Mark Twain said there are “Lies, damned lies, and statistics”, but it could just as easily be said that there are facts, damned facts, and statistics—sometimes all three at once.

I recently visited with Ewing High School’s creative writing and journalism classes, and particularly in the latter, we talked about the ways that certain information and statistics, while factual in the strictest sense, can be manipulated or misleadingly presented to steer the reader’s opinion one way or another.

Toothpaste and gum commercials used to boast that “4 out of 5 dentists recommend” their products. But those statistics, while not untrue, didn’t provide a complete picture. How were the dentists compensated? What choices were they given, and what was the wording of the questions asked? How many dentists began the survey but were dismissed when it became clear they weren’t supplying the desired answers? Pharmaceutical companies play this game regularly, manipulating data to produce statistics that mislead the FDA and consumers, with stakes much higher than selling gum for a dollar a pack.

Compstat, a computer program used by law enforcement all around the country, has proved itself a great tool, but as police departments and promotion boards grew reliant on the statistics generated by the program, little things—like the less quantifiable impact of seeing a police officer walking a beat or talking with neighborhood kids—fell by the wayside. Abuse crept into the system, with major offenses often downgraded to lesser charges to better preserve the desired conclusion: that the frequency of serious crimes was dropping under the reign of the police chief or captain in question. Never have so many dangerous litterers and jaywalkers roamed the streets.

As Janet Norwood, former Commissioner of Labor Statistics, once said, “The real problem is that people often want a number to tell them everything.” Many of us fall into this overdependence on stats—we track exactly how far we’ve walked, and exactly how long it took us. Instead of just walking more and sleeping more, we count steps on Fitbits and track calculations of deep sleep percentage. This can be useful information, but the danger of missing the larger point is greater today than it’s ever been. The same errors and oversights that gave people a false sense of security regarding the likelihood of “Hundred Year Floods” and nationwide financial collapse also indicated that Donald Trump had little chance of winning the 2016 presidential election. (Far be it for me to compare his administration to a disaster, mind you).

Statisticians, and a lot of other people, can easily lose sight of the fact that even if the data is flawless—which it rarely is—models are created and interpreted by humans, who make mistakes and usually have economic incentives to spin the truth.

The ever-increasing dominance of technology has inspired a multitude of science fiction dystopias; the use of statistics and other predictive measures, meanwhile, was cast in a favorable light in Isaac Asimov’s classic Foundation series of novels. Today, people still rule the planet, but, in violation of Asimov’s rules for his fictional science “psychohistory,” the influence of predictive technology is beginning to affect the behavior it’s supposed to be measuring, in a kind of feedback loop. The tail has begun to wag the dog.

In baseball, the practice of applying rigorous data analysis to improve results (famously recounted in the book and film Moneyball) has become ubiquitous. So many baseball stats now exist—some disconcertingly violent-sounding, like WHIP (Walks Plus Hits Per Inning Pitched) and WAR (Wins Above Replacement)—that one has to wonder if the perfect manager of the future will be a robot, maybe with a human assistant manager to run out and kick dirt on the umpire once in a while. But again, the feedback loop is what interests me—because for every manager who shifts his infielders to counter the statistical proclivities of the opposing hitter, there’s a hitter who can see the gap that’s created, adjust, and poke a hit where the infielder normally would have been.

Perhaps the clearest demonstration of the power and prevalence of “big data” is in shopping online. Predictive analytics is not quite the same as statistical modeling—the former uses algorithms to learn from historical data, while the latter requires ongoing human involvement—but both have similar goals: to gain a glimpse into the future. When I visit Amazon’s website, it uses collected data from past purchases to suggest other items I might like. It’s convenient and helpful, but also a bit disturbing—am I really so predictable?

In protest and pure unproductive spite, I recently attempted to outwit the algorithms by thinking of the least likely thing I’d typically buy. “Handmade” and “Luxury Beauty” are two Amazon categories largely foreign to me, and, seeing that luxury also meant “expensive,” I was soon browsing pages for a Handmade Natural Dead Sea Mud Soap Bar, a Handmade Gift Box of Handmade Bath Bombs (not as exciting as they might sound), and a Handmade Set of Walnut Stained Yard Dice, for those who’ve always dreamed of playing Yahtzee outdoors.

The thing is, I couldn’t get myself to pull the trigger. I don’t really want any of those things, and Amazon knows it. In frustration, I recalled the cry of Patrick McGoohan as Number Six in the 1967 television show The Prisoner: “I am not a number, I am a free man!” Echoing this statement to Amazon Echo’s intelligent personal assistant interface, Alexa, it/she responded, “I don’t understand what you mean.”

No, Alexa, you wouldn’t. Quotes from The Prisoner haven’t been programmed in or “machine learned” yet, but I’d be willing to bet that pretty soon, they will be; even so, machines may never truly understand a person’s chagrin at being reduced to a number or a statistic. And I’m not the only one who thinks so—4 out of 5 readers of this column agree.

Peter Dabbene’s website is peterdabbene.com. His essay “The Golden Days of Bowdlerization” can be read at storgy.com. His latest book, The End of Spamming the Spammers (with Dieter P. Bieny) is available on Amazon.