Samaritans Radar should be opt-in only

The Samaritans Radar app has had a disastrous launch: by initially refusing to allow individuals to opt out of being monitored and by responding badly to criticisms, the Samaritans have created a lot of bad feeling. The app is now working at a very large scale: the Samaritans report that after day 1 it was monitoring 900,000 Twitter feeds. Unfortunately, although allowing opt-outs is a big improvement, the way the app is being handled is still a mess and many people clearly don’t find the app’s monitoring acceptable. I will therefore argue that Samaritans Radar should be made opt-in only.

Monitoring public spaces

As I’ve said, I think that launching Samaritans Radar with no opt-out was inexcusable. An uproar was generated when the Samaritans initially refused to allow individuals to opt out and treated the public spaces of Twitter[1] with what looked like distain: arguing that “All the data used in the app is public, so user privacy is not an issue”. While these online spaces are public, they are also valued by many and are covered by norms of acceptable behaviour. As Mark Brown notes, “For some, Twitter is the only place they have felt able to meet others with mental health difficulties and to be honest about their true feelings.” Online spaces are important.

The way the Samaritans behaved would be viewed as unacceptable in an offline public space.[2] Imagine, for example, that I went into a busy local park and told users I’ll be monitoring what they do, storing records and alerts about their activity whether or not they like it – after all, everything they do in a public place is public. When people argue back, I tell them to think about issues of privacy in parks. This would not make me popular. At best, I think I’d be told to go away.

Not surprisingly, the Samaritans have received a lot of negative responses to their app launch and lot of people clearly view Samaritans Radar’s monitoring of public space unacceptable. I wouldn’t feel comfortable monitoring a public space after many people from the community had told me to stop, and I don’t think it’s appropriate for Samaritans Radar to continue monitoring Twitter after having had such a negative response. Changing the app to work on an opt-in basis – so that only those who want to be monitored are monitored – would be much more acceptable.

Opt-in and opt-out observation

I think that opt-out observation of public spaces can be acceptable in some circumstances, and when I was planning this post yesterday I was going to argue for Samaritans Radar’s work to just be made more overt.[3] However, opt-out monitoring by Samaritans Radar isn’t acceptable now. When I was trying to defend the potentially benefits of an opt-out project – saying that an opt-in project would be smaller scale – @MLBrook pointed out that when I say smaller scale I “mean it might have included just those who actually consented”. This argument for an opt-in approach is persuasive for two reasons. Firstly, given the response the app’s launch has had, I don’t think that Samaritans can take the fact that users don’t opt out as implied consent for monitoring (indeed, some people object to the Samaritans storing their details on an opt-out list). Secondly, the project is scooping up a lot of sensitive personal data (it will be collecting more the longer it runs for) and this type of project needs very careful handling. However, as well as the problems at launch, the Samaritans Radar mess continues today. For example:

  • @Endless_Psych reports that, after they sent a tweet to trigger Samaritans Radar, “Someone got a notification about the tweet. The next day.”
  • The Samaritans have still not made clear whether opting out from Samaritans Radar will stop the app from monitoring my Twitter feed or stop the Samaritans from collecting, storing and analysing data from my feed. The Samaritans statement re the introduction of an opt-out for individuals says that the opt-out is for “individuals who would not like their Tweets to appear in Samaritans Radar alerts” but doesn’t offer further information about the impact of an opt-out on their data collection and processing. They have since stated that opting out does stop monitoring – and I appreciate the prompt response to my question – but this should have been made clear from the start.
  • There is no clear way to report users who are abusing the app or get them blocked.
  • Samaritans have claimed – repeatedly and incorrectly – that people who don’t follow them can DM to opt out.

Given how badly things have been handled so far, I think the best thing for both the Samaritans and those they are trying to help would be for Samaritans Radar to start out opt-in only and on a much smaller scale.

I’ve covered problems around the launch of Samaritans Radar here. I’m posting this after a long day – let me know if you spot any typos, if anything seems unclear or if I’ve missed out a link I should have included. I should also say that I think the Samaritans are generally a great organisation and I’m sure Samaritans Radar was launched with good intentions; this makes what has happened with the app all the more disappointing, though.

Update 31/10: @Endless_Psych has updated me that “they got the notification [from Samaritans Radar] about two or so hours after I posted but they were asleep”. I’ve also updated the post to add the point that it’s not clear whether opting out from Samaritans Radar will stop monitoring, data collection etc.

Update 2 31/10: @J5nnRussell has clarified that opting out stops Samaritans Radar from monitoring your tweets. Post updated to reflect this.

[1] Sorry, I’m a geographer – I like spatial metaphors. I also like footnotes.

[2] I’m not arguing that there’s a clear online/offline divide, but talking about online and offline spaces seemed the clearest way to express my point in this blog post.

[3] To be open about my own biases, I have previously argued for and carried out online observation which has been opt-out rather than opt in (although I have done this in very different ways and on a very different scale to Samaritans Radar).

Problems with Samaritans Radar

The Samaritans Radar app is an interesting – and potentially valuable – idea. However, the app relies on the covert monitoring of Twitter users and will probably be collecting and processing lots of sensitive personal data. There is also the potential for the app to be used to target people when they are vulnerable. I will argue that the covert nature of the app’s monitoring and the lack of any apparent way for people being monitored to opt out are both unacceptable and that the Samaritans have not evidenced adequate safeguards against abuse.[1]

The app is presented as “a chance to help friends who may need support”. Some users will no doubt use it in this way. What the app actually is, though, is a means to get alerts when certain words or phrases crop up in the tweets of people a Twitter user chooses to follow (as long as their accounts aren’t private and they haven’t blocked me).[2] Monitoring is not transparent to those who are monitored: Samaritans make clear that “Samaritans Radar is activated discreetly and all alerts are sent to you alone…The people you follow won’t know you’ve signed up to it”.[3] I can’t find any way to opt out of being monitored by the app – the decision about whether to use it is made solely by the user who is signing up for alerts about people they follow.

Samaritans argue that “All the data used in the app is public, so user privacy is not an issue. Samaritans Radar analyses the Tweets of the people you follow, which are public Tweets”.[4] This is a bad argument. By way of analogy, my office window looks out on a public street – whatever people do there is public. There would still, though, be privacy issues if I installed a video camera in my window to tape what people did outside; there would be bigger issues if, say, I allowed interested parties to subscribe to alerts when person X or Y walks past my window drunk. It would be even more worrying if person Y found out about this and was upset but I didn’t offer any way for them to stop me from monitoring them or sending alerts.

There is also a real risk of harm here. People might, for example, feel less able to share their feelings and seek support on Twitter if this brings them a raft of well-meaning but unwanted contacts from followers or if they felt they were being surveilled in an oppressive way. As @Sectioned_ points out, an alert from the app “could seem like open encouragement to platitude-bomb someone when they’re feeling rubbish”. More worryingly, abusive people might use the app in order to find out when a target of theirs is feeling lousy: as @claireOT argues, “there’s a worrying lack of safeguards against ppl using the app to target vulnerable ppl”.[5] Even if someone is aware that they’re being targeted in this way and wants to stop it, I can see no way to opt out from being monitored. I also can’t see any way to report someone who’s using this app for abuse or to get them blocked from using the app.

The Samaritan’s Radar app is a nice idea, but the lack of any clear way to opt out seems inexcusable – and increases the likelihood of the app doing harm. I haven’t seen evidence of adequate safeguards against abuse of the app. If the app is popular, its launch will mean the covert monitoring of many Twitter accounts along with the collection, analysis and storage of a lot of sensitive personal data. It might be possible to justify this – and I’m sure the Samaritans have good intentions – but I haven’t seen anything like an adequate justification from the Samaritans.

UPDATE: Samaritans Radar is now covertly monitoring (or “supporting”, as they put it) 900,000 Twitter feeds. This is a large-scale monitoring, data collection and processing project, and really does need to have appropriate privacy and risk mitigation measures in place.

UPDATE 2: the Information Rights and Wrongs blog now has an excellent post on data privacy issues around Samaritans Radar. I now probably won’t write a post on data protection and the app – I don’t think I could do anything better.

Update 3 (30/10/14): Samaritans have announced that they will allow individuals to opt out from being monitored by the app. I have added strikethroughs to the post to reflect this.

I’ve tried to acknowledge sources here, but I may well have missed people making similar points about the app on social media. Please tell me, and I’ll add in appropriate links.

 I’ve kept this post brief-ish, but I also have a half-written post about data protection aspects of this an another looking at how issues like this are dealt with from the point of view of research ethics (I submitted an ethics form for some online ethnographic work not that long ago). I’ll try to write these up at some point – so there’s the excitement of discussions of data protection and research ethics still to come! I’d also like to write something about how this type of app might work in a more ethical, and less intrusive, way.


[1] I appreciate that people can leave Twitter or make their accounts private. However, people should not be forced to make their Twitter account less public in order to escape this type of monitoring.

[2] Though I imagine I could set up an anon sockpuppet account to follow anyone who blocked me and I still wanted to monitor.

[3] Clearly, some of those using the app may tell those they follow that they are doing so and some Twitter users may actually ask to be monitored. However, the app does not tell people that it is monitoring them.

[4] It is worth noting that much of the data collected will fall under the Data Protection Act’s definition of personal data – for example, I’m easily identifiable from my Twitter handle @JonMendel.

[5] I appreciate that an abuser can also just read a public Twitter feed, but this app is potentially making this far easier.

Should we care if the NHS is number 1? Questions about the Commonwealth Fund Scorecard

I’ve been hearing a lot of claims recently that the NHS has (as the Guardian puts it) been rated as “the world’s best healthcare system”. These stories draw on a recent Commonwealth Fund report. However, I don’t think the question of whether the NHS is number 1 in the world is a good thing to focus on. I’m also not convinced that the Commonwealth Fund report is a good way of measuring NHS performance. Read More…

Horizon’s Longitude Prize episode: overselling the benefits of GM crops and fish oil

Horizon’s recent Longitude Prize episode discussed interesting research using Genetically Modified (GM) crops to make omega 3 fish oil (not found naturally in plants). Horizon stated that fish oil is “thought to protect against…cancer”. This surprised me, so I asked Longitude Prize for the source of Horizon’s fish oil claims. The sources they provided aren’t adequate to support the claims made – Horizon oversold the benefits of fish oil and of GM crops that produce it. Read More…

UKIP cite Europol statement about Romanian gangs as evidence of Romanian criminality. The irony.

UKIP leader Nigel Farage arguesh that people have a right to be concerned if a group of Romanians move in next door. His reason for this is that “Police figures are quite clear that there is a high level of criminality within the Romanian community in Britain.” UKIP cites four tabloid articles (and no other sources) to back up this claim. I’ll take a look at one of them: a Standard article on gangs.

The article is reporting Europol worries about Romanian gangs. Clearly, this does not in itself make it reasonable to be ‘concerned’ about Romanians moving into one’s neighbourhood – even assuming Europol’s worries are well-founded, the great majority of Romanians are not in gangs. It is touching, though, that UKIP now have such faith in a European Union agency like Europol – I’d argue that it’s important to critically assess the arguments made by EU agencies, but maybe UKIP are more in favour of blind trust?

More interesting, though, is that in the same article Europol state their belief that “the threat from Romanian gangs will not rise significantly when restrictions on the type of work Romanians and Bulgarians can do in the UK are lifted in January” 2014. I wonder if UKIP also find Europol convincing on this point?

UKIP have an exclusionary and reactionary idea of the local

I’ve lived in Dundee for just under five years. I work in the city, really like it here, and see myself staying here for many years. I was therefore surprised that, according to UKIP councillor Suzanne Evans, I wouldn’t count as ‘local’. In last week’s Westminster Hour*, Evans argues for prioritising services for “local people”. Evans doesn’t seem to view just living and working in an area (even for several years) as enough to be ‘local’; instead, she states that UKIP’s idea of ‘local’ would require “specific roots to an area…having been born in an area, having parents or grandparents there…we think that’s only fair.”

According to this logic, I could never be a local in Scotland. First generation immigrants to the UK – even people who have been here longer than I’ve been alive – could never be a local anywhere in this country.** UKIP have an incredibly – and worryingly – exclusionary idea of what it means to be local.

UKIP have recently been complaining about being called fascist. I don’t believe that they are fascist – the term implies rather more organisation and centralisation than UKIP have been able to achieve. However, this reactionary and exclusionary idea of what it means to be local puts them in an unfortunate political position – their position on this is actually more regressive than some far-right parties which prioritise the nation over the local (and are therefore much more positive about movement within the nation).

What’s interesting here is that UKIP seem to be constructing a place-based, very exclusionary type of politics which is in some ways making them less nationalist – emphasising the local over the national. Definitely not fascist, though, so that’s a relief!

* listen from about 25mins in. I had to listen several times to convince myself that Evans was really making this argument.
** Unless, perhaps, they brought parents/grandparents to the UK with them. However, I am not sure that this is a goal of UKIP policy.

Science Translational Medicine and a non-correction correction

In response to the concerns I raised, Science Translational Medicine have added a competing interests declaration to Ma et al.’s article on Vitamin C and cancer. This article now states that

JD holds an unpaid advisory board position for a grassroots advocacy group, the Alliance for Natural Health-USA. She had previously served as a board member for the American College for Advancement in Medicine. QC holds an unpaid position on an advisory board for the International Society of Integrative Medicine. None of these organizations are connected to the supplement industry

This correction is, however, not flagged as a correction at all. It’s also only visible once you have got through the journal’s paywall. Given the huge publicity that this research got while it was missing this declaration, this is strikingly inadequate.

I will be contacting Science Translational Medicine to ask what’s going on with this, and also query some other aspects of this declaration. However, I wanted to put this up so this statement of competing interests can be accessed outside of the journal’s paywall and so there’s a public note of this correction – both things that, as far as I can tell, Science Translational Medicine have failed to achieve.

Science Translational Medicine, Vitamin C and cancer: a potential CoI and a lack of transparency

The recent Ma et al. Science Translational Medicine article on “High-Dose Parenteral Ascorbate Enhanced Chemosensitivity of Ovarian Cancer and Reduced Toxicity of Chemotherapy” has received a lot of attention recently – for example, the BBC initially reported it as suggesting “Vitamin C keeps cancer at bay”.* However, Jeanne Drisko (one of the corresponding authors) has a potential Conflict of Interest (CoI) which was not noted in the article. Science Translational Medicine (STM) will be correcting what they view as an “oversight” after I e-mailed to let them know about this, but their response has generally been disappointing: they have not been nearly transparent enough in handling this potential CoI’s omission from the published article.

Ma et al. (2014: 10) states that “The authors declare that they have no competing interests.” Drisko, though, is listed as Chairman of the Alliance for Natural Health USA (ANH-USA): a campaigning organisation that describes itself as “dedicated to promoting integrative medicine and sustainable health”.** Intravenous vitamin C is, as the Kansas University press release promoting this article notes, “currently administered intravenously to thousands of patients by practitioners of complementary and alternative medicine”. STM’s Conflict of Interest policy asks that authors declare “all affiliations…or management relationships related to the reported research, including those that could be perceived as potential sources of bias”. When an author has a senior role in an organisation campaigning on issues related to the article, this could reasonably be viewed as a potential source of bias and should therefore be declared.

I don’t know whether this potential Conflict of Interest (CoI) was declared to STM when the article was submitted by the authors or accepted for publication: the journal refused to tell me, arguing that

the specific content of CoI forms is confidential

This is a long way from an open CoI policy!  What STM did tell me is that

We have looked into this issue and discovered that the potential conflict of interest you highlighted (which is entirely non-commercial and non-financial) was inadvertently left off due to a misunderstanding. We are now working to correct this accidental oversight.

I am pleased that STM are working to correct this, but this lack of transparency about whether the potential conflict was declared is unfortunate.*** STM also declined to clarify whether their interpretation of the journal’s CoI policy is the same as mine, arguing that

Although I could provide an interpretation of the statement on the website, I don’t see how that would be fair to the authors. The authors do not normally get customized interpretations from journal editors, but instead have to make their own interpretations. You now have exactly as much information as the authors did when working on their forms.

Alongside STM’s disappointing response, it’s rather a shame that it’s left to me to publicise this potential CoI – after the end of most of the media puff around the piece.  Given the article’s topic, STM and the journalists covering the story should have realised its potential to have high impact outside of academia – intravenous and oral vitamin C are both heavily marketed to cancer patients and other vulnerable groups.  If there’s ever a topic to take extra care over, this is it.  Whether or not the article’s authors declared this potential CoI to STM, Drisko is quite open about it elsewhere – her position as Chairman of ANH-USA is noted on the organisation’s website, so it doesn’t exactly take a courageous feat of investigative journalism to find out about this role.  Despite a lot of media coverage of the article (Google News lists more than 300 pieces) none of the journalists writing about this topic appears to have noticed the potential CoI.

It is important for potential CoIs to be declared, so that readers can keep them in mind when assessing published work. Even if STM now makes some minor changes to the article, the damage has already been done by the media hype around this article’s launch: vitamin C has been widely publicised as a useful cancer treatment (although the Ma et al. article did not find any statistically significant difference in survival between patients treated with vitamin C and those receiving standard treatment). This STM article and the media coverage that accompanied it will be extremely useful to those campaigning for – and selling – the type of ‘integrative medicine’ that ANH-USA promotes.

* The publicity around the article often overstated its implications for cancer treatment in humans: NHS choices has done a nice job of summarising some limitations of the research and problems with this publicity.

** I learnt about Drisko’s role in ANH-USA through reading David Gorski’s critique of the article

*** Oddly, though, STM seem to view some financial details of this potential CoI – that it was non-financial – as not confidential: at least, they shared this information with me.

Postscript: my own potential conflicts of interest

Given the topic of this post, it seems reasonable to be open about my own potential CoIs. I have a general statement of interests here.  Specific to this topic, I should also note that I have been involved in a campaign for more open declarations of UK doctors’ interests (this has been unpaid, although I have applied for funding for some related work). I have also done some voluntary work for one of the UK’s cancer charities. In terms of my personal views, I strongly disagree with most of the positions taken by ANH-USA (but would also defend the right of academics to engage in campaigning work, and have been involved in this type of work myself).

Trials, Lisa Bonchek Adams and Bill and Emma Keller

Emma Keller and Bill Keller both recently posted op ed pieces about Lisa Bonchek Adams – for the Guardian and NYT. Adams is a mom living with Stage 4 breast cancer, who has been tweeting and blogging about it – writing beautifully about things like trials and surgical procedures, and getting lots of followers as a result. Both op eds were crass and offensive – some of the reasons why have been discussed on The Nation, The Atlantic, Medium and on Twitter, and the Guardian article has been taken down “pending investigation”. One unpleasant aspect of Bill Keller’s piece that hasn’t had enough comment, though, is his passive aggressive snark about Adams’ choice to participate in clinical trials. Bill Keller seems to view volunteering for trials – something which can have really altruistic effects – as something it is negative for Adams to write about. However, we should be grateful to those who volunteer to participate in trials – future patients can and do benefit from their altruism. I wouldn’t be here to write this if it wasn’t for participants in past trials in different conditions.

Bill Keller’s article states that

There is always the prospect of another research trial to excite [Adams’] hopes

as if this is intrinsically a bad thing. Without knowing the detail of the trials Adams participated in, Bill Keller writes that

Whether or not this excellent care has added months or years to [Adams’] life, as she clearly believes, is a medical judgment, and her doctors, bound by privacy rules, won’t say. Most trials of new drugs aim to determine safety and calibrate dosages, and make no promise of slowing the disease in the participants.

Bill Keller appears to be complaining that Adams

has been an eager research subject…Adams has been a cheerleader for cancer research in general and Memorial Sloan-Kettering in particular.

Clearly, many patients choose not to participate in clinical trials – as is absolutely their right, and a decision that should very much be respected. I don’t know what decisions I might make myself if I became seriously ill in future and am given the option of trial participation. However, those who do participate in trials are doing a public good – what is learnt on these trials can improve or prolong the lives of others, whether or not it helps the trial participants. Indeed, I’d almost certainly have died several times by now if it hadn’t been for volunteers for previous clinical trials (on various different diseases); the same can be said of a number of people I care about. Snarking about a cancer patient’s use of social media is pretty low in itself, but writing in a way that makes trial participation seem like a bad thing for patients to do or to write about is particularly unpleasant.

Additional Conflict of Interest: I volunteered to participate in one trial myself, which stood almost no chance of benefiting me personally but will hopefully help the NHS and future patients a little (in case that seems particularly noble, it also involved pretty low risk to me and very little hassle!) I’m not going to say what the trial was – ironically, given the context, doing so would tell the internet more than I want to about my previous health…

The Nudge Unit should publish more of their research – now

The Cabinet Office Behavioural Insights Team (or Nudge Unit) has done lots of interesting work – in particular, around using Randomised Controlled Trials in order to guide public policy. Some of their reported results – for example, when looking at getting people to pay tax more promptly – have been really impressive. I therefore submitted a Freedom of Information Act request to find out more detail about their research – in order to properly interpret research findings, it is important to be able to look at what was done and assess the strengths and weaknesses of the work. Sadly, the request was refused and, when I asked the Cabinet Office to review this response, they upheld this refusal.

The Cabinet Office gave two reasons for refusing my request:

  1. The Cabinet Office stated that some information on Nudge Unit research is already published. I’ve no arguments with this – I’m not asking the Cabinet office to republish what they’ve already published.
  2. The Cabinet Office argue that “It is the intention of the team to continue to publish the findings from its work on its website and, where appropriate, to publish findings in academic journals…academic publication can take some time, and it is important that the publication is timed so that it is released into the public domain as a coherent, consistent and accurate body of work”. I am not happy with this justification, as I’ll discuss in more detail below.

I’m pleased that the Nudge Unit is pro-actively publishing findings from its work on its website and is planning to publish in academic journals (which I entirely understand can take some time). However, I don’t think this is an adequate reason not to publish more information right now because:

  • Nudge Unit plans to publish “findings” in future aren’t sufficient – one needs detail about how the research was done in order to know how to interpret the findings. No research is perfect, so it’s important to be able to look at how the research was done rather than just trusting in the findings.
  • Researchers can publish some details of their trials (for example, trial protocol) prior to publication of the trial in an academic journal.  This is seen as good practice with clinical trials, and there are moves by some social science journals to insist on pre-registration of the trials they publish. I fail to see how publishing detail of trial methodology, hypotheses being tested etc. would prevent subsequent academic publications.
  • It’s not uncommon for policy research to be published in one format (for example, a policy report) and then subsequently written up as journal articles. While academic journals won’t want to run something that is identical to something already published elsewhere, the differing publication formats and audiences mean this often isn’t too hard to work around.
  • The Behavioural Insights Team makes bold claims for what some of its research has found – claims with major policy significance, and which seem to suggest adaptations to Government practice – without providing the information needed to critically assess these claims.  For example, in February 2012 the Nudge Unit argued that “the team has been able to demonstrate effects that, if rolled out, could save hundreds of millions of pounds.” That would be a dramatic achievement, and the paper discussing this research is an interesting read. However, the short summaries of trials given in the paper don’t give enough detail for one to judge how reliable they are. These bold claims were made more than 18 months ago and there is, still, not enough publicly available information for them to be properly assessed. If issues of coherence and accuracy are important to the Cabinet Office and they really feel they can’t publish quickly in more detail, the better approach would be to hold off on making claims until a fuller publication is possible – making the bold claims without a fuller publication is not helpful.

I now need to decide what to do with this Freedom of Information Act Request – I may appeal to the Information Commissioner’s Office. It may be that some Nudge Unit research will be out as academic publications shortly – in which case, it could make sense to wait – but this seems very unlikely to be the case across all their research. I have been disappointed by the lack of a more open response from the Cabinet Office.