January 28, 2016 [7 favorites]

That doesn't get you published, mittens.
posted by Justinian at 5:37 PM on January 28, 2016 [29 favorites]

What's awesome is that Justinian's use of mittens' name can read as both extremely polite and/or scathingly dismissive. Like a Schroedinger's reply.

...I wonder if there's a paper, there?
posted by Ryvar at 5:46 PM on January 28, 2016 [8 favorites]

I cynically assume that this objective system for assessing books will let us know how wrong we are to like what we like.
posted by Going To Maine at 6:08 PM on January 28, 2016 [2 favorites]

Of the books charted in the main link I've only read one, so can't comment there. But this statement, that 'stream of consciousness writing is the most comparable to fractals', seems self-evident. I get fractals as the patterned growth in nature across scales, so a grain of sand is like a pebble like stone etc. So it follows you won't see fractals in a writing style with rules for sentence structure and length, since that's an imposed order.
and the comment section there has many thoughtful posts.
posted by TDIpod at 6:11 PM on January 28, 2016 [2 favorites]

I cynically assume that this objective system for assessing books will let us know how wrong we are to like what we like.

"dude everyone knows women are bad at math, this must be the reason i don't like their books!"
posted by sallybrown at 6:21 PM on January 28, 2016 [3 favorites]

an imposed order

A Second Look at the First-Draft Version of “Finnegans Wake”

posted by mittens at 6:21 PM on January 28, 2016

Zipf's Law, Benford's Law

"Natural scientists recognize in 'Zipf's laws' the counterparts of the scaling laws which physics and astronomy accept with no extraordinary emotion - when evidence points out their validity. Therefore physicists would find it hard to imagine the fierceness of the opposition when Zipf - and Pareto before him - followed the same procedure, with the same outcome, in the social sciences." - Benoit Mandelbrot
posted by 0rison at 6:25 PM on January 28, 2016 [2 favorites]

Finnegans Wake is just a long string of odd Gaelic jokes loosely translated into English, and the multifractals are hilarious.
posted by ovvl at 6:41 PM on January 28, 2016

I know what fractals are. I know what networks are. I know what power law distributions are. I know what novels are.

I do not understand this.

What are the zoomed in/zoomed out levels in sentences in a novel? So let's say there are sentences that are 17, 25. 47 words long. Does that mean the paragraphs follow the same ratio in the number of sentences they ahve? And the chapters follow the same ratio in the number of paragraphs they have? Wuh? If it just means lots and lots of short sentences and a few outrageously long sentences, I see how that's power law, but I don't see how that's fractal.
posted by If only I had a penguin... at 7:44 PM on January 28, 2016 [2 favorites]

Here's the preprint of the actual paper.

If you know what the Fourier transform is, well, one of the things they're saying is that if you make the list of the length of each sentence in the text and take its Fourier transform, you get a power law spectrum. (This is different from having a power law in the histogram of sentence lengths.) Fractals also happen to have a power law spectrum. Beyond that, you got me.

I started reading this page to try to understand the implications of power law spectra, but it's not particularly positive about them either.
posted by a car full of lions at 8:07 PM on January 28, 2016

I'm sorry but this is completely stupid. Many long passages in _Finnegan's Wake_ are not sentences in any meaningful sense of the word sentence, as they are neither punctuated nor assembled as such.

How does it even make sense to talk about SLV (Sentence Length Variability) when the representamen (or most notable specimen, whatever) has units that cannot be meaningfully compared to other members of its class (e.g. _The Waves_).
posted by mistersquid at 8:13 PM on January 28, 2016

You know where this is headed. If we can take the FFT of books, we can pitch shift the books. Auto tune for books!
posted by RobotVoodooPower at 8:50 PM on January 28, 2016 [8 favorites]

So you think you have a power law for other intrinsic data structure chicanery and model-selection nonsense
posted by lalochezia at 8:57 PM on January 28, 2016 [4 favorites]

Is just another incarnation of power law dynamics. It's an attractor in the renormalization group space. Basically, if it's symbolic in nature, it will have that multifractal structure. Linguistic networks have been studied before. Money. City size. Societies. Can't do anything with it, although a group tried.

For real expected that link to be to Asimov's Foundation series.
posted by solotoro at 9:03 PM on January 28, 2016 [5 favorites]

If you look real closely, the scientists themselves never claim that the texts are fractals:

The results of our analysis of this text are virtually indistinguishable from ideal, purely mathematical multifractals.

See? They did something with the texts. And the results of their analysis, not the original text, resembled a fractal.

But it wouldn't be surprising if stream-of-consciousness texts resembled a fractal. After all, the fabric of our space-time continuum is equivalent to a hologram. And when you fold that down to a single consciousness, you get a fractal. And the fact that I'm writing this sends ripples through the holographic space-time continuum that (if the right higher-order being notices) will eventually get me published. Or lifted to a higher dimension. Or both.
posted by sour cream at 12:37 AM on January 29, 2016 [1 favorite]

Is this just an artifact of the analysis method itself? What do other books turn out like? What do similar analyses of other types of data look like?

I"m not really seeing that this is necessarily an interesting result.
posted by mary8nne at 1:59 AM on January 29, 2016

To this I say:

(bababadalgharaghtakamminarronnkonnbronntonnerronntuonnthunntrovarrhounawnskawntoohoohoordenenthurnuk!)
posted by chavenet at 3:30 AM on January 29, 2016 [2 favorites]

This just in; James Joyce wrote beauty pieces.
posted by ZaneJ. at 3:57 AM on January 29, 2016

My seminal work, 'On The Use Of Certain Transformative Functions In Information Synthesis", detailing a rigorous method of turning any old nonsense into academic papers, will be available from Springer in late summer (45pp, $550 hb, $445 pb, Kindle edition $430).
posted by Devonian at 4:33 AM on January 29, 2016 [2 favorites]

In a letter about the novel, Work in Progess as he then knew it, he told Harriet Weaver: “I am really one of the greatest engineers, if not the greatest, in the world besides being a musicmaker, philosophist and heaps of other things. All the engines I know are wrong. Simplicity. I am making an engine with only one wheel. No spokes of course. The wheel is a perfect square. You see what I’m driving at, don’t you? I am awfully solemn about it, mind you, so you must not think it is a silly story about the mouse and the grapes. No, it’s a wheel, I tell the world. And it’s all square.”

I could not read the entire article. Did they mention that Joyce drank a fuckton?
posted by bukvich at 6:28 AM on January 29, 2016

sour cream: I highly suspect that any text, not just Finnegan's wake, would produce this spectrum pattern. So any system of symbols. But not things that are not systems of symbols.
posted by hleehowon at 8:47 AM on January 29, 2016

This just in; James Joyce wrote beauty pieces.

Truly, we can barely imagine what horrifying and alien fractal geometry must occur in the works of H.P. Lovecraft.
posted by Going To Maine at 10:12 AM on January 29, 2016 [1 favorite]

Interesting, thanks!

(A lot of comments in this thread sound like people didn't read the article.)
posted by jjwiseman at 11:57 AM on January 29, 2016

"When the only tool you have is a hammer, every problem looks like a nail."
posted by bricoleur at 7:02 PM on January 29, 2016