In September, the collective deflation of Twitter was almost audible. The famous, beloved account @Horse_ebooks was unveiled as a hoax–or, well, it was unveiled to be something, at least. Art? Maybe.

We learned there was a pair of humans behind the account, and that they’d manufactured its tweets. The operators were mostly silent on the subject, but a _New Yorker _profile out this week finally helps us understand their motivations.

The account–a malfunctioning text-robot, at first–tweeted surreal, occasionally profound inanities, like “Everything happens so much.” People loved this horse. I loved this horse. So when it turned out the account was under the control of human beings (at least for most of its existence), people felt betrayed. The profile, penned by Susan Orlean, who first unmasked Horse_ebooks, is out this week, and the wizards behind the curtain, Jacob Bakkila and Thomas Bender, make the case that their horse was an elaborate work of art native to the Web.

The duo had cut their teeth on “performance mischief” like a series of videos on how to pronounce words (“haute couture,” and eventually, “Timothy Olyphant”) and a faux tourism ad for Milwaukee.

Bakkila found the Horse_ebooks account to be an ample canvas for another performance, and purchased it in exchange for $250 worth of e-books from the Russian Web developer then running it. Bakkila began posting bits of text to the accounts, mostly sliced-up, dadaist sentences culled from real websites. The account became more and more popular before the big reveal: there was a human there all along.

Even now, as the profile points out, opinion is divided on whether this was a work of art, or a huge scam. (The profile, which is pay-walled, is really worth the price of admission for a print copy, by the way.) Orlean writes about what made the tweets so borderline-beautiful: “They were more peculiar and more evocative than bots usually are; they were like found poetry in an otherwise crass medium.”

Except, as wonderful as it was, this isn’t what most of the account’s followers had signed up for. They’d hoped it was an unintentional work, that its poetry was really “found,” and not by people. Eager Twitter users followed the machine for a laugh, and were shocked when the machine looked back and laughed, too. Is that art? Or just a simple hoax?

It’s not like this horse kicked off such a debate; we’ve been having it for a long while. Consider the case of Ern Malley, a 1940s “poet” who turned out to be fictitious. A pair of writers submitted purposely terrible work under the fake author’s name, to prove that modernist poetry was drivel. The hoax was eventually uncovered, but not before striking a serious blow to the form in Australia. Bizarrely, and in a way the creators of Horse_ebooks would no doubt appreciate, the work eventually became seen as a prime example of playful surrealist poetry. This is the inverse of what happened to Horse_ebooks: instead of a hoax that turned out to be unintentional art, the account was intended to be a work of art and taken to be an elaborate hoax.

That won’t do robot-based art, something that can be seriously amusing, even sometimes poignant, any favors. Take Darius Kazemi (on Twitter, @tinysubversions), who’s created gorgeous text-shuffling robots, like one that displays instances of executed Texas inmates saying “love” in their last words. That’s a perfect example of what makes Web art great: it cedes some, and only some, control to the machines. The results are only half intentional, and the beauty and humor comes from everything else. That’s why Horse (and “Malley’s” poetry) was so wonderful at first, and so unfortuante after. Without that element of chance, it’s no experiment–it’s closer to the Turk, a supposed chess-playing machine, mentioned in Orlean’s profile, that turned out to have a human being stuffed inside. If that was art, it certainly wasn’t very good art.