What happens to a Twitter bot after its maker dies?

There’s no good way to ask people if they think Twitter will outlive them.

I don’t mean any offense to Jack Dorsey and Co., but one must ask this awkward question when trying to find out programmers’ plans for their Twitter bots when they die.

Bots (for lack of a better term) are programs that use an application program interface (API) to post to or pull information from various services, and they’ve have become ubiquitous on Twitter. There are bots that sort pixels, suggest Marxist startup ideas, or behave more like teens than teens. Twitter said in a 2014 SEC filing that 23 million of its users were automated.

The people behind these artsy bots are a friendly, loosely-organized community of programmers, artists, journalists, and anyone else who feels like making bots. They’ve got hashtags, most prominently #botALLY; a roving Bot Summit which will take place in London this year; and a “botifesto” on Motherboard that describes the present and future state of bots.

As part of that future, botmakers must consider what will happen to their script kiddies after they die. When I emailed Darius Kazemi, a prolific internet artist and bot maker, about the question he quickly responded. “I think about this a lot,” he said.

Usually, it’s “in the context of robustness in general,” which is to say the bot’s survival if routine changes happen, such as an issue with his web host or an API change.

Some of Kazemi’s bots have already run untouched for 3 or so years because they don’t require much maintenance. Those could continue after his death as long as someone paid for the server where they live. In other words, the bots’ existences are linked now not to Kazemi’s survival but that of his credit card.

Bots are “processes, not artifacts,” he said. A bot’s survival beyond his lifetime or the lifetime of Twitter is “about longevity of the algorithm, more so than the actual piece of code running.” He offered the example of Two Headlines, a popular bot he created that, as its name suggests, takes two recent news headlines and mashes them together:

“If I died and Two Headlines went down, somebody else [could] read my article about how it worked and [make] their own version of it,” said Kazemi. “And maybe Twitter’s gone by then, too, and they make a version of it that works for whatever new service is around. And it might be a little bit different and it might make more sense in the context of that service.”

But to Kazemi, that means his bot would still be alive, and more importantly, relevant.

By way of comparison, he offered the example of John Cage’s 4:33, the musical composition which consists of four minutes and thirty-three seconds of ‘silence,’ but which is really meant to highlight the sound created by and around the audience. “John Cage did not predict that peoples’ cellphones might be going off during a performance, but the process that he laid out…that still works today, and it brings into the fold these environmental things,” he said.

Pragmatically, though, Kazemi is pretty sanguine because “I’ll be dead so it won’t matter to me.”

However, there’s also the issue of how to deal with bots that aren’t entirely yours. For instance, Brett O’Connor, a programmer and artist from Colorado, hosts a number of what he describes as “twitter_ebooks bots” for others. They take the tweets of a specific user and mix them together to create new tweets. This leads to slightly broken, uncanny versions of those people. For example, here’s one O’Connor built based on my friend Caroline:

While the bots are created by O’Connor, and hosted alongside his other bots on a laptop, they’re drawn from someone else’s tweets. He says he’s often wondered what to do if the person the bot is based on dies.

“Should I keep it running as some kind of memorial?” he said by email. He hasn’t posed this question to most of the people on whom his bots are based. Though he has had at least one person tell him they want their bot avatar to live on after their death. “They don’t care if it’s creepy. I rather like the idea,” he said

But O’Connor is still concerned about bots without stewards, and told me he’s “often thought about adding a feature…where if someone replies with ‘delete,’ it immediately deletes [an offending tweet] without my intervention.” He loves the idiosyncrasy of bots, but says that “a bot is right twice per day in the same way [as a broken clock], and that serendipity is what makes the fun.”

“Unnnnforrtunatlly it also says ‘heil hitler!’ twice per day too,” he added.

O’Connor also hopes his bots will live on through other people playing around with them. Like many bot makers, he keeps the code for some of them on Github, so that others can run it when he can’t anymore. “Bots are lucky that their DNA is so easily transferrable and restorable,” he said. “It’s like Jurassic Park every day on Github, right?”

thricedotted, another bot maker who specializes in accounts that create portmanteaus and bad jokes, has been thinking about contingency plans since 2014, when they posted this tongue-in-cheek tweet in their first months making bots:

Now thricedotted says these bots shouldn’t live on without a human caretaker.

“I don’t think this is true for every bot, but for my own I feel strongly that there should be a sentient being behind them at all times,” they told me via email. “I would be happy to leave them in the hands of a trusted caretaker…somebody who can go in and remove any offensive material that might slip through.”

It doesn’t matter who the caretaker is, says thricedotted. “The bots “speak” for themselves.”

Like other botsmiths I spoke to, thrice is ultimately pragmatic, and has no plans for “an estate that funds a server running into eternity where only the bot has the keys to the Twitter account.”

Jim Kang, a developer and artist who runs a number of bots, the most popular of which is @godtributes, had an idea I hadn’t heard before that would replicate the “passive consumption situation” in a post-Twitter world: create physical bots, in the form of signs or some other updating device. Kang came up with the idea based on a Boston-area liquor store chain called Sav-Mor that’s known for having funny sayings on their signs.

“Get a sign up in a publicly visible space, then have bots post to that,” Kang said. “Maybe they couldn’t respond to tweets, but they could talk about…local issues and weather?”

And really, as far as post-death robotic leavings go, what more can we hope for than one that talks about how nice it’ll be out.

This is part of our week-long series on the future of death.

Ethan Chiel is a reporter for Fusion, writing mostly about the internet and technology. You can (and should) email him at [email protected]