I first heard about SamaritansRadar yesterday (October 29th) – that it was an app, and that its function was to help identify someone, a friend, who was feeling vulnerable. Then I began to hear more concerning comments about the way in which it was being applied and how the algorithm underpinning it could not manage sarcasm. Funny? Not really; I thought it probably couldn’t handle excerpts from professional papers or books, discussion amongst professionals, for instance, about suicide and depression, and also, more amusingly, the tweets my fiction writing friends often put out to publicise their books or stories. I decided to try it out, to register and see what the process was and what filters it allowed me to apply. But there were none, it was a simple activation process that then embedded the app in my twitter account and set about monitoring all the people I follow. I know very few of them in any other context and the lack of constraints, consent, or any kind of privacy protection was alarming. I revoked its permission straight away and emailed the Samaritans with my concerns. A good idea poorly implemented? Maybe, but let’s please not drive twitter users away by well-meaning but covert surveillance.
UPDATE: 31 October
It appears Samaritans have silently tweaked their FAQs (so the text near the foot of this post no longer appears). They now say tweets will only be retained by the app for seven (as opposed to thirty) days, and have removed the words saying the app will retain a “Count of flags against a Twitter Users Friends ID”. Joe Ferns said on Twitter that the inclusion of this in the original FAQs was “a throw back to a stage of the development where that was being considered”. Samaritans also say “The only people who will be able to see the alerts, and the tweets flagged in them, are followers who would have received these Tweets in their current feed already”, but this does not absolve them of their data controller status: a controller does not need to access data in order to determine the means by which and the manner in…
View original post 1,302 more words