By John Gruber
WorkOS: APIs to ship SSO, SCIM, FGA, and User Management in minutes. Check out their launch week.
I find Twitter to be both fun and useful, but, rather infamously, it’s also always been rather spotty with regard to uptime. The entire service has had a tendency to go down somewhat frequently ever since its inception.
Over the last week or so, however, I’ve noticed something different: the service has been up (for the most part), but individual tweets are seemingly being dropped. The main place I’ve noticed this is with “@replies” (a message sent to “@username” is treated by Twitter as a reply or reference to that particular person). I subscribe to an RSS feed of replies sent to me, which lets me see all “@gruber” replies, regardless whether I’m following the person who wrote it.
For the last week or so, some of these don’t seem to be getting to me. I tweeted about this a few hours ago and numerous people replied that they’ve noticed the same thing. Twitterrific developer Craig Hockenberry even has proof. [Update 1: Hockenberry also has confirmation that Twitter engineers are aware of the issue. Update 2: Screenshot evidence from Peter Hosey.]
Obviously, the ideal case is for Twitter to be able to handle all tweets all the time, and I’m sure that’s their goal. But in the meantime, I much preferred the old Twitter. The old way, you at least knew instantly what the deal was. Twitter was up, or Twitter was down. If it was up, you could pretty much count on your tweets working. Now, though, you can’t count on anything. Psychologically, I find it far better to have a big problem (the entire service being down) that everyone knows about than to have a smaller problem (occasional tweets silently disappearing into the ether) that, because of the very nature of the problem, you don’t even know about.
In short, making Twitter’s reliability indeterminate is worse, not better.