I sat bolt upright in bed at 3 AM this morning, turned to the framed picture of Richard Stallman I keep on my nightstand, and told him I finally figured out how to fix Twitter. It came to me in a dream. Jeff Goldblum and I were in some future cantina, bellied up to the chromium bar and talking cryptocurrency futures, when two bounty hunters broke down the door.
“Those men are out for blood,” I whispered to Jeff as I nonchalantly rested a hand on my trusty Klout (TM) Wokeblaster800.
“Not men,” he replied. “Only made to look like men. Check the badges.”
Clearly visible on each bounty hunter’s chest was a polished round badge: BOT. This is about where I woke up.
My solution is very simple, I explain to Richard. Twitter just needs to run a sort of inverse Verified program, wherein accounts that act like bots (or like human-assisted “cyborgs”) are labeled as such. A little silver badge would show up on their profile picture, plus maybe a line in their bio. Easy. And then the real magic would begin.
At this point I am pacing the room, portrait in hand. My wife, long since wise to this sort of thing, has earplugs in and is gently snoring, face to the far wall.
Once you’ve labeled the bots, you can begin to reprioritize humans in the network. One of the most effective changes would be to not include bots in the stat counts for a tweet’s likes, retweets, and replies. This would reward people who are interesting enough to garner actual human followers while depriving botnets their visibility. Viral tweets would actually reflect the demographics of Twitter.
In response, I imagine your average state-run botnet maintainer would have to get creative, both in how bots are registered and how they spread ideas. You’d probably want to monitor new accounts: do they suddenly change behavior after some boring reputation-building? Badge them!
And I imagine bots would have to move from mass-liking and -retweeting to a more grassroots pattern of replying to individuals, so give users controls around how they interact with bots: don’t notify them when a bot replies, and provide an option in a user’s settings to block all bots.
Richard Stallman nods.
There are two complications to these suggestions that I think are important to consider. (I sit on the edge of the bed and lower my voice slightly.) First, a robust appeals process would be necessary in case a real, actual human was categorized as a bot. I realize that this question is, like, a fundamental one in computer science, but Twitter would need to put a system in place that kindly helped users prove they are who they say they are. Luckily, they already sort of have such a system.
Twitter would also need to be careful not to ban away the other kind of bot — the sort that uses the APIs to (legally, and often beautifully) contribute to the network, either with teen angst or algorithmically-generated moths or just endless screaming. These accounts are valuable and part of what make Twitter special. And they can be drowned out by the botnets, just like humans.
This is just to say that, like all software, this change should be carefully, deliberately, and thoughtfully implemented. The result, I think, would be a social network that reflects who we really are.
Richard Stallman reaches out of the frame to place a finger on my lips. “We are not alone,” he says. My wife rolls over and is not my wife but Jack Dorsey. He takes the earplugs out of his ears.
“You’ve been incepted,” he says. “The seed of doubt has been sown. Now no one will believe the efficacy of this simple change to Twitter. We will continue to be peopled by bots, and our shareholders will rejoice at our numbers.”
I pinch myself but cannot wake. From everywhere at once I hear the macOS startup sound, unending, a sun that never ceases to rise. “This will make a great thread!” I shout, but no one seems to hear.