Twitter set to release all tweets to scientists? Not exactly.


This Scientific American article with the headline Twitter to Release All Tweets to Scientists has been circulating around our favourite microblogging platform recently, creating a mixture of “could it be true?”-style excitement and alarm. The article picks up on a discussion of the ethics of using ‘big data’ from Twitter for research prompted by this ethical framework for Twitter research [pdf] published by a couple of epidemiologists, hinting that these issues (which are already very much alive in the internet research community, by the way) are about to explode as Twitter opens up access to its full archive of tweets to scientists everywhere. If this news were true, it certainly would be a game-changer.

Unfortunately, not only isn’t it true, but it isn’t news either. Twitter has made no such announcement recently. And as you can clearly see at the bottom of the piece, it was originally published under the title “Twitter opens its cage”; in added weirdness it was forward dated to 1 June although re-published in May. I don’t know what’s going on at Scientific American, but my interpretation (along with others) is that the original article was published in February as a secondary response to the announcement of the Twitter Data Grants program. The original announcement and call was in February, resulting in thousands of submissions, with the 6 winners announced in April. The data grants scheme – small-scale at least for now – may represent a trial for future scientific access, but it is also nice PR for Twitter’s recent acquisition of data retail company Gnip, which, like competitor DataSift profits from the commercially legitimated trade in access to ‘big social data’ from a range of social media platforms, but especially Twitter.

So while Twitter data is incredibly useful to a wide range of academic disciplines, it has become really hard to get access to it at scale without a lot of money or specialist infrastructure. As Twitter data has become monetizable, open API access to it has been choked, not only for marketers but for scientists and not-for-profits (e.g crisis mapping organisations) as well. I’ve previously written about this with my German colleague Cornelius Puschmann in our paper The Politics of Twitter Data.

So a more accurate title for the Scientific American piece would be: “In news that was new half a year ago, Twitter has selected a tiny number of scientific research teams to gain access on a limited basis to historical tweets via the recently acquired data reseller Gnip”. Not so exciting after all.