Heretical Things Statistics Tell Us About Fiction | Hacker News Reader

Heretical Things Statistics Tell Us About Fiction

newyorker.com

90 pointsPetiver35 comments

The movie industry, unfortunately, has this figured out. The classic guide was "Screenplay" (1979) by Syd Field, which lays out a fairly rigid structure for movies. Then, in 2005, Blake Snyder published "Save the Cat", which is a rigid guide to what has to happen in each set of pages in a screenplay. Here's the summary on one page.[1]

This pattern is so effective that almost every big-budget movie from Hollywood now follows it. Very closely.[2] So closely that audiences expect it. Hollywood got a design pattern that works, and they're sticking with it.

[1] https://goodinaroom.com/wp-content/uploads/2013/02/Blake-Sny... [2] http://www.savethecat.com/beat-sheets-alpha

As someone with some relevant industry experience, I can tell you that nearly 0 professional screenwriters or development executives are consciously sticking to a rigid structure as prescribed by some book. Instead, what you're seeing is the difference between after the fact analysis and the process of creation. Things like the three act structure and certain beats appearing in many movies exist because those are just natural ways to tell a story. The three act structure is, at the most basic level, just the series of beginning, middle, and end. Of course any movie has these sections by necessity and there are better and worse ways to navigate through each.

Just to be clear though, no one is sitting around talking about Snyder's beats or terminology. In the industry, the only terms you'll here from these books are discussions of the three act structure because that is so vital to this sort of material. All that said, I do think the books are a good starting point for someone who wants to think critically about story, but don't think that these are blueprints for Hollywood success. If they were, every screenwriter would be paid scale.

As a counterexample, see Paul Reubens on writing "Pee-wee's Big Adventure":

"I bought the Syd Field book on writing screenplays, and me, Phil, and a guy named Mike Varhol did exactly what was said in that book. I mean, exactly, right down to getting three-by-five file cards and writing each beat down and putting it on a bulletin board. We did every single thing like that. As a result, Big Adventure is extremely classical in its structure because we were literally doing it by the book. So on page thirty, Pee-wee’s bike gets stolen. On page sixty, it gets found. It’s a whole thing."

https://www.villagevoice.com/2016/03/15/pee-wees-big-friends...

> As someone with some relevant industry experience, I can tell you that nearly 0 professional screenwriters or development executives are consciously sticking to a rigid structure as prescribed by some book.

I'd like to take your word for that, but numerous other insiders have said the opposite, and also noted a significant convergence around Snyder's precise timing breakdown since the book was published.

> The three act structure is, at the most basic level, just the series of beginning, middle, and end. Of course any movie has these sections by necessity

Traditional three act structure is more specific than that, it specified in broad terms what happens in each part, and, no, it's not fundamentally essential; there are other story forms, though they are rarely used in well-funded American feature films.

To someone outside the industry, the idea that much Hollywood screenwriting follows a rigid formula seems plausible. There may be creative auteurs creating bizarre plots with delightfully unpredictable pacing, but those scripts aren't being filmed and distributed to the public.

Indeed, it's getting to the point now where all TV/movies seem to be using the same tricks around pacing, creating audience involvement with characters, generating tension etc., to the point where I recognise it easily and then cannot continue watching. In contrast, and obviously I'm generalising massively but still, I find that older media doesn't seem quite so desperate to build me up into a frenzy of emotion, and seems happier to let a story develop at its own pace without resorting to shallow tricks to liven things up.

I mean sure, just like anything there is a certain method to writing a story. That doesn't mean that writing a movie is just paint-by-numbers like you seem to be implying. A programmer doesn't invent a new language every time she's going to write an application.

Writing a good movie is really, really challenging.

Would you agree they don't need to consciously stick to it, because that structure is internalized already as "the best way to write a mainstream movie"?

Just to be clear, I'm not a writer. That said, there are certain structural elements one expects to see in a well-crafted story... the same way you expect to see certain things in a good program.

Memento, Cloud Atlas, Moby Dick, 2001 space odyssey, melancholia, American psycho...

Not that I entirely disagree with you, but I think it's more analogous to say that Hollywood is in the business of making burgers. You expect certain structural elements in burgers: buns, centre, sauce. They're incredibly popular for a reason.

The problem comes when we say burgers are basically the superset of decent or nutritious cooking.

Looking for patterns in storytelling is a popular thing to do, see for example "The Hero with a Thousand Faces" by Campbell (1949) or "The Uses of Enchantment" by Bettelheim (1976).

The music industry is now doing this too. Using statistics and machine learning to recognize songs that are catchy and could be potential hits. I don't remember most of the details, but there's a decent amount on this in the book Automate This. Which is already a few years old and I assume it's only gotten worse since then.

Whatever they're doing doesn't seem to be working. What are the great music acts coming out of the industry in the last five to ten years? It is all a bunch of copy-cats without any large impact on society, unlike music produced up to the 70s.

Glossing through save the cat- it doesn't seem like a restrictive guide, but rather a set of reasonable narrative components, like stating the theme and having a narrative catalyst. Frankly this seems fine, if not beneficial. I would be more concerned about the formulas that drive your average Michael Bay quality film or second-by-second construction of trailers.

The key new idea in Save the Cat is that both dramatic elements and pacing are standarized. Save the Cat tells you on what page of a screenplay the various elements should appear. Screen time allocation is rigid.

Yeah... and that's not really accurate.

Applying statistics to works of fiction is mildly interesting, much as applying geometry to paintings. But after you tell me all about your statistical analysis of Faulkner's prose or how the Golden Mean is distorted in Van Gogh's paintings, you still haven't touched the thing that makes the works of Faulkner and Van Gogh much more interesting than those of their less-known contemporaries. That's why it is called "art" and not "science."

I've always wanted to conduct an experiment where we put some higshcoolers who wanted to major in art history in isolation (except for sterilized food and some math textbooks slipped under the door) and had them reconstruct the periods and choose important artists from scratch.

How much of the canon would turn out the same?

You'll be amused to know that similar experiments have been run for music:

2006: "Experimental Study of Inequality and Unpredictability in an Artificial Cultural Market" http://www.princeton.edu/~mjs3/salganik_dodds_watts06_full.p...

Salganik & Watts 2009 "Web-Based Experiments for the Study of Collective Social Dynamics in Cultural Markets" http://www.princeton.edu/~mjs3/salganik_watts09.pdf

2008: "Leading the Herd Astray: An Experimental Study of Self-fulfilling Prophecies in an Artificial Cultural Market" https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3785310/

Similar to the plot of "Pierre Menard, Author of the Quixote"

https://en.wikipedia.org/wiki/Pierre_Menard,_Author_of_the_Q...

I wouldn't agree. Menard is about authorship of a rewritten identical work - the GP was talking about people selecting works to potentially produce a different canon, and there's no question of an author.

Right... but it's similar.

At "How the Mona Lisa became so overrated" from Vox, you can see a unique circusntance that made the picture legendary beyond its initial merit.

https://www.youtube.com/watch?v=d2wy7Fp2fqw

Well maybe, but see http://www.gutenberg.org/files/2398/2398-h/2398-h.htm#leonar... from 1873.

Reminds me a little of the thousander maths in Stephenson's Anathem.

Reminds me of the Auditors in Pratchett's Thief of Time.

Sure but that doesn't mean we can't understand why some works of art resonate with us. The computational folkloristics field has found some very interesting patterns (see for instance http://www.mftd.org/index.php?action=atu). In "Religion Explained" Pascal Boyer describes why certain myths resonate with us and others do not.

On authors favorite words: "The words must be used in half an author’s books, at least once per hundred thousand words; they can’t be proper nouns... Isaac Asimov prefers 'terminus,'"

"terminus" is the proper name of a location in Asimov's Foundation series. Did Ben Blatt's analysis fail to account for common words used as proper nouns, or does Asimov use terminus outside the Foundation series?

Apparently the word has to appear in half their books. I don't know if he took the whole of Asimov's 900+ books

Wikipedia puts him at 40 novels, enough short stories for maybe 40 short to mid-sized collections (382 stories), 280 non-fic books, and 146 as editor or annotator, which surely don't count. 504 books included edited/annotated. If one were only using fiction and excluding duplicates (I'm sure there were collections of short stories of his which were mostly or entirely repeats from earlier collections) then that's maybe 80 volumes of unique material (I'm sure there were tons of dupes in various story collections), tops. They'd easily fit in 40-50 fairly ordinary vols. I bet if you were willing to put multiple "books" in one (a lot of his juvenilia are really short) and probably 10-15 LOA-style volumes.

Not that that's a small amount of fiction to have written, but even including the non-fic it's nowhere near 900.

Yes, seems my off-the-cuff memory's is counting an individual short story as a book & even then doubled the value

http://www.asimovonline.com/oldsite/asimov_catalogue.html

This is a really good point. It seems likely that proper nouns were not taken into account.

Another dead comment made a similar point: https://news.ycombinator.com/item?id=14886678

I'm curious to see how someone like Neal Stephenson would be identified. His writing style seems to change to fit the times and persons he writes about.

I don't know if this was taken into account, but in Isaac Asimov's Foundation series, "terminus" was very much a proper noun, being the name of the very planet where most of the plot takes place.

I wonder if he named it that because he liked the word so much, or if the word became his "favourite" after using it that way? It would be cool to see how author's usages of these words changed over time.

Reply on news.ycombinator.com