It's worse when nobody's listening

I'm going to assume readers are familiar with the whole fracas surrounding the revelation that the government is doing more-or-less exactly what everybody (should have) assumed the PATRIOT Act and FISA re-authorizations of the past decade allowed them to do.  I don't want to get into a recapitulation of the whole thing here.

I also don't want to get into a whole long thing about privacy, which is something I think I've sounded off on in the past (here, for instance; or here).  What I'll just say in brief is that I think the NSA data collection is problematic for a lot of reasons that involve policy, politics and civil liberties but not so much "privacy" per se, which I still think is a mostly outdated concept.  It's worth mentioning that the real issues we're talking about here involve the retention of data that's created by phone companies and online service providers in order to make technologies like cell phones and e-mail work at all, and who has the data, which (prior to the PATRIOT Act and FISA re-authorization bills) was solely in the possession of private companies until those companies destroyed the data, unless a government actor subpoenaed them to preserve it and turn it over.  (Which brings up another point that's being widely overlooked: state and government actors have always had potential access to phone and online records, what has changed isn't the fact of access, but the conditions: a subpoena or warrant can be challenged by the receiving party, doesn't reach to data that's already been destroyed, etc.)

No, what I really wanted to get into is that the business of collecting and indefinitely keeping records raises civil liberties concerns that don't have much at all to do with privacy, if there's still any such thing.  The privacy issue may even be a distraction.

I think we might start with: critics of the NSA collection policies have bounced the adjective "Orwellian" around a good bit, which is unfortunate because it usually reflects a widespread misunderstanding of what was going on Orwell's 1984.  The common perception is that Orwell described a surveillance state where everybody was being watched; this is incorrect.  What Orwell described was a state in which members of The Party, i.e. pretty much everybody in any kind of bureaucratic position, might be under surveillance.  Orwell recognized that even in a somewhat allegorical work, the idea that any government bureaucracy, however wide-reaching, would have the time and resources to monitor everybody was simply absurd; what a totalitarian state might do, however, is foster a paralyzing sense of paranoia amongst everybody who might turn into a troublemaker by giving all of those people a rational basis for thinking they could be under surveillance at any particular point in time, including a point in time in which they were actually making some kind of trouble (however slight).  In 1984, the ruling caste of Oceania knows the ever-present cameras in their Party apartments aren't always being monitored, they just don't know when they're being watched and thus must behave as if they're always watched.  The Proles--who make up most of Oceania's population--aren't really under surveillance at all.  (Furthermore, since there's possibly no longer any actual counterrevolution occurring--it's implied that Emmanuel Goldstein is long dead and was never the traitor Big Brother makes him out to be, an echo of how Stalin used Trotsky after his exile and assassination--the Party must continuously eat itself so that there's an appearance of continuing counterrevolution for the people to unify against: hence the entrapment of impotent misfits like Winston Smith and Julia.)

But if government collection of data that's generated by everybody's acquiescence and consent to participate in technologies (nobody's forcing you to have a cell phone or GMail account, are they?) isn't "Orwellian", is there a better adjective for what might be wrong with it?  There is, and I think Daniel J. Solove nails it in a piece posted at The Chronicle Of Higher Education, "Why Privacy Matters Even If You Have 'Nothing To Hide'":

Another metaphor better captures the problems: Franz Kafka's The Trial. Kafka's novel centers around a man who is arrested but not informed why. He desperately tries to find out what triggered his arrest and what's in store for him. He finds out that a mysterious court system has a dossier on him and is investigating him, but he's unable to learn much more. The Trial depicts a bureaucracy with inscrutable purposes that uses people's information to make important decisions about them, yet denies the people the ability to participate in how their information is used.

The problems portrayed by the Kafkaesque metaphor are of a different sort than the problems caused by surveillance. They often do not result in inhibition. Instead they are problems of information processing—the storage, use, or analysis of data—rather than of information collection. They affect the power relationships between people and the institutions of the modern state. They not only frustrate the individual by creating a sense of helplessness and powerlessness, but also affect social structure by altering the kind of relationships people have with the institutions that make important decisions about their lives.

Kafkaesque.  I've been a fan of Franz Kafka since I was a teenager, and I'm a little embarrassed I wasn't using this word weeks, if not years ago.  Yes.  The common thread through much, most, possibly all of Kafka's fiction is the idea that the universe in fact has a basic, fundamental meaning, an underlying truth and is governed by a rational system: all of which is denied to you; you will not be told the truth (nor will you be lied to), you will not be granted the meaning, you will not be able to see how the system works altogether.  There are rules, you just won't be told what they are.  There are facts, but if you don't already know them it isn't anyone else's responsibility to share them.  Kafka's characters are always on the outside, almost inevitably trying to figure out how to get in: there is a way inside the castle even if you can't get in, there is a message from the Emperor Of China even if you never hear it, there is a crime that has been committed and for which you will pay even if the indictments remain sealed, you deserved to be turned into an insect even if you're not sure why, all of your bad acts and consequences thereof will be carved into your skin although you'll die without ever having a mirror to read what has been written--et cetera.

There is an irony in the collection of metadata that I sense only bothers me, and something people keep saying which I suspect only bothers me.  And at the risk of being labeled merely contrarian I'm going to try to explain what it is and why it bothers me and how Kafkaesque it is (and therefore, perhaps, why the NSA data collection program really is wrong).  At the heart of it is something that critics and defenders and people who are merely blasé about the whole thing seem to agree upon: "The NSA is only collecting metadata--where and when the calls were placed, the lengths of calls, things like that--they aren't listening into the actual calls or recording the contents of the calls."  And they agree, critics and defenders and people who are merely blasé, that this is a good thing.

It isn't.

Now let me see if I can explain why in a sensible way.  Because this seems counterintuitive, because if you're used to thinking about privacy issues in Orwellian terms (properly applying Eric Arthur Blair or not), it seems like collecting data is less intrusive than snooping, and therefore better (even if you also feel the data collecting is wrong).

There are a lot of things that can be done with all this data.  One of them is nothing: the data can be piled up on the server equivalent of the big warehouse at the end of Raiders Of The Lost Ark, more information than anyone can possibly process even with the use of sophisticated computers.  Aside from the unlikelihood of this scenario, given the leaps and bounds of computer processing power and exponential improvements in software development, it's also a de facto argument for not collecting and keeping the data at all: if it's useless, why do you even want it in the first place?  I think we can assume it's not useless.

Another way the metadata can be used is to connect people.  Kieran Healy wrote a primer if you're interested in how it can be done.  You might do this after you've decided a phone number is suspicious, which is what William Saletan is pretty sure is happening, using the metadata to figure out who might be connected to a suspected terrorist's cell phone; and Saletan is probably right, since another option--preemptively connecting people by defining every local social network in the country--is astronomically challenging (right now, with the resources available, but I assume it'll eventually be feasible if the government's interested in doing it; not that I think that part of it matters all that much if they can).

So if you decide that phone number 1-555-555-5555 is a terrorist's number, you can go back into the accumulated data and figure out all sorts of things from who has called that number or been called by that number and where and when and how.  And you can use that information to build a picture of the social networks around that number, and the networks around those networks, and potentially generate a kind of organizational chart of activity that might be terrorist related.  And that might save lives and stuff.

So what's the problem?

The problem is that you have networks devoid of content and context.

Here is what is going to happen, eventually (if it hasn't happened already).  A government agency will arrest somebody.  They will arrest this person as a matter of public safety, because the way he's embedded in various networks suggests a high probability he's up to something and knows others who are up to something.  They will arrest him to prevent something from happening the government thinks has been planned, or he will be arrested in the aftermath of something that couldn't be prevented.

The government will not want to charge him with a crime.  Not in a standard courtroom.  Because charging him with a crime will trigger certain rights: he will be entitled not to have his own statements used against him unless he was notified of his rights and offered a lawyer (Miranda), he will be entitled to exculpatory investigative materials in the state's possession (Brady), he will be entitled to face his accusers (Crawford), etc.  The government will not want to publicly disclose the tradecraft used to build the case against him, they will not want to "burn" possible undercover or turncoat witnesses by exposing them in court.  The government will want to have no trial at all, or a trial in some kind of tribunal setting in which the suspect has fewer or no rights.

They may not even want to tell the suspect what he's suspected of.

But they will want information.  They may tell the suspect this much, "We know you called so-and-so five times between this date and that one.  What did you talk about?  Who else is involved?  What else do you have planned?"  They may even resort to extraordinary rendition, or at least threaten it: "If you won't come clean with us, maybe you'd like to talk to these gentleman who are here to take you to Yemen when your plane lands over there."

They will do these things because they are good people.  This is the first thing I want you to understand.  They will not arrest someone without formal charges and deny him of his rights because they are evil, cruel, or villainous.  They will do it because they joined a government agency and accepted less pay than they might have earned elsewhere and more stress and longer hours because they wanted to serve their country, wanted to save lives, wanted to stop evildoers, wanted to give something to a thing larger than themselves.

And the second thing I want you to understand is that most of the people they do this to will be guilty as all hell.

But there's going to be this one guy they arrest.  They're going to ask him what he knows.  They're going to tell him what they know.  They're going to say, "We know you called your brother five times between March and June," for instance, and something like, "we know you called him five times before the incident and we know you're connected to a bunch of his associates.  Come clean.  We know you're involved."  And he's going to say, "I know he did a terrible thing, but he was my brother; we never talked about terrorism except one time, and I told him I thought he was making terrible friends and our mother would be ashamed of him.  Maybe I should have called the police, but he was my brother and I didn't know how involved he was, and then after it happened I was scared.  But we only talked about anything like that once, and I told him he was with bad men."

And they will say, "Sure."  And they will lock him up, or send him overseas to a black site or one of our less-scrupulous allies.  And if he ever gets a hearing of any kind, be it a trial or military court or just some kind of kangaroo tribunal, they will say, "He called his brother five times between March and June and do you think they never talked about the most important single thing in his brother's life?"  And he will say (if they let him be heard), "I said our mother would be ashamed of what he might be doing," and they will reply, "Of course he would say that, wouldn't he?  Five times, between March and June."

And what will he do?  How will he fight circumstantial evidence with missing evidence, if he's really telling the truth.  (And we don't know, do we?  If he's telling the truth?)  We have metadata, not data; we have time and place and how long and who else and everything except what he really said.  And he tells us he said something harmless or even tried to talk his brother out of it, but what would you expect him to say, the man's a suspected terrorist.  He'll say anything to save his life, and who wouldn't?  (He possibly even confessed to it, when he was in Yemen.  Or there's the conversation his blockmate swears happened when they were cell-by-cell in the prison camp.)

Here's a kicker.  If the NSA were really listening to phone calls, they might not even arrest the guy.  Would probably bring him in, sure, debrief him: "We know you tried to talk your brother out of this.  You really should have called someone.  Can you think of anyone else we should call?  Here's our card, if you think of something."  Because they aren't, you know, evil, cruel, or villainous; they did, you know, want to serve, to protect.

Do you see what I'm saying here?  Collecting the data without saving the calls is actually worse than listening to the conversations.  Because gods know, listening in is a ridiculous breach of privacy, an insult, an intrusion, a violation.  But it's also the only way you can use the data without relying on inferences that may be incontestable.

We have a ridiculous idea in the post-Enlightenment west.  We have several.  But there's this one: "Better," we say, "to let ten guilty men go free than punish one innocent man."  This is stupid.  If you care about public safety, really care, it's better to randomly murder a hundred people on the off chance one of them might eventually do something terrible.  But we consider that capricious and cruel, the kind of thing tyrants and madmen do.  So we have the former idea, that you're better off sometimes allowing something good to happen to the iniquitous than to allow the terrifying weight of the state to fall upon a single innocent.

I see two ways to protect the innocent here, and I realize I'm utterly mad and everyone will think this is the stupidest thing they've ever read.  The first is to keep nothing, at least not without specific warrants issued upon the basis of probable cause, a mechanism that has mostly worked in America for two centuries.  And the second is to keep everything, so that if a pattern emerges from the data--and patterns will emerge--you can dig deeper and find out whether the pattern means what you think it might; you might, if you're feeling squeamish about someone listening to your conversations, create double-blind mechanisms: e.g. the call is recorded, but no one listens to it without a warrant and/or other appropriate due process.

This is not some kind of modest proposal, some kind of attempt at satire.  As I see it, we have created or are creating a Kafkaesque system in which evidence is collected against people who are not suspected of a crime, for use later if they are suspected, without telling them what the crime they might eventually be suspected of is and without recourse to the law in responding should they ever be made aware of the accusation of the thing that hasn't happened yet.  You are the friend of a friend of someone who is suspected, and both of you are friends of somebody else, and so you're probably a terrorist, but nobody knows all you ever talked about with any of them was sports, or the weather, or how to poach eggs.  Perhaps you knew they were all ne'er-do-wells and the only reason you ever talked to any of them was to discourage them: I hear Christ spent a considerable amount of time hanging out with prostitutes, thieves, rabble-rousers and radicals and have concluded he was a First Century racketeer.  (No, wait: I know what he supposedly said to all of them.)

Keeping half is worse than keeping all, or nothing-at-all.