Twitter CEO promises downloadable tweet archive again

Want to revisit all the witty one liners you have ever tweeted? Twitter CEO Dick Costolo has promised yet again that archived tweets will be available for download in a few weeks' time.

In a public talk at Ford School of Public Policy and School of Information at the University of Michigan, Costolo confirmed that engineers were working on making a full archive of tweets available to users by the end of this year. He was there to discuss Twitter’s role in the future of global communication and democratised access to information.



Costolo had made this statement previously too. He first announced that this feature will be made available in an interview with The New York Times in July. During a keynote conversation at the Online News Association conference in San Francisco, he gave end of 2012 as the tentative deadline, taking into consideration the engineers’ capacity.

Flying over fuller to Facebook

Download your tweet archive soon


During the Ford School talk, Costolo went into deeper detail about why it has taken so long for Twitter to come up with this feature and why the 2012 deadline might actually seem to be too ambitious. Costolo explained that the team had been too busy trying to keep with the breakneck speed at which users and number of tweets on the site were growing. At 350 million tweets being posted per day, according to the CEO, the team hasn’t had time to build a solid archive retrieval feature.


Using Twitter's API, only the latest 3,200 tweets posted by any user are available at any given point of time. The tweets are arranged in the newest first order making it difficult to find a tweet you posted, say, a few months ago, especially if you happen to be a compulsive tweeter.

Costolo described the situation when an audience member asked him to explain why users ‘aren’t allowed to’ download their own tweet history. The CEO joked that this made it sound like he didn’t want them to. “It’s really horribly suited to archive search and archive distribution. So if you wanted to do a search against our user database, our user DB for that entire history, it would be so slow that it would slow down the rest of the real-time distribution of things.

He went on to say that they were working on creating a different kind of archival system for these tweets to enable users to download their search history. The current system works well with creating real-time search and distribution on the timelines at the rate of 15,000 tweets per hour, Costolo said. He elucidated this with the example of the night of the US Presidential elections, “There was a point at which we were serving 1.3 million timelines. A timeline is my home timeline of all the tweets of the people I’m following; 1.3 million timelines per second. So keep in mind that’s every second 1.3 million timelines going out that are threading together every single tweet that’s coming in from around the world at 15,000 tweets per second, and organizing them in chronological order.”

Costolo was very cautious while promising the end of year deadline, but joked that the engineers working on this feature were mad at him and didn’t necessarily agree that they’ll be done by the end of 2012.


You can read the transcript of the entire talk here.

Published Date: Nov 27, 2012 13:24 PM | Updated Date: Nov 27, 2012 13:24 PM