Is writing style determined by a computer algorithm?

[Math Is Hard]

I tried some e.e. cummings (my sweet old etcetera) and got James Joyce as a result.

Sylvia Plath (Lament)was recognized as Arthur Conan Doyle.

T.S. Eliot (The Love Song of J. Alfred Prufrock) yields Charles Dickens.

 

[GeorginaS]

I tried some e.e. cummings (my sweet old etcetera) and got James Joyce as a result.

Sylvia Plath (Lament)was recognized as Arthur Conan Doyle.

T.S. Eliot (The Love Song of J. Alfred Prufrock) yields Charles Dickens.

That's just party pooping.

But yeah, no kidding. At the very least you'd expect the code to have access to a database of popular writing.

 

[Pythagorean]

I tried some e.e. cummings (my sweet old etcetera) and got James Joyce as a result.

Sylvia Plath (Lament)was recognized as Arthur Conan Doyle.

T.S. Eliot (The Love Song of J. Alfred Prufrock) yields Charles Dickens.

Yeah, but to be fair, if you put in text from Charles Dickens, you get Charles Dickens, and you get Arthur Conan Doyle for works by Doyle.

1 of the 5 trials for Dickens, I got Stephen King. All the inputs came from A Christmas Carol, so I don't think they cheated and just put a text identifier in, I think it actually does some kind of analysis.

 

[Pythagorean]

I did a nonsensical rant with a lot of curse words and I got Cory Doctorow (whoever that is)

 

[Pyrrhus]

I pasted my writing from a journal paper... It says I write like Edgar Allan Poe, WOW

If you just spam "elf" a bunch of times you get J.K. Rowling

The word murder? Edgar allen poe.

I suspect when they say it analyzes your word choice and writing style, they really just mean word choice. Probably a simple document comparison algorithm based on word counts to see which author's writing you compare most similarly too

Interesting exercise: pick a word, and guess which author you will be compared to when you put it in (you'll need to repeat the word a bunch of times to get past the minimum word limit... use copy paste for this).

Perhaps for some words, but in my journal article the word murder is nowhere to be found.

 

[xunxine]

I did a nonsensical rant with a lot of curse words and I got Cory Doctorow (whoever that is)

He's a blogger!

blog entry + nonsense = Cory Doctorow?
Is he any good?

I guess it's database of authors is limited, ...

I'm inclined to agree with this.

I got Cory Doctorow, whom I've never heard of before.

But the first paragraph from Sarah Palin's book gets James Joyce...Scary!

Maybe Sarah Palin writes like James Joyce? Who writes strictly original stuff nowadays? The way we write is very much influenced by and mixture of the things we read.

 

[Pythagorean]

when I put garbage in: kajfd ;lkaelkm afcsl;ekj l;a,smef lkjesar

I get James Joyce

 

[GeorginaS]

when I put garbage in: kajfd ;lkaelkm afcsl;ekj l;a,smef lkjesar

I get James Joyce

That, actually, entirely makes sense.

 

[TheStatutoryApe]

when I put garbage in: kajfd ;lkaelkm afcsl;ekj l;a,smef lkjesar

I get James Joyce

Well obviously you found yourself a thunderword.

 

[Pythagorean]

huh, interesting:

http://atlantapoetsgroup.blogspot.com/2007/10/query-on-joyces-thunderwords.html

 

[Jack21222]

I entered 3 essays from my ethics class I just finished, and came up with 3 different authors.

H.P. Lovecraft
William Gibson
Dan Brown

Until this thing can come up with a fairly consistent output, I'm not sure I trust it.

 

[TheStatutoryApe]

I entered 3 essays from my ethics class I just finished, and came up with 3 different authors.

H.P. Lovecraft
William Gibson
Dan Brown

Until this thing can come up with a fairly consistent output, I'm not sure I trust it.

Ever consider that maybe you are the one that isn't coming up with some consistent output? ;-p

 

[Office_Shredder]

Perhaps for some words, but in my journal article the word murder is nowhere to be found.

I'm not really sure what your point is here...

 

[Jack21222]

Ever consider that maybe you are the one that isn't coming up with some consistent output? ;-p

My writing style is fairly consistent by my own estimation. This website only goes by word selection, and not things like sentence structure. It is funny, though, that my essay about anthropogenic climate change matched up with Lovecraft, and my one about the value of science to society came up as Gibson (the steampunk writer).

 

[hypatia]

Eh gads...Bram Stoker

 

[Danger]

I went into the test upon first provocation, and so have missed several pages of responses. My input was the original opening of my novel. (I have since instituted a new opening that consists of a few paragraphs ahead of what I submitted.)
Leo Tolstoy?
I love and respect old Uncle Leo, but when the hell did he ever write about a hypersonic fighter aircraft ripping up the Thule Greenland NORAD tracking station?
Methinks that this site extrapolates a tad overly.

 

[Noxide]

I bet you if you type random garbage it will tell you you write like some great author. I'll paste the following and see what it says:

"When Adam sidewalk into fairground red the elephants tube slid before the night."

Your Badge
I write like
Margaret Atwood.

I wouldn't be surprised if it just picks a random author. If it helps your ego fine but don't take it seriously.

To be fair, Margaret Atwood is a rather garbacious writer.

 

[Pythagorean]

This website only goes by word selection, and not things like sentence structure.

The site says:

Check which famous writer you write like with this statistical analysis tool, which analyzes your word choice and writing style and compares them with those of the famous writers.

I'm not sure what they mean by 'writing style', but how have you ruled out sentence structure?

 

[Office_Shredder]

I'm not sure what they mean by 'writing style', but how have you ruled out sentence structure?

Sentence structure is really hard to define/analyze with a computer. Word count is really easy.

 

[BobG]

Words per sentence? Length of words?

How does Microsoft Word analyze the reading level of any paper you write? It at least looks at the number of passive sentences.

 

[Danger]

'm not sure what they mean by 'writing style', but how have you ruled out sentence structure?

My thoughts exactly. The site (alleged site?) asked for a sample, but didn't specify what sort of sample. I've written comedy, serious ****, in between... one novel, one movie screenplay, half a dozen TV scripts, a graphic novel inroductory issue, who knows what else... and this thing boils me down to Tolstoy based upon the intro to a novel that is already 30 years out of date?

 

[Jack21222]

I'm not sure what they mean by 'writing style', but how have you ruled out sentence structure?

From an article about the site. The creator of the site seems to agree with this conclusion, if the article is even remotely accurate.

Chestnykh modeled the site on software for e-mail spam filters. This means that the site's text analysis is largely keyword based. Even if you write in short, declarative, Hemingwayesque sentences, its your word choice that may determine your comparison.

Most writers will tell you, though, that the most telling signs of influence come from punctuation, rhythm and structure. I Write Like does account for some elements of style by things such as number of words per sentence.

Chestnykh has uploaded works by about 50 authors – three books for each, he said. That, too, explains some of its shortcomings. Melville, for example, isn't in the system.

But Chestnykh never expected the sudden success of the site and he plans to improve its accuracy by including more books and adding a probability percentage for each result. He hopes it can eventually be profitable.

"I think that people really like to know how they write, even if it's not accurate results," said Chestnykh. "Still it's fun for them."

Source: http://www.huffingtonpost.com/2010/07/17/i-write-like-website-goes_n_650037.html

Emphasis mine.

 

[Pythagorean]

I see. I was impressed that it could identify the authors in its database, at least.

He hopes it can eventually be profitable.

:P

 

[Danger]

[spookyquote]This means that the site's text analysis is largely keyword based. Even if you write in short, declarative, Hemingwayesque sentences, its your word choice that may determine your comparison.[/spookyquote]
Hmmm... that should be it's your word choice that might determine... etc.. How the hell is this guy going to analyze the English language when he's illiterate in it himself?

 

[Office_Shredder]

[spookyquote]This means that the site's text analysis is largely keyword based. Even if you write in short, declarative, Hemingwayesque sentences, its your word choice that may determine your comparison.[/spookyquote]
Hmmm... that should be it's your word choice that might determine... etc.. How the hell is this guy going to analyze the English language when he's illiterate in it himself?

Because analyzing word choice doesn't require any knowledge of the language

 

[Danger]

Because analyzing word choice doesn't require any knowledge of the language

I can see that you have succeeded Artman as my nemesis. You are putting me into Commander Data mode: "Your analysis is correct, Captain, and quite intriguing. I request a few minutes and Geordi's assistance to temporarily negate the 3 Laws so I might be able to introduce my boot to your ***."

 

[rootX]

I believe it is not very hard to develop an algorithm that uses criteria and authors large amount of texts as inputs to determine different groups. You can continue refining the criteria until each group has only 1-2 authors. Once that is done you just need to take user inputs, determine their groups based on the criteria, and provide them author that shares the same group.

 

[Danger]

I believe it is not very hard to develop an algorithm that uses criteria and authors large amount of texts as inputs to determine different groups. You can continue refining the criteria until each group has only 1-2 authors. Once that is done you just need to take user inputs, determine their groups based on the criteria, and provide them author that shares the same group.

So the next time I take the test, should I totally ignore my novel and my teleplays and instead submit my meagre contributions to Penthouse Letters? They're cheezy, but pay well...

 

[HeLiXe]

First I pasted an assignment on religion from my humanities class and it said I write like Dan Brown.

Then I pasted an assignment on cognitive and behavioral psychology from my psych class and it said I write like William Shakespeare

 

[Danger]

Then I pasted an assignment on cognitive and behavioral psychology from my psych class and it said I write like William Shakespeare

Was that, by any chance, based upon spelling? Old Willie signed his own name at least half a dozen different ways. That's one of the reasons why it was so difficult to determine which of the plays attributed to him were actually written by him.

 

[HeLiXe]

Well in which regard? The humanities assignment had words which could be considered as misspellings, as they were from another language, while the psych assignment had no foreign languages, abbreviations or anything else which could be considered a misspelling.

 

[Pyrrhus]

I'm not really sure what your point is here...

My point was in agreement that the algorithm could be heavily procedural in nature (e.g. Check word count -> compare word to a list, and so on). I doubt it is a fair assessment of writing style, as most here have discussed.