"I write like" meme

[armb]

http://nielsenhayden.com/makinglight/archives/012497.html
"H.P. Lovecraft, At the Mountains of Madness (from here): George Orwell.
George Orwell, Burmese Days, chapter one (from here): Margaret Mitchell.
Charles Manson quotes (from here, both pages, with the words “Charles Manson” removed): Kurt Vonnegut.
Mao Tse Tung (in translation, sources of quotes removed, from here): Kurt Vonnegut.
Unabomber’s Manifesto (from here): Mario Puzo.
Hitler’s Mein Kampf, chapter one (in translation, from here): Ernest Hemingway."

"Foo. Wanted cool; got balonium."

Comments

[keristor.livejournal.com]

See comment #20, it's apparently a Bayesian system. Personally I'd have added some things like type of words (allegedly Stephen King is allergic to adverbs, and some of the others are prone to adjectives) and punctuation (Arthur C. Clarke, for instance, is prone to use ellipses, which is where I get my penchant for using them) to differentiate a bit more. And added a metric for unrecognised words so it could throw out Latin and nonsense (possibly the latter could be assumed to be Lewis Jabberwocky and his famous Carol).

It might be fun to try to write one, at that. But I'd probably have to leave it for a few months after this one has died before starting it as a meme...

[uk-sef.livejournal.com]

It would also depend on what type of thing one was writing / submitting. A scientific publication would look inherently different from a romantic novel (one hopes!).

News reports also have a distinctive standard style (eg of awkward sentences where biographical asides are being wedged into a narrative using as few words as possible). It might be very hard to program a system to understand when puns are being employed.

Idiosyncratic spelling choices/mistakes would be much easier to recognise and count. Failure to use capital letters doesn't imply cummings-esque pretentions these days. It's more the norm - as is txt spk. You definitely need to include as a possible result "you write like a LOLcat" though.

[filklore-on-lj.livejournal.com]

Submitting one of my blog posts, at random, it says I write like David Foster Wallace, which I guess is kind of cool. However, another post of mine is analysed as Dan Brown, so that text is about to be nuked into oblivion.

[keristor.livejournal.com]

Oh, definitely the LOLCat result! IMO that by itself justifies such a project. (Hmm, can LOLCat be distinguished from TXTSPK? Not by use (or not) of capitals, many people including myself keep normal capitalisation in LOLCat. I suspect that it might need to look for phrases, like "can ha[sz]" and misspellings which are common in one but not the other.)

Heh, I'm talking myself into this one. As though I needed anothe timesink project...

[keristor.livejournal.com]

Just add a load of words no one else understands, then it will probably reclassify the post as James Joyce.

[uk-sef.livejournal.com]

Follow-up post:

http://nielsenhayden.com/makinglight/archives/012502.html