UK suicide-prevention charity Samaritans checks Twitter timelines for worrying posts

The Samaritans, a charity devoted to preventing suicides, has launched a service that plugs into people’s Twitter accounts, then notifies them if anyone they follow is sounding like they might become suicidal.

The Samaritans Radar web app appeared on Wednesday, promising to “turn your social net into a safety net.” Once authorized to view the user’s timeline, the system looks for phrases such as “help me” and “tired of being alone,” then sends an email to the subscriber so they can offer support to the person typing such phrases.

“Samaritans Radar is in its infancy and won’t get it right every time — it’s not good at sarcasm or jokes yet! But there’s a way for you to give feedback on whether a Samaritans Radar alert was correct, so the service improves for everyone as it learns more,” the website reads.

I think this is a great idea, in theory at least. As people gather round the virtual rather than physical water cooler, it’s easy to miss the sort of social cues that might suggest a friend or colleague is having a rough time. And, if you follow a lot of people on [company]Twitter[/company], potential cries for help might get lost in the stream without some kind of flagging aid.

However, many people have a big problem with what the Samaritans are doing here — mainly because some people just use Twitter to get stuff off their chest, without the expectation that others will try rushing to their aid. Here’s a sample of the negative reactions on Twitter (there are plenty of positive ones too):

https://twitter.com/textuallimits/status/527456606565720064


https://twitter.com/gbelljnr/status/527467810398941185

A lot of this comes down to whether Twitter’s “public space” really is fully public, or whether people should expect to be able to hold semi-private conversations on it too. My colleague Mathew Ingram has written about this a couple times, and I must confess I haven’t made up my mind on the subject. Yes, Twitter is like a public space, but in a physical public space people can still hold discussions within a closed circle; a kind of expectation of obscurity, perhaps, if not outright privacy.

A spokeswoman for the Samaritans stressed to me that people opt into using the service — however, that obviously doesn’t apply to the people whose tweets are being flagged. “It’s simply highlighting a tweet that the subscriber may choose to respond to. It’s not like Samaritans is intervening in private discussions as such,” she added.

Incidentally, she also told me that the Samaritans briefly keep a record of the tweets that their algorithms flag up, but that the subscriber chooses not to act upon, so that the systems can better learn what is and what isn’t a cry for help. However, the data isn’t shared with any third party.

The Samaritans already work with Twitter and Facebook as the social networks’ U.K. and Ireland suicide prevention partner – if someone reports someone else’s post as being of concern, Samaritans contact details are shared in those countries.

This article was updated at 7.50am PT to include discussion of the criticism of Samaritans Radar.