Tag: big data

Trying out Samaritans Radar

I’ve (temporarily) signed up for the Samaritans Radar app. I thought this would be a smartphone app, but it’s not. It’s a little twitter app that reads posts looking for keywords and sends an email if it sees anything it thinks might be alarming.
I’ve got two notifications so far. The first one comes from a tweet from before I added the app:

Oh, irony is not dead.
The other notification was for this tweet:

It sent me an email about 20 minutes after that was first posted, which is a bit late for a ‘I’m going to kill myself … right now’, but possibly ok for people building up more slowly. (A later notification for a different tweet arrived after only 7 minutes.) Gmail autofiltered the email to my Updates tab, which means it did not generate an alert on my phone. All of the updates I got have been in a single thread, so I could move it to the primary inbox if I were concerned.
The email told me:

Dear @celesteh,

Radar has spotted a tweet from someone you follow who may be going through a tough time.

Read the tweet here and find out how you can offer support.

View Tweet Now

If you need more information on how to help your friend or if you need some support yourself, please visit www.Samaritans.org

Clicking through on the ‘View tweet now’ takes me to a web page.

My Radar

Hi @celesteh
This is the tweet that Radar has spotted

Are you worried about this Tweet?
[yes] [no]

Clicking yes gets a small pop up window overlaying the tweet in question:

What to do:

You’ve indicated that the tweet might be worrying. Here are some tips on what to do next:

  • Have a look at their tweets, is this a one off? Out of character? Has it been going on for long?
  • Try sending them a tweet (or perhaps a DM, Email, Text?) gently asking how they’re doing.
  • Could a mutual friend have the same worries and help approach them together?

You might want to try and meet up with your friend, or arrange a time to chat on the phone?

  • Offer Samaritans contact details and suggest they call if they want to speak to someone anonymously : 08457 90 90 90
  • For more information and help with knowing what to say visit www.samaritans.org/radar

And that’s it. None of the pages had any queries or information relating to the nature of different relationships I might have with the potentially troubled tweeter. For example, if I got a tweet like that about an underling that I line-manage or have power over at work, that is not addressed. Nor is any mention made of the equalities act.
This tool does not give me access to any information I did not already have access to. It only shows me tweets already visible in my timeline. Therefore, concerns that it could out somebody as trans or LGB do seem misplaced, as it only shows me what people have already decided I should be able to see. I don’t know, but I suspect it will also work with protected tweets, so it will send me alerts about tweets that aren’t public, but, again, the person decided I could already see them. (This does raise larger concerns about how twitter apps run by big data companies are a way of circumventing privacy controls set by individual users.)
Again, this app draws my attention to tweets that would have been in my timeline anyway, but that I may have missed because of the massive volume of tweets that go by, or I may have scrolled past it without noticing. I can see why the Samaritans were surprised by people getting upset about privacy concerns, as this is much less invasive than is all of facebook or a lot of other twitter-based data gathering.

Previously on this topic: Posting to the Internet

Posting to the internet

I’m seeing a lot of people upset over an app made by the Samaritans. This is a twitter app that listens for some keywords and phrases that might indicate if one of your contacts is having a hard time. If they are, it gives you some nudges to reach out to offer them some support. This is obviously well-intentioned, but many are concerned about potential misuse, for example, by employers who want to discriminate against people, or bullies.
I would think the vast majority of people installing an app like this have good intentions. If your manager sees that you’re having a rough time via this app, well, they’re already following you on twitter and could very easily stumble across your unhappy posts by accident. Having this app indicates that they at least like the idea of supporting people, but, yeah, they could very well make poor decisions about what to do with a struggling underling. It’s not clear whether this app makes that any more likely. It does call attention to possibly unhappy tweets by the employee, but hopefully a well-intentioned manager is aware of the Equalities Act (and hopefully this scenario occurred to the programmers, in the user interaction between the app and the concerned person).
The bully argument is a bit more troubling, as, alas, it’s all too easy to imagine someone installing this just to keep tabs on varied target lists. This linked post is concerned about transphobic abuse, which is all too common, but its also easy to imagine this being exploited by GamerGate. In the hands of sociopath, this use is, indeed, troubling. However, it is merely automatic a task that they’re possibly already doing manually.
I think people’s major concern is the inherent creepiness of a piece of software following you and attempting to diagnose your mood and mental health problems via your public posts. This concern is not misplaced, but the targeting the Samaritans probably is. The truth is, untold numbers of apps do this already. We don’t know how many, because they’re under no obligation to tell us. Twitter makes available via their API a raw stream of everything posted to twitter. Every single post in real time. You or I cannot make use of so many tweets blasting out at us, but Big Data can.
Right now, there are programs running, looking at twitter, scanning every single tweet. They are looking for patterns. Some of this is completely benign. In the relatively recent earthquake in Napa, California, analysts were able to use tweet times and geolocation to figure out where the quake was big enough to wake people up. A lot of people, after making sure they were safe, reached for their smart devices and tweeted they’d felt the earthquake. Researchers were able to make a very nifty map, showing where it woke up the most people to where it seemed to waken virtually no one. That’s nifty, right? But it does require doing a lot of looking at where people are and whether they would normally be tweeting at that time of night. It’s not an invasion of privacy, because it’s public and maybe it doesn’t seem squeamish at all, but this is an unusual example.
Many twitter applications are looking for mentions of brands and what emotive words are used next to them. Many years ago, I tried to get a job blogging for a brand and they mailed me in the information pack every single public live journal post that mentioned their brand. All of them were public, but many of them were kind of embarrassingly personal. The authors never imagined them being disseminated to corporate types. This same thing still goes on, but a much more massive scale.
And its not unheard of to try to trace individuals. Target’s datamining certainly includes twitter.
We’re fine with giant corporations, security agencies and, yes, our bosses knowing everything we tweet. How do I know we’re fine with it? We all use twitter and nobody objects. Until some well-meaning do-gooder tries to make an accessible, helpful version of this.
Twitter’s entire business model relies on sharing this kind of information. This is not a case where they could be reformed into better behaviour. If you don’t want to be mined and you don’t want to tweet anonymously, consider using a privacy aware service to make posts like this – one that will share your tweets only with the people you say you want them shared with. So if you’re having a really really rubbish day and post, ‘I feel so down today, I’m at a loss and just don’t know how long I can carry on’ it will be seen by the people you want it to be seen by and not everyone on the entire internet and not a corporation out to sell your moods and data to anyone who will buy it.
If you want privacy control for posts, use Disapora for them. If you want the world and everyone to know what you’re thinking, use Twitter. There are times were it makes sense to use either. But even if this app gets withdrawn with muttered apologies, don’t assume you know who is reading your tweets. They’re public and anybody could be – and certainly is – gathering them.