Preston Richey


Meerkat, the now-defunct live-streaming app predating Twitter’s own Periscope by only a few precious months, was released to the general public in early March 2015. The app, in what would eventually become its downfall, lacked its own infrastructure, so Meerkat streams were necessarily distributed through one’s own Twitter feed. Yet there was no way to archive a stream for later viewing (as Periscope does now), so clicking on a link (which was accompanied with Meerkat’s automated stream description, “|LIVE NOW|”) often took you to a ‘stream over’ landing page, since most streams, especially early on, were live for just a minute or two.

This frustrated me. It frustrated others, as well. John Constine, in his March 1st launch article for TechCrunch, wrote:

“The only real downside to the reliance on Twitter is that Meerkat links live forever in tweets, even through the streams they lead to may be over and gone in a few minutes. That’s one thing Twitter could improve on if it built or bought something like Meerkat.”

So, I set out to do something about it. The idea I came up with was Roadkill Collector (@roadkillbot), my first real Twitterbot. The idea was simple: watch for tweets with Meerkat stream links, check if the stream had ended, and if so, tweet at the user asking them to delete the stream tweet (in retrospect, the copy I used could have been a bit less cheeky and more direct). Twitter users could also follow the account and be able to see the bot’s reply to the original Meerkat stream inline (since with Twitter you’ll only see @reply’s to users you also follow), saving them the time of clicking on the stream.

The bot ran for just 2 days, March 22nd & 23rd, and replied to 759 individual tweets. During this timeframe, @VH1 tweeted a link to a Meerkat stream of a Q&A with Shawn Mendes which was retweeted over 1,300 times and received nearly 300 replies. Due to some faulty logic in my script, the bot saw replies to a dead Meerkat link as individual tweets that themselves contained dead links, so it replied to lots of tweets mistakenly. Lots of users were also either perturbed by the bot, or just confused. I added a default response to any users who tweeted back at the bot, informing users that it was an automated service and that they could yell at @prestonrichey (me) if they wanted to.

I disabled the service because I started to question the ethics of such an automated, naggy (and somewhat passive-agressive) service. Building the bot was a fun exercise, but in the end I figured it should be up to Twitter users if they wanted to delete Tweets to dead Meerkat streams (even if I still believed that’s what should be done). Twitter has since added a clause in their Terms of Service that prevents bots from reaching out to users, unprovoked, something I learned later on when building @menemnesa.