Machine Madness and Big Al Gore
Let’s think about PRISM. The problem isn’t that someone is going to be listening to your telephone conversations or looking at your telephone records. They may well do so, no matter what the President says. Indeed, they will be because that is the nature of human curiosity. However, that is not the problem.
The problem is the machine that is PRISM will be looking at every telephone call you make and every email address you send something to or receive something from. Likewise, the machine will take note of every web site you visit, and, if you have a blog or website, it will make note of every address that visits you.(I like to use the term machine. The difference between machine and some sissy term like data base is the difference between calling an old man who likes kids a troll or calling him Santa Clause.)
So what if PRISM is looking at all your records? Good question. The machine is programed to cross check your telephone and web activity with a set of sophisticated heuristics designed to make assumptions about your usage and, based upon those assumptions, classify you based upon your potential threat to the state and the nation. Again, so what? “I am not a threat to anybody,” you say. While that is almost certainly true you need to keep in mind that the only thing the machine knows is what it has been programed to know based upon what it “sees” when it analyzes your records.
Many of you are professional writers. From time to time, you are seized by an uncontrollable urge to pen a novel or short story. Some lucky few of you occasionally are employed to write an article on a specific subject of interest to a periodical’s publisher or editor. Such an article, novel or short story may require you to do a little research into a topic that may be of concern to our security forces.
Let’s say, for the sake of argument, that you are writing a story that involves a nuclear suitcase bomb. In the old days you’d take your self off to the library at Georgia Tech or MIT or some place where technical literature relating to nuclear weapons in general and suitcase bombs in particular can be found. Now days, however, you’d first hit der google or da badda bing or ye old yahoo. Eventually, given the subject matter, you’d end up using all three, and you’d use them a lot.
In the course of your research you may even have occasion to correspond with some well known nut case who advocates the right of a citizen under the Second Amendment to keep and bear a suitcase bomb. (After all, the amendment does not say you have the right to keep and bear guns, it says arms and a suitcase bomb is an arm, a damn powerful one. Also,the article may require a little proof that the whole concept sought by the editor doesn’t completely lack credulity.) You may well do this because very journalist knows, to prove validity, you must find somebody and demonize him. It is even better if you can find a real demon (Hence your conversations with the nut case) and don’t have to make one up.
In any event, you are doing nothing other than conducting research in preparation to exercise your right to freedom of speech, expression and the press. What could be more American than that? (Nothing, unless you did it all the while keeping a loaded AK47 in your lap, that would be a somewhat better expression of American values). Indeed, this activity is practically a working definition of mom, baseball and apple pie.
Unfortunately, the machine will not look at the web based pattern and emails/telephone calls to and from nut jobs as anything so innocent. The machine will not do this because it hasn’t any real opinion about you or your activities, you are merely an item to be classified. Seeing such a pattern, it will flag you as a potential trouble maker because it has been programed to do so.
You see, the machine and all these terra bits of data and the operating system that runs it is a giant information system. Its sole purpose is to flag potential bad guys and classify them.
I don’t have any inside information on PRISM. I don’t know anybody associated with the NSA or any security organization. I do understand the general outline of information systems and the enormous power over our everyday lives we have ceded to them. They all exist to do exactly what PRISM is doing. They are decision support systems designed to wade through mountains of crap and find a diamond or two. They do this by classifying stuff, in PRISM’s case, us.
I have no idea what level of responsibility has been assigned to the machine. I do not know if it is empowered to flag an “enemy combatant” or a “potential terrorist” or a candidate for the no fly list. I do not know if the system is embedded with an artificial intelligence program that would allow the machine to “temporarily” place, pending human review, a name on any or all such lists without prior human intervention. I doubt anybody in Congress knows. I doubt the President knows. I suspect General Clapper doesn’t know. Any of these guys will have to ask somebody to tell them who they should ask to find out. More than likely, whoever they ask will have to ask somebody else. The guys that could answer questions like that do not hang out in the executive suite.
We all need to know exactly how much power this machine has been given. You know how big organizations work. It doesn’t matter whether the organization is in the public sector or the private sector. If the machine flags a name, nobody wants to take the risk of overruling the machine. What if the machine is right? You override the machine and it turns out to be right, your career is shot. It ain’t gonna’ happen.
What will happen, assuming there is even a mandatory human review of machine decisions, is the name flagged, maybe your name, will be sent down the chain to be reviewed by the local FBI office. They will, in all likelihood, have never heard of you. They’ll have to do a little background research that will involve local authorities (pray you haven’t pissed off the local police bitching about break ins in the neighborhood) and some of your neighbors. Maybe they run into someone with a grudge against you and maybe not.
Due to budget constraints, the local FBI can’t spend a lot of time on this, so they never get around to talking to you. Therefore, they don’t know about the article you’re researching and won’t until it is published, if it ever is. They report back saying the matter is inconclusive, so you stay on the list the machine put you on.
Now, sweet Jesus, the story your working on leads you to pipe bombs and pressure cooker bombs and assault rifles and body armor. PRISM picks this new activity up and cross references it with what they already have. Maybe now, bells and whistles really begin to chime and blow in a great cacophony. Maybe now, you (you already having a classification and a FBI file) really come into focus for the machine and its “masters.” Maybe now, you are really screwed and it fell on you before you even noticed you were under consideration as a terrorist. What a prize you just won.
As a practical matter, the problem is information systems classify people based on gross summations of activity evaluated by mathematical heuristics you couldn’t understand if the programmer explained it to you. You couldn’t understand, not because you’re stupid, but because you don’t speak the language nor do you think the way they think.
Now, the security forces have a file on you, a rather fat file, and they are invested in your status as a bad guy. You could spend the rest of your life trying to unravel this mess. Worse, you may not even know why life has gotten so difficult because every blessed thing associated with the government’s ongoing concern with you is classified. You can’t see it, your lawyer can’t see it, warrants can be issued by secret courts you don’t know exist and to which neither you nor your lawyer have access. Hell, you’ll never even get notice you’re under discussion until the hammer falls. (It could be, if, through some comedy of bureaucratic blunders and intimidation of all sentient humans by the information system’s conclusions about you, you win the security lottery and are classified an enemy combatant. In that case, the hammer could be a drone that takes out your neighborhood.
I know this is highly unlikely. However, if we are going to have this inconceivably huge information system studying some of our activities and making material decisions about us (or recommendations about us), we need to have a thorough and complete debate about what the decisions are, how are they determined, what powers have been delegated to the information system, what data and activities are the decisions based upon, what rights of notification and appeal exist humans caught in this trap and more stuff besides. Lost in the debate between the liberty/privacy side and the security side of the current debate is the detail of how this system operates and what decisions it is empowered to make.
PRISM could be an okay thing, maybe even a good thing, but it has to be subject to our democratic processes and the rule of law. If we don’t insist on that, no one, not Barrack Obama, nobody, can truthfully say anything about what this will do to individual citizens or our society. We are not, as it is currently constituted, ceding massive new powers to the federal government. The government already has these powers. The problem is we are ceding these massive new powers to an information system over which no human system of politics and law has ultimate control. We really do not want to do that.
HAL (2001-A Space Odyssey) is real, its name is PRISM and we are being asked to live with it. Like war being too important to be left to the generals, a massive and ubiquitous and intrusive information system like PRISM is too important to be left to the programmers and the system designers and the politicians. None of them understand the human/information system interface.
I know President Obama won’t want to hear this, but he needs to get on the phone with Al Gore and ask him how to fix this. There may be others who understand the technology and the politics and the international terrorist threat and and the human/technology interface as well as Big Al, but none of them has recently been interviewed on the subject.