Example image

Image courtesy of Adam Sharp.

It is with mixed feelings and after a decade+ of downtime that I decided to revive istwitterdown.com.

I will not even try to summarize the drama that unfolded as Elon Musk tried to, tried not to and then was forced to purchase Twitter and take it private.

But I think things have gotten to the point that the site going down is a real possibility. Something that has not be likely for a long time.


This is the backstory.


Twitter launched in 2006 and grew like crazy (by 2006 standards). I knew of the site from day one, but waited a few months so sign up, giving me user id 19853.

I was working at technorati.com at the time (kids ask your parents), coincidentally one of the first consumers of the original Twitter firehose.

At this time, Twitter was notorious for regular outages, which featuring that infamous whale graphic chosen by Biz.


On 2007-11-15 Twitter went down and I got an idea.

Inspired by sites like islostarepeat.com (ask your parents) and thanks to a fast domain registrar, low engineering standards and an existing VPS, I conceived of, built, and launched istwitterdown.com during the time that Twitter was down (IIRC like 90 minutes).

My tweet announcing it is here. Note there are no native retweets or replies… because those didn’t exist on the platform at the time. (I can’t explain why there are few likes, then called favorites.)

The site was intended to be a joke - during outages it was common to hear a cubicle-mate lean over and and ask “is twitter down?” — therefore the implementation was a joke. The logic for checking was rudimentary (try to load the home page once a minute) and the infrastructure was not well maintained.

Those weaknesses spawned isistwitterdowndown.com and downforeveryoneorjustme.com, but it wasn’t worth maintaining as it was… a joke.

2009, etc

In March 2009 I joined Twitter as employee 33. It wasn’t until my first week of work that I realized that my coworkers who just hired me didn’t know I had built istwitterdown.com. I was worried. (I told them and they didn’t care.)

I spent much of that year working long hours doing hardcore but fulfilling work to fight the fail wail. Especially fixing a bunch of issues related to tweet storage, like twitpocalypse I and II, launching snowflake, trying to make cassandra a thing (and realizing how bad an idea that was). After leaving the core storage team I helped start, I started a team which became Platform Analytics and then later moved on to more organizational work, like founding Flight School, our engineering onboarding and education program.

At some point we made Twitter’s availablity pretty good. The fail whale was retired. The company was so big at that point that I didn’t even get one of the t-shirts they made of Moby Dick’s Captain Ahab standing on top a dead whale.

I let istwitterdown.com die. I didn’t give up the domain name because I didn’t want anyone to build something dumb with it. And who knows, maybe… I would need it again…someday.


It’s long been unlikely for twitter to experience the sort of easily detected outages that it had in 2007. In the past fifteen years, Twitter has been re-architected to be much more resilient than it was in 2007. It doesn’t really go down down.

Parts can fail, but the service overall can gracefully degrade. Of course that works better if there human operators to help it do so.

In the real ending of Moby Dick, Ahab doesn’t conquer the whale. Instead, his obsession leads him on a final, days-long chase. At the end, the whale drags him down into the depths of the ocean, entangled in his own harpoon.