Interactive Tool — Musk-era Twitter Emulator

A blog post by Albert Semple

30 Nov 2022

I made a Twitter Emulator!

"I used machine learning to analyse 1.3 million tweets and taught my AI to respond to left-leaning Tweets, emulating how the Musk-era Twittersphere responds to such opinions."

is what I could have written about this tool, but it's a complete lie.

I didn't use machine learning or AI. This is an art/poetry project, targetting the following audience:

It just randomly selects from a corpus of vile phrases, stitches them together to make a coherent response using this template:

{Optional excalmation!} Never mind that, what about {different observation}? You obviously think that's OK, you {name calling}, sitting {performing a mundane/degrading activity} {optional setting in your mundane/degrading scenario}. {Optional imperative to do something mundane/degrading}!

Inspiration

I published a poem called Things that don't exist which included a list of popular terms for things that aren't real. In this list I included the term "The southern England migrant invasion.". One of the responses to the poem went thus:

ReplyGuy85216: Maybe ask the poor old bastard that woke up to two albanians munching the contents of her fridge about the second one. Just because it doesn't exist inside the 4 walls of your pish stained wee cooncil flat doesn't mean it doesn't exist.

Me: Doesn't meet the definition of an invasion — [https://en.m.wikipedia.org/wiki/Invasion](Wikipedia article defining invasion)

ReplyGuy85216: So you think it's ok that pensioners that have paid into the tax system are being robbed at point of entry? Come on to fuck.

(I've changed the respondents name to protect the guilty).

I know, I shouldn't have engaged him, but I couldn't help it.

But I thought the response — or rather the "whataboutist" pattern of the response — was interesting and could be used as a template to respond to any social media post.

And in showing how tedious and formulaic these kind of responses are, I hope it will disarm them by helping people understand how easy it would be to set up a bot to share mis-information and hate.

How easy? My code, including all the underlying text options, comments and whitespace to keep it legible, totals less than 200 lines and took a couple of hours to write.

It's not worth engaging on platforms (i.e. Twitter) that seem to regurgitate, tolerate and potentially promote those kind of interactions between people and/or machines.

Subsequent Poem

So I wrote a concrete poem call "Whataboutism" which showed 6 options for each phrase, suggesting the user rolled a dice for each one to define the response, and shared it on Reddit in a left-leaning Sub-Reddit. It seemed to go down well, people seemed to get the point, and had a laugh at the ridiculously angry responses. Six phrases in four positions = 6^4 = 1,296 permutations of possible responses.

I've created a new tool that allows anyone to experience what it's like to express left-leaning opinions on Twitter without having to actually sign up or even visit twitter.

Automation

That was my Alpha, now moving to the Beta, I've automated it. No need to roll a dice and note the answers — just enter your hypothetical tweet in the box and it will generate a pseudorandom response.

I say pseudorandom rather than truly random, because I've built it so that you get the same responses for the same string entered. So if you get a particularly funny response you want to share with people, you can share the URL and they'll get the same response back.

With the increase in the number of phrases and extra positions, there are now more than 500,000 possible responses to whatever text is entered.

I hope this tool achieves it purpose. In summary — mothball your accounts on toxic platforms and move elsewhere. Mastodon, for example, seems nice. You will feel much happier.

Unlike other pages on this site, it won't work if you don't have JavaScript enabled.