Allegedly “inspired” by my recent browsing history—a 3 lb bag of Gala apples that I am quite certain I neither viewed nor purchased (I’d buy Greenmarket apples, hello?)—Amazon.com has helpfully suggested the following products for me:
And now that I have actually clicked on and viewed these three items, OMG I cannot wait to see what “inspired” product suggestions Amazon has for me next! How about a giant rubber fish head mask? Glow-in-the-dark toilet paper? An iPhone case made out of astroturf? Ooh! Ooh! A Donald Trump board game!* The possibilities truly boggle the mind.
Another thought occurred to me, and I really have to check myself here because it sounds more than a little paranoid. See, I’m pretty sure Amazon.com employs some of the most sophisticated software available to generate these product suggestions. Their programs keep track of everything I’ve ever searched for, viewed or purchased via Amazon, and probably every other online transaction I’ve ever made and website I’ve visited, my Facebook posts and Tweets, what I ate for lunch yesterday, plus whatever else can be hoovered up by Big Data bots. Naturally, Amazon also tracks what other customers bought immediately after viewing a 3 lb. bag of Gala apples, and all of this information is compiled and analyzed for the purpose of generating suggestions for the products and services I am most likely to purchase. That’s the entire raison d’être: to extract the maximum amount of cash from my coffers as quickly and efficiently as possible. And somehow, all of this data mining and number crunching has culminated in a prediction that I just might be in the market for a $10 bottle of coyote urine.
Are you with me so far? Okay good.
The United States government presently deploys thousands of federal, state and local organizations to engage in domestic surveillance for the purpose (allegedly) of counterterrorism. These programs are spectacular failures in this regard, and one never needs to look very far or wait very long for even more evidence of this rather crucial fact that is never, ever mentioned in mainstream media coverage. Based on what we know for certain, surveillance agencies such as NSAC use big data analysis just like Amazon does, except that instead of generating very tempting suggestions to purchase coyote urine, the objective of state surveillance (again, allegedly) is to discover terrorist suspects and so-called “clean skins.” That is,
“people with no known affiliation to terrorism or crime, needles in a giant haystack that don’t necessarily look like needles. Or people who aren’t needles at all, but who might become needles in the future and thus warrant observation today.”
In other words, the data sets and search algorithms are different, but the underlying concept of predictive software is the same. So how good is the government’s technology at identifying potential terrorists? Does it serve up better or worse results than coyote urine?
The question is more than academic. Because what’s at stake here is much bigger than the US market for coyote urine (I wonder how they even collect that stuff...You know what? Never mind, I don’t want to know). What’s at stake is the likelihood that any one of us can be designated a potential terrorist threat by the state, based on data that may not even be accurate: recall that I neither searched for nor viewed (much less purchased) any 3 lb. bag of apples in the first place. It’s one thing to write fun essays about Amazon’s bizarre product suggestions and speculate about their software, but just try arguing that the government’s data is wrong at your military tribunal.
And it’s not just the false positives that matter— it’s the misses. Mohammad Youssef Abdulazeez, a 24-year-old Jordanian national born in Kuwait who visited the Middle East last year, just shot up a military recruitment center in a Chattanooga strip mall before attacking a Naval reserve center seven miles away, killing four Marines and wounding three others. He used an AK-47 style weapon and had tons of ammo in 30-round magazines. Yet before the shooting, none of his activities generated any surveillance alerts or investigations by law enforcement? Nada?
Richard Reid (the “Shoe Bomber”), Umar Farouk Abdulmutallab (the “Underwear Bomber”) and Faisal Shahzad (the “Times Square Bomber”) were all spotted and stopped by civilians, not the state. And despite the lies to the contrary by our highest government officials, none of these programs have ever prevented a single terrorist attack. I’ve pointed out many times that the Boston bombing case should have put an end to the entire surveillance state apparatus, immediately and permanently: the U.S. government was tipped off twice about Tamerlan Tsarnaev by the Russian Federal Security Bureau, and he had been on a terrorist watch list watch for eighteen months before he and his brother set off bombs that killed three people and injured 264 others at the Boston marathon. If mass surveillance cannot prevent incidents like this one from happening, what is it good for at all?
All of which is to say—yet again—that mass surveillance is worse than useless, at least for its purported purposes.
But okay. Maybe I’m wrong about all of this, and as a result I’m missing out on something very important. Hey, does anybody have any good recipes for coyote urine?
*Yes, these are all actual things that exist in the world and are sold on Amazon. I’m sure you will agree this fact alone is pretty solid evidence that our civilization is imminently doomed.
[a version of this column appeared at perry street palace.]