the cops know who wrote desert

8 posts / 0 new
Last post
Hack The Words
the cops know who wrote desert

By Hack The Words

Stylometry is the study of writing style, used to find the true author of an anonymous or disputed text. A stylometric analysis finds similarities in word choice, punctuation use, sentence structure, and countless other ‘features’ that are assumed to stay mostly constant across an author’s works. Since the late 20th century, the field has been dominated by computational methods that can process millions of words a second, drawing from massive datasets of digital text to attribute authorship with a high degree of accuracy.

This is a serious problem for those of us who want to remain anonymous. The Anarchist Library is a beautiful project, fostering creativity and communication and action; but it’s also a centralized source for the writing of almost five thousand anarchist writers, coming in at almost 42 million tokens (individual words and punctuation marks) in size. If there’s a text under your name in the library, it could easily render any anonymous story, essay, or communique you publish not-so-anonymous after all.
a case study

To test my concerns about stylometry, I programmed a simple method that’s been around since the 90s. For those interested, it’s a single-layer perceptron model trained on 1000d stop-word frequency embeddings. The point is, it’s a very rudimentary model; researchers, businesses, the cops, all have access to fancier models with higher accuracy, and they have been used in solving criminal cases. If I can extract some stylometric meaning out of The Anarchist Library in my bedroom, so can they with their state-of-the-art methods and supercomputers.

For a start I fed my model some known texts. Sure enough, it was accurate for a most of the works I tested, selections from Noam Chomsky, Errico Malatesta, Emma Goldman, and Bob Black, all authors with a large number of texts on the library. It performed worse for authors with fewer works, for foreign language writers with English translations, and fiction. The better models used by ‘legitimate’ people will surely work much better than this, but it gives us a baseline.

Next I wanted to satisfy my curiosity and try an anonymous text. I chose Desert, the most popular anonymous text in the library at the moment. I hit enter, and within a few minutes my model spat out a name with a 74% confidence score. Sure, there’s a chance that the author of Desert hasn’t written anything else on the library under their name, but if they have, this is it. After all, that’s a higher confidence than it gave for David Graeber having written Bullshit Jobs.

I have no interest in who wrote Desert. The cops probably don’t either, but if it were instead a communique about a crime, they would be, and stylometry gives them a way of finding out who wrote it that works no matter how many digital security measures the author took.

P.S. I would love to share my test in more detail, but turns out code is even easier to do stylometry on than the English language. My email address is at the end if you have any specific questions.
stylometric nihilism

I haven’t given much thought to it, but I think our distinctive writing styles are a learned behavior. From the time we’re taught how, the act of writing is mixed up with a cloud of social constructs. We’re taught that what we write should be unique, so that it doesn’t infringe on the intellectual property of others, and that it has to be consistent with other things we write in order to have true ‘artistic vision’. The insistence that our words should work to differentiate ourselves from others is what pushes us to develop a unique writing style that can be used to identify us.

There’s already some brilliant theory about how negating these constructs and values around art, particularly intellectual property, is necessary. I think that it’s also essential for protecting our anonymity. Plagiarism, for example, confuses your writing style with that of those you plagiarize. Unlearning these things isn’t easy (I couldn’t find a way to write this little report without leaning on entirely my own words, in my distinctive style), but I hope it’s something that we can all take small steps towards. For some, it could be what saves them from years in prison, or worse.
hack back

I also hope that we see more stylometry applied to creating anarchy. Necessarily burdened by the imaginary constructs of their beloved status quo, the right and the center are easily de-anonymized using these methods. It isn’t beyond the ability of a few punk programmers to, say, create a database of fascist writers’ styles that updates in real time, drawing on information from already existing leaks to de-anonymize those behind new groups and hateful actions.

If you want to discuss or collaborate (with me of all people), write to hackthewords@protonmail.com. Encrypted mail only please.

anon (not verified)
Dear OP,

Dear OP,

Link us to the source code repository or be laughed out of town as an alarmist faker.

Also, if you wanted to know who the anonymous author of Desert was you could have just asked, m8.

anon (not verified)
weird flex, but ok cop

weird flex, but ok

cop

Brian Anleu (not verified)
was your linguistics class

was your linguistics class fun?

Isaac (not verified)
https://iwl.me/

https://iwl.me/

I just put in my buddy's criticism of metaphysical arguments in radical politics. It said he's Agatha Christie.

I got Asimov's, so I'm not going to complain.

anon (not verified)
This shoulda ended with (btw

This shoulda ended with (btw I'm a cop, that's how I know... coz you know that's what I'm doing here on this very site *oink oink*)

So yeah, sure, maybe the author(s) didn't know about stylometry back in the days. But what if... what if they knew and you got mislead by some brilliant writer?

Any competent, experienced writer can play with their rhetoric and writing style, as they know more than any cop how to deconstruct language. Of course probably not a Dan ShitBrown or a Tom Clumsy... but you know the actual real writers with skills and talent, and one who looks at the details. The hell... if I wanted to I could equally use CrimethInc's typical writing style in order to incriminate a few persons, but I won't.

So it's still not an exact science, and leaves space for error... Just like any cop's "evidence".

Aaaaaand here's a fun challenge for the author:

Identify who I am, judging by these words. Anything like an online pseudonym (nick) will do.

Let's see if you're that good.

anon (not verified)
David Foster Wallace

David Foster Wallace

anon (not verified)
Got me...

Damn!

*takes private jet to some Pacific island*

Add new comment

Filtered HTML

  • Web page addresses and e-mail addresses turn into links automatically.
  • Allowed HTML tags: <a> <em> <strong> <cite> <code> <ul> <ol> <li> <dl> <dt> <dd>
  • Lines and paragraphs break automatically.

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.
CAPTCHA
This question is for testing whether or not you are a human visitor and to prevent automated spam submissions.
r
2
A
*
f
d
h
m
Enter the code without spaces.