Researchers wrestle with a privacy problem
nature.com
nature.com
The value that we know as "privacy" is far more arbitrary and historically contingent than most of us realize. In societies where resource scarcity forces people to share bedrooms, the acceptable spheres of privacy are radically different from the ones that modern American life inculcates.
Are other social boundaries equally flexible? And if so, should we test them?
Perhaps so. There is an intrinsic tension between the pursuit of privacy and the hacker ethos that information wants to be free. By definition, privacy is an act of sacrifice: a denial of information to other nodes in the network that forms civilization. Our laws already recognize narrow circumstances in which the costs of that sacrifice outweigh the benefits, such as when there is probable cause to believe that an individual is abusing his or her right to privacy in order to undertake criminal activity.
Yet there is a social taboo attached to radical opposition to privacy. For example, an integrated network of ubiquitous surveillance could offer incredible benefits to humanity. Violent crime could be virtually eradicated. People could be held accountable for anti-social behaviors, like littering or cutting off other drivers. Our entire society could operate more efficiently in innumerable ways. But open advocacy of a surveillance state is only a few degrees away from fascism according to our modern political mores.
In the example from the story, it's clear that our collective welfare could be increased by analyzing reams of restricted data. Are we willing to intentionally tolerate greater human suffering in order to preserve the amorphous right to privacy?
I contend that it's more useful to look at the discrete harms that flow from disclosure of specific information. How could people be harmed? What kinds of safeguards could be put in place to prevent that concrete harm?
Those are useful questions, which allow us to make informed choices instead of switching off our brains and discussing "privacy" in the abstract.
The same folks who are in favor of unlimited surveillance of the public tend to be opposed to public access to their own back-room decisionmaking process, and in favor of censoring embarrassing information in the name of “security” interests.
The “hacker ethos that information wants to be free” is going to run up hard against majoritarian attempts to scapegoat, shame, change, or destroy the different or weird. The life in a small village full of nosy neighbors which you mention was normal in the past was also a life of social conformity and harsh punishment of nonconformists. There’s no a priori reason to believe it would be different in a higher-tech society.
“Violent crime could be virtually eradicated”, except in the case of police corruption, which is in many places endemic, and which ubiquitous surveillance makes only more alluring.
Right now, we seem trapped in the worst of all worlds. The government engages in mass surveillance, so a rogue actor within the state could exploit the accumulated data for the unlawful purposes that you mentioned. And private corporations have access to almost as much information, with equal opportunities to abuse it but fewer democratic checks on their power. At the same time, we forgo virtually all of the potential societal benefits of data aggregation.
If surveillance is justified to prevent terrorism (and I realize that most here would dispute that), why is it not appropriate to use the same data to protect people from risks that claim thousands more lives every year?
I don't think that ubiquitous surveillance necessarily makes police corruption more alluring. One of the most recent policy responses to police misconduct has been the nationwide push for body cameras on individual officers, which is itself a form of surveillance. But it's one that obviously increases accountability, so it doesn't have all of the usual stigmas.
When placed within a proper governing structure, I see no intrinsic reason that other forms of surveillance could not be enlisted for the same purpose. Wouldn't it be better if we had actual surveillance video of what happened in the Trayvon Martin or Michael Brown cases? Do you believe that constantly being monitored in that fashion would increase police abuses of power?
In a broader sense, most of the responses to my comment illustrate the very phenomenon that I'm describing: i.e., an instinctive defense of "privacy" in the abstract instead of a detailed analysis of the costs and benefits of increasing transparency in specific circumstances.
I'm not faulting the authors. After all, I have raised this issue in a very broad sense, and so it's only natural that they would reply in the same fashion. But the way that we seem to be talking past one another illustrates how difficult it is to have meaningful discussions about "privacy."
If I could make three major points, they would be:
1. The need to acknowledge the futility of restricting access to certain forms of data
Most privacy advocates cite benefits that do not correspond with the facts. For example, the government already has the power to tail individuals in public without a warrant and to watch everywhere that they go. Thus, a rogue operative committed to malfeasance has the ability to compromise the privacy of his target. We will never eliminate that power.
The question is, given its existence, would there be benefits to a more comprehensive surveillance system? More specifically, would the marginal benefits of broadening the system outweigh the attendant costs? In my opinion, there is a real debate to be had.
That is especially true in modern America, where we have essentially delegated unlimited surveillance powers to the government, subject only to legal restrictions on the purposes for which the collected information can be employed. Any actor willing to flout the law in order to pursue coercive goals will be able to summarily defeat the obstacles intended to deny access to the current fruits of state surveillance.
If anything, providing the NSA and other members of the intelligence community a monopoly on access to this information seems to increase the risks of misuse, by making it easier for a modern day J. Edgar Hoover to seize power by exploiting the asymmetry between his access to information and that of his foes within the government. I would be more comfortable in a society in which the state transparently employed the data it collects for worthwhile ends instead of pretending that it only uses it for narrow purposes like counter-terrorism.
2. Not all data is created equal When we talk about privacy, people naturally begin framing the debate in Orwellian terms. Isn't it possible for us to approach this topic with more nuance?
For example, if we allowed the government to know which books we are reading (something the NSA already knows...), then political activism could be directly compromised, and the benefits would be limited. The same is not true of expanding public surveillance on city streets or highways, to give one example.
The problem with "privacy" in the abstract is that it simultaneously encompasses legitimate concerns and ones that cannot withstand scrutiny. Why is access to our medical information so privileged? Do we want to subsidize the livelihood of people with medical problems by artificially eliminating the ability of employers or insurance companies to take into account all of the available information?
The answer might well be yes! I certainly benefit from that policy judgment. But I'm not sure that it's the correct one, and I wish that we could enjoy an open debate about whether the gains in aggregate economic efficiency would outweigh the costs to individuals. At minimum, shouldn't we explore alternatives like increasing social welfare payments for medically disadvantaged individuals while simultaneously allowing information about their medical histories to improve the accuracy of public and private decision-making? Maybe that approach wouldn't work, but I have not even heard it discussed.
3. Are we utilitarians or deontologists? One of the major dividing lines in these discussions is an implicit disagreement over the appropriate moral calculus.
I default to consequentialism, so in my mind the foreseeable harms from criminality are equivalent to the potential for police abuse. Yet most people process the two in very different ways emotionally.
My well-being is threatened much more directly by criminal gangs than it is by an oppressive state. Only a few months ago, I nearly lost my life when I was attacked in an alley by two thugs.
Should I be forced to be a martyr for privacy when increased surveillance could have deterred their actions?
There's a reason for that: we know what fascism looks like. Historically, oppressive regimes have depended on surveillance, both directly through the state and indirectly through a culture of informants, to enforce ever more fine-grained and brutal policies, and programs of oppression against religious, political and ethnic populations. You seem to be describing privacy as the product of a cargo-cult mentality, as if there were no credible precedent for that point of view. Those "modern political mores" didn't just emerge from a vacuum.
Bad things do sometimes happen when powerful entities correlate deanonymized personal data in aggregate. Your utopian description of the benefits of a ubiquitous surveillance state ignores an entire dimension of human malice that privacy as an ideal seeks to avoid. Although you may be right that this ideal can be taken further than necessary, you haven't really disproven its utility.
(IANAE; just a decent metaphor) In algorithmic finance, one way of determining if a stock will do well in the future is if it was doing well in the past -- consistently (momentum portfolio management). Makes sense. These stocks then do better because algorithms then invest in them. As a human, I have a tendency to fuck shit up every now and again. The government may take pity on me, but if I'm looking for a job it's fairly natural to want to optimize for the best employee possible -- based on existing datapoints. We do this already of course, but it's fairly well known what is considered the most significant stuff and what's going to be on display in the future. (Notice how it already influences us to a great extent...)
Then there's a balance of power. In open source most people can make it to a position of power in a project with some elbow grease. We don't always have the luxury of opportunity for power in real life. Also information isn't going to be public domain; somebody has control over it. I'll give a good reason right now: storing and processing that information takes a fuckton of resources; not to mention it's incredibly valuable. Ultimately sacrificing privacy is giving up a piece of the individuals power for unknown or questionable results.
Or violent crime might continue unabated while a large taxpayer funded workforce uses the surveillance infrastructure the taxpayer paid for to look at hot girls.
I'm all for sanctioning anti-social behavior when it directly harms innocent people. My problem is this:
The surveillance proponent claims to have a solution for some societal ill: Street violence, red light running, smuggling of explosives into airplanes, etc.
The proponent benefits, personally, regardless of how effective the solution is: They sell the cameras, or the organization that employs them gets revenue from penalties imposed on those who are caught by the solution.
Sometimes the proponent isn't personally subject to the new privacy invading system: Proponents of invasive airport screening who fly private.
> In the example from the story, it's clear that our collective welfare could be increased by analyzing reams of restricted data.
Even with this relatively benign example, I see the same case of perverse incentives.
Did we really need to look at tax returns to learn that better teachers have a positive impact? Did our collective welfare improve at all here, or did the researchers just help their own careers more than anything else?
> I contend that it's more useful to look at the discrete harms that flow from disclosure of specific information.
This question is like asking how people could be harmed by leaving their keys in the ignition while they run a 5 minute errand.
The potential for harm depends on the circumstance of the individual.
A multimillionaire leaves his car outside the country club with his driver in it. A poor man leaves his car outside 7-Eleven. Now imagine the poor man is required to appear in court in 45 minutes for a child custody hearing.
There's a significant difference in the possible discrete harms that flow from those situations.
I therefore contend that it's more useful to keep things discreet, unless the individual explicitly opts to release the information.