Discussion: View Thread

As a paid-up member of the Yentas With Attitude local, Maxine has been snooping diligently into hashslingrz, before long finding herself wondering what Reg has gotten himself into and, worse, what he's dragging her uncomfortably toward. The first thing that jumps out of the bushes, waggling its dick so to speak, is a Benford's Law anomaly in some of the expenses. Though it's been around in some form for a century and more, Benford's Law as a fraud examiner's tool is only beginning to surface in the literature. The idea is, somebody wants to phony up a list of numbers but gets too cute about randomizing it. They assume that the first digits, 1 through 9, are all going to be evenly distributed, so that each one will turn up 11% of the time. Eleven and change. But in fact, for most lists of numbers, the distribution of first digits is not linear but logarithmic. About 30% of the time, the first digit actually turns out to be a 1-then 17.5% it'll be a 2, so forth, dropping off in a curve to only 4.6% when you get to 9. So when Maxine goes through these disbursement numbers from hashslingrz, counting up how often each first digit appears, guess what. Nowhere near the Benford curve. What in the business one refers to as False Lunchmeat.

"Once I got access to a real computer," Luke resumed, "I did a Bernoulli distribution. Do you know what that is, Mr. Smith?" The blond man shook his head. "He does, though," Kalisha said. Her eyes were merry. "Right," Nicky agreed. "And doesn't like it. The Whatzis distribution is not his friend." "The Bernoulli is an accurate way of expressing probability," Luke said. "It's based on the idea that there are two possible outcomes to certain empiric events, like coin flips or the winners of football games. The outcomes can be expressed as p for positive result and n for negative result. I won't bore you with the details, but you end up with a boolean-valued outcome that clearly expresses the difference between random and non-random events." "Yeah, don't bore us with the easy stuff," Nicky said, "just cut to the chase." "Coin flips are random. Football scores appear random if you take a small sample, but if you take a bigger one, it becomes clear that they're not, because other factors come into play. Then it becomes a probability situation, and if the probability of A is greater than the probability of B, then in most cases, A will happen. You know that if you've ever bet on a sporting event, right?" "Sure," Tim said. "You can find the odds and the likely point-spread in the daily paper." Luke nodded. "It's pretty simple, really, and when you apply Bernoulli to precognition statistics, an interesting trend emerges. Annie, how soon was the fire after your aunt had her brainwave about keeping her sons home?" "That very night," Annie said. Luke looked pleased. "Which makes it a perfect example. The Bernoulli distribution I ran shows that precognitive flashes-or visions, if you like that word better-tend to be most accurate when the predicted event is only hours away. When the time between the prediction and the event predicted becomes longer, the probability of the prediction coming true begins to decline. Once it becomes a matter of weeks, it pretty much falls off the table and p becomes n." He turned his attention to the blond man. "You know this, and the people you work with know it. They've known it for years. For decades, in fact. They must have. Any math wonk with a computer can run a Bernoulli distribution. It might not have been clear when you started this thing in the late forties or early fifties, but by the eighties you had to know. Probably by the sixties." Smith shook his head. "You're very bright, Luke, but you're still just a child, and children indulge in magical thinking-they bend the truth until it conforms to what they wish were true. Do you think we haven't run tests to prove the precognitive capabilities of our group?" His lisp was growing steadily worse. "We run new tests every time we add a new precog. They're tasked with predicting a series of random events such as the late arrivals of certain planes . . . news events such as the death of Tom Petty . . . the Brexit vote . . . vehicles passing through certain intersections,...