Aesthetics for Birds

Aesthetics and Philosophy of Art for Everyone

Algorithmic Satire


We are witnessing the birth of a new comedic form: satire by algorithm. You want to make fun of some category of thing, and show how empty and mechanical and simplistic all the examples of that thing are. So you make a bot that randomly generates new examples of that thing. And the entire point is that it’s a bot. And often, it’s utterly crucial that it’s a dumb and obvious bot. This is why isolated exposure to the only one or two bot-Tweets or bot-memes doesn’t get you the full package. The real sharp end of the joke hits when you start to catch on to the rules, when the raw and obviously algorithmic nature of the bot reveals the utter banal predictability of its target.

Take, for instance, the Brooklyn Bar Menu Generator, which randomly generates bar menus for imaginary hipster gastropubs.

Screen Shot 2017-10-31 at 4.27.44 PM

“Seaweed with homespun salt” and “free-range sardine”

Screen Shot 2017-10-31 at 4.28.40 PM

“Activated chorizo with kale”

Screen Shot 2017-10-31 at 4.29.01 PM

“Farm-to-table water tacos”


Part of the joke is in how narrow the visual archetype is — the fonts for the menus barely change across a narrow range, the layouts are almost identical. The deeper part of the joke is how narrow the range of terms and ingredients is. And, of course, there are the occasional winks which give the game away — Brooklyn Bar Menu Generator spits out items like “tormented booze” and “naive oyster ramps” which wink to the reader at the brokenness and jokiness of it all.

Twitter’s full of these things, like Thinkpiece Bot, which auto-generates pseudo-intellectual clickbait headlines:

The first level of the joke here is obvious: the construction of these sorts of headlines is so obvious and so mechanical that a bot could have done it. (It’s a variation on that age-old complaint about modern art: “My kid could have done it!”) And, of course, the best way to demonstrate that is to actually go and actually build the damn bot.

But the subtle content of the comedy builds through many instances. You can’t just read one Thinkpiece Bot tweet, you need to read a bunch. Because it helps to see Thinkpiece Bot’s total failures (“Stop Blaming Fictional Character ‘Vivian James’ for Donald Trump”), the tweets where it careens hilariously against the outer boundaries of language and reveals it’s artificial bot nature (“Why Millenials Are Consuming Blueetooth-Enabled Cheering When Lawmakers Who Want Poor People To Die From Lack Of Health Insurance Get Cancer”), and the ones that just end up sounding like totally plausible think-piece headlines that would fit in just fine on the front page of Slate (“Millenials Are Using Callout Culture To Get Sex”).

The many instances are crucial, because you need to get a feel for how the bot is working. It isn’t funny if the bot’s workings are too obscure. Over the many instances, you learn the rules, and that’s the whole point. A well-made algorithmic satire is a way of demonstrating how easy it is to catch on to the rules of some trend. It exposes the robotic nature of some bit of our culture by making, you know, an actual robot.

This isn’t the only kind of bot comedy; algorithmic satire is a very specific kind of bot-comedy, made for tearing at those parts of our world that are fake, automatic, and suffocatingly predictable. There are other kinds of bot comedy. Take, for example, that the incredibly popular Inspirobot, which makes fake inspirational posters. I don’t think Inspirobot is a kind of algorithmic satire.

And though some of Inspirobot’s fake inspirational messages do come close to real ones, that’s not what people love about Inspirobot. People seem to adore Inspirobot at its most linguistically broken and absurdist, when it uses the form of the motivational poster and goes somewhere unexpected. Inspirobot is a joke about language; it gooses our logic centers. If we’re laughing at anything, we’re laughing at Inspirobot itself. Thinkpiece Bot is a very specific piece of satirical murder, aimed very precisely at modern think-pieces.

And of course, like other satire, sometimes algorithmic satire exposes the mechanical workings of things we took to be human so brutally, that the laughter is more cold than gut-busting.



  1. I am not sure about the usefulness of “algorithmic” as a a general descriptor. It seems that there is something distinct between formulaic bots (like ThinkpieceBot, cf. ) and Markov chain bots (like Brooklyn Bar Menu Generator). The former seems pretty continuous with Mad Libs, just slightly more automated; which makes me question how new the comedic form is. The latter might be genuinely new.

    • Petty comment: algorithms are a pretty general concept. I buy Dom Lopes’ view, that an algorithm is a rule-based system executed in a rule way, and that you can have algorithms run on the human brain. So even if Mad Libs were non-computer based, they might still be algorithmic.

      Better comment: what matters to me isn’t the kind of program running the algorithm, but the degree to which it generates that satire-by-a-asense-of-rule-based-ness. Mad Libs isn’t satirizing anything – I think it’s more like the Inspirobot, where you’re experiencing the chaos of language and generating absurdity. It may be algorithmic, but it’s not algorithmic *satire*. What unites Thinkpiece Bot and Brooklyn Bar Menu Generator is their aim of targeted, critical comedy.

      • I think I am okay with all of that. But then worry is that if you understand algorithm so broadly such that algorithmic satire just means “satire-by-a-asense-of-rule-based-ness” then it’s unclear how new the particular examples are. Genre conventions are rules, so there’s a trivial sense in which all satires are satires by a sense of rule-based-ness. More significantly, many writers/comedians/whatever have rules they internally follow, even if they rarely consciously articulate it. So I thought what’s more interesting about the Markov chain satires is that the rules are not even in principle accessible to the creator.

        (I suspect we might then get into what is a rule…)

  2. In the 1990s there was program called Kant Generator which output paragraphs patterned after bits of the 1st Critique.

    I subscribed to a Kant studies mailing list, and someone wrote in with a paragraph asking if the paragraph was from Kant and if so from where. They’d been given the passage by a friend. It turned out (unbeknownst to the person writing in) to be from Kant Generator.

  3. If you think the algorithm measures up to what an actual human might come up with, you should read more from the genre…

Leave a Reply

Required fields are marked *.