Ed Lorenz back in 1961 was running, on a computer!, a simple weather model, with twelve full parameters. This was in the days of punch cards and paper print outs.
He wanted to re-run part of the model, but computer time was expensive then, so he decided to start the simulation mid-stream. This meant he had to type in the parameter values at the given time, which he retrieved from the print outs.
Then something strange happened. The output diverged from the previous run, not just by a little, but by a lot. This is strange because the model was fully deterministic. If a deterministic model starts at the same point, the mathematics says, it should always end at the same point.
And so they do. Which meant Lorenz did not start at the same point, only the approximately same point, a point so close to the original starting point that any would consider it “close enough for government work.” Although it wasn’t, and couldn’t be, “close enough”. It turned out, he learned later, it has to be exactly identical, and the the slightest deviation led to wildly varying solutions.
Thus was born, or re-born, the science and mathematics of chaos theory, though a better name might be sensitivity theory. There is nothing chaotic, in the plain-English meaning of that word, in deterministic models. They are boring, these models, always doing what they are told to do (like all models!). They are just difficult to predict because they are sensitive to the initial conditions.
This sensitivity Lorenz discovered turned out to be impossible to ignore, because meteorologists used to try to predict future weather based on analogues. If the weather on a date back in, say, 1952 “looks like” the weather today, we have found an analogue. To forecast the weather a week from today, look at what the weather was a week from your analogue date in 1952.
This approach works reasonably well for near-term forecasts. The day after your date in 1952 will be close to tomorrow, but two days after your 1952 date won’t be as close to two days from now. And so on. The analogue breaks down as time progresses. Weather in real life, like weather in a simulation, is a system that is sensitive to initial conditions.
For simple deterministic models, sensitivity theory (there is no chance this new name adopted) is by know well understood mathematically. It is a fun topic. All we need to know about the math is that sensitivity is found in many non-linear systems which are non-linear. These systems have all kinds of feedbacks and peculiarities that make them difficult to track.
There’s much more too all this, of course, but for our purposes it’s sufficient. Because we’re trying to understand the usefulness of historical analogue forecasting, not just in weather but in any area. This, after all, is the prime season of forecasting by analogue in politics.
Pundits are saying what it was like in the 2016 election when Trump was ahead or behind, or how Democrats did this or that well in an economy similar to this one, and on an on. Some eschew political analogues. Conrad Black writes “This campaign, and this president, are like no other”, but this rejection is more of a figure of speech than a philosophy.
Analogues were sought frequently back when people watched sports. You’d hear guesses that tonight’s pitcher was going to do well because he done so against opposing teams that had so many recent losses on Tuesday evenings when the stock market was up.
That’s a big problem with historical analogues. What counts as the analogue. With weather, it’s more or less clear. The temperature and pressure fields have a certain shape, and this shape is correlated with the dynamics of movement. There’s more to the weather than just that, of course, but even if we had every aspect that was causative of the outcome, we’d still have to match our analogues perfectly in order to avoid sensitivity.
With politics, sports, and other human behavior, it’s far less clear what counts as what measures should be examined to find analogues. Sensitivity to “initial” conditions (the start of our analogue) is almost a given. That we can ever know with certainty, as we can in mathematical models, all initial values to sufficient precision is obviously true. Groups of men are still composed of men and there’s no way to know the full mind of even one man, let alone groups of them.
This doesn’t make historical analogue forecasting valueless. We know it has value, which is one of the reasons we study history. To see what people do in given—imperfectly known—circumstances. We look around us and compare ourselves or our countries with how it was then, wondering whether what happened to these other people will happen to us.
It’s going badly now in the culture, and the natural question is how bad will it get? This is not necessarily the same as how bad it can get, but we’d like to know if these two coincide. The search for analogues begins.
Gary Saul Morson might have found one in
pre-revolutionary Russia, summarized in his article Suicide of the Liberals. His analogue applies only to the intellectual class, because it is obvious there are large differences between Tzar-led aristocratic Russia our our elite-led oligarchy. Nevertheless, the similarities between their intelligentsia—which for good reasons he translates as intelligents—and ours is frightening.
I can only give a small flavor of the analogue here: read his piece to experience the full nausea as you recognize each similarity.
Most important, and of greatest concern, was how intelligents thought. An intelligent signed on to a set of beliefs regarded as totally certain, scientifically proven, and absolutely obligatory for any moral person. A strict intelligent had to subscribe to some ideology—whether populist, Marxist, or anarchist—that was committed to the total destruction of the existing order and its replacement by a utopia that would, at a stroke, eliminate every human ill…
Though some liberals recognized their differences from the radicals, most acted like intelligentsia wannabes who were unwilling to acknowledge, even to themselves, that their values were essentially different. Socialized to regard anything conservative as reprehensible—and still worse, as a social faux pas—they contrived ways to justify radical intolerance and violence as forced, understandable, and noble. They had to, since the fundamental emotional premise of liberalism—hostility to those ignorant, bigoted, morally depraved people on the right—almost always proved more compelling than professed intellectual commitments.
This analogue, should it prove to be one, has the obvious and immediate forecast. We know what happened in Russia. The intelligents ate themselves first, and after the Party gorged on them, they turned on the right remnant. Will the same thing happen here? Or are the “initial” conditions just too far apart from ours for this analogue to be of any use?
To support this site and its wholly independent host using credit card or PayPal (in any amount) click here