I started the account in response to a coworker in what must have been 2007 or 2008. He shot up from his chair, enraged by an email he’d just read that confused “your” and “you’re”. I was amused, and being someone who finds that sort of thing irritating, too (although nowhere near as irritating as my coworker finds it), I instantly came up with the concept of building a bot. I’ve long had a penchant for projects like this, large and small, that can quickly be built, and then built upon, as necessary. Dating back before the commercial Internet, I’ve always enjoyed stringing together public services and communication tools to do cool, if not necessarily useful stuff.

Within a half hour, I had it up and running. Instead of replying to the wrongdoer, it retweets and publicly shames them. The retweet contains the name of the recipient, allowing them to join the bot in poking fun at their friends. Many respond angrily to the corrections, typically for one of three reasons:  

1. They’re upset by being shamed in public
2. They think [Wrong!] refers to the sentiment being expressed rather than how it was expressed.
3. They believe the bot is a human being who spends his entire day criticizing people.

Of course, there are plenty of good-natured responses as well. 

Here are some recent examples: http://bit.ly/TXjKBw

The bot detects a very specific use case and does get it wrong sometimes. I’d update it if I could, but I no longer recall where it’s running. The bot is just out there in the world somewhere. To be honest, I’m kind of surprised it’s still functioning. I’m delighted that it’s doing its job, but I never imagined it would be approaching 300,000 corrections all these years later. I can’t imagine what that number would be if I hadn’t artificially limited the number of people it could call out in a day. 

I’m hardly following the AP Stylebook when I tweet, but I find it pretty easy to avoid the your/you’re mistake. That’s not to say that I don’t live in fear of eventually being corrected by my own creation.