New APPS: Art, Politics, Philosophy, Science

Philip Agre on Privacy and Big Data (part 1)

By Gordon Hull

A current paper by Mireille Hildebrandt sent me to a paper from 1994 that I’m embarrassed to say I hadn’t read before: Philip Agre’s “Surveillance and Capture.”  Agre’s paper has been cited over 300 times, but it’s missing in a lot of the privacy literature I know.  After reading it, I’ve decided that’s a mistake, and it’s time to make amends.  I’ll begin by saying why I think Hildebrandt is exactly right to bring the paper up in a context of big data.

Agre starts with quotidian examples of tracking, things like employee ID cards that let systems know where they are, to UPS package tracking.  The core argument of Agre’s paper is that there’s two conceptual models of privacy that need to be distinguished in making sense of phenomena like these: one, the one that we all talk about all the time, is surveillance.  The surveillance model grew out of experience with state bureaucracies, particularly in the Soviet bloc, and features visual metaphors (typically Orwell and Bentham); assumes that the watching is nondisruptive and secret; involves territorial metaphors like invasion of space, which tend to then lead to a dichotomy between coercion and consent; involves centralized orchestration; and is identified with the state.  This model, though ubiquitous, isn’t the only or even the best one.  Agre spends most of his time developing the alternative model, though he notes that “when applied as the sole framework of computing and privacy, the surveillance model can lead to oversimplified analysis” and suggests that it lends itself to caricature and easy dismissals such that “genuinely worrisome developments can be seen as ‘not so bad’ simply for lacking the overt horrors of Orwell’s dystopia” (116).   It’s not hard to say that Agre got this one right: whether it’s the well-trodden problems with notice-and-consent or the stubborn persistence of the “I’ve got nothing to hide” deflection, the surveillance model isn’t adequate to privacy worries now.

The second model, which Agre ties to computer science and information systems management, is built around the idea of “capture.”  The capture model also has five characteristics, which Agre develops as a contrast with surveillance.  Capture: uses linguistic metaphors for human activities; assumes that linguistic parsing of activities is active intervention; defaults to structural metaphors; is decentralized; and is driven by to reconstruct human activity “through [its] assimilation to a transcendent (‘virtual’) order of mathematical formalism” (107).  The core concept here is grammar: the basic mechanism of capture is to pick a set of atomic elements (people, UPS packages) and to then develop a grammar that describes their possible movements or changes.  This model is then imposed – sometimes coercively – on the activities it purports to describe, mechanisms are used to then track and measure the activity according to the grammar; the system can then be further calibrated.

More on Facebook and Mental Health Privacy

By Gordon Hull

It’s not news that Facebook generates a lot of privacy concerns.  But it’s nonetheless worth keeping up a little, just to indicate how seriously we need to be concerned about the connection between Facebook and data analytics.  We’ve known for a while that automated analysis of Facebook likes can predict basic personality type.  A 2017 paper showed that targeting of advertising based on personality type was highly effective, as measured by clicks and purchases.  Facebook appears to be marketing predictive-advertising (so an advertiser can know to send you a nudge if FB software predicts you may be about to switch brands of something).

Now there’s this: an individual’s Facebook posts can apparently predict depression independently of subjective reporting (discussion here). From the study:

“Each year, 7–26% of the US population experiences depression, of whom only 13–49% receive minimally adequate treatment. By 2030, unipolar depressive disorders are predicted to be the leading cause of disability in high-income countries. The US Preventive Services Task Force recommends screening adults for depression in circumstances in which accurate diagnosis, treatment, and follow-up can be offered. These high rates of underdiagnosis and undertreatment suggest that existing procedures for screening and identifying depressed patients are inadequate. Novel methods are needed to identify and treat patients with depression.

By using Facebook language data from a sample of consenting patients who presented to a single emergency department, we built a method to predict the first documentation of a diagnosis of depression in the electronic medical record (EMR). Previous research has demonstrated the feasibility of using Twitter and Facebook language and activity data to predict depression, postpartum depression, suicidality, and posttraumatic stress disorder, relying on self-report of diagnoses on Twitter or the participants’ responses to screening surveys to establish participants’ mental health status. In contrast to this prior work relying on self-report, we established a depression diagnosis by using medical codes from an EMR.”

It seems to me that these paragraphs speak to why this technology can be so difficult to wrap one’s head around.  On the one hand, it is really important to take mental health seriously, and we live in a neoliberal society that both generates mental health problems and offers problematic band-aids to solve them, including moves by Pharma to drum up purchases of drugs with dubious clinical efficacy.  The study is classic big data: by looking at actual EMR records and correlated FB posts, the authors were able to train the algorithms with existing known cases of depression.  What it discovered was that certain words in posts predict a future diagnosis of depression pretty well, better as you get closer to the diagnosis.  And it doesn’t rely on self-reporting. Nor does it rely on a primary care physician spotting the problem, which the study suggests is part of why depression is so under-diagnosed.  As the authors put it, “the potential exists to develop burdenless indicators of mental illness that precede the medical documentation of depression (which may often be delayed) and which, as a result, could reduce the total extent of functional impairment experienced during the depressive episode.” There’s a chance here to do some real good for public health at a relatively low cost (though one should raise a digital divide worry: those who do not leave a sufficient social media trail – say, the homeless, who are at very high risk for mental health issues – will fall further through the cracks).

On the other hand, want to buy a list of depressed patients to sell those dubious remedies to?