Bad Predictions Mean Bad Models, Even If You’re A Distinguished Expert

Bad Predictions Mean Bad Models, Even If You’re A Distinguished Expert

Suppose you’re an Expert, or even an ordinary person, and you say, “If the world is like X, then Y will happen”. Or you say or imply, “If the world is like X, then Y has this-and-such high probability of happening”.

Suppose it turns out the world was like X but Y did not happen. What could this mean?

It means you were wrong (or way off).

This was the case with distinguished professor Allan Lichtman, who predicted Kamala would win and Trump would lose. He said that the world was in a certain way or condition, X, and he listed the items in this condition (“a formula of 13 true-or-false questions”; we looked at this before the election, blog/Substack). He also told us that he had verified that the conditions were met. This led him to announce boldly that Y, a Kamala win, would happen.

It did not. He was wrong. But how was he wrong?

First understand there is a little hidden something in statements “If X then Y” (or the same with some probability, which is a pain to keep typing and reading, and it doesn’t matter here anyway). The hidden something is the “glue”, or the tie between the X and Y. That glue is the model. The model has any number of premises, including perhaps old observations, some premises perhaps mathematical, or even pages of math and code. It doesn’t matter. This collected list of premises, no matter how large organized or disorganized, is the model, which says “If X then Y.”

Sometimes in science we get to see these models, and sometimes we don’t. We don’t with Lichtman because he never really wrote it down. He only told us of its existence. He said “If X then Y”, from which we deduce he has a model.

Lichtman, we have been told, was usually right with his model. Which, of course, means he was sometimes wrong.

He was wrong this time. He should have blamed himself, for he is the model. His thought processes I mean.

There are only two things that can break a prediction: the model itself, or those X. It’s obvious, for instance, that bad math (that which ties Xs to Ys) can lead to bad models. But so can bad X break a model. If you say the world will be like X, and it isn’t, then even if your model is a lofical deduction from X to Y, containing no uncertainty or error, then your prediction still dies the death because you got the X wrong. This applies too all models of any kind; yes, even AI.

Lichtman assured us the world was like X, so by rights and logic he should only blame himself, i.e., blame the model.

But he tried to blame the X! He came out last week and said the world was not like X after all! That somebody had pulled X away, coated it with “disinformation” (his world), and released it back into the wild, where it only appeared to be like X. So close was the resemblance, that it fooled—tricked!—Lichtman himself.

Lichtman found a camera which he could look into and then accused the world of “…disinformation. We’ve always had disinformation, but disinformation has exploded to an unprecedented degree. You talked about a grievance election, but a lot of that grievance was driven by disinformation.” 

Lichtman pointed to conservative media platforms and Musk, who poured millions into President-elect Trump’s campaign and has become one of his loudest media cheerleaders, as a factor in his inaccurate prediction.

Musk had helped fuel the spread of false or misleading information online about issues like immigration, hurricane relief and the war in Ukraine, Lichtman said, effectively “putting his thumb on the scales.”

“And you know, as scholars have shown, once you dissolve truth, democracy dissolves along with it, the way authoritarian takes hold, and it’s taking hold all over the world, not just here, is not through force, but through the manipulation of information, as George Orwell warned in 1984 you know, in that dictatorship, war is peace, famine is plenty,” Lichtman said.

I know how the poor fellow feels. I, too, though not an Expert (an Enlightened person with credentials), have made several public predictions, and have fallen right onto my keister. It stings. It is unpleasant. You wish it were not so. The temptation to grab on “If only…” is difficult to resist.

Lichtman, poor fellow, did not resist. I have sympathy for him. He has been sought, and lauded, every four years, because he has brought good news to the managerial class. He did this time, too. Alas, for him, the news turned out badly. He likely won’t be called next time. He knows this, and it hurts.

Incidentally, the crude modified Briggs Persistence Model beat Lichtman (same links: blog/Substack). This was the model that says the incumbent wins if he is running, else the opposing party does, unless there is a major event like an assassination (Kennedy). Even, as it turns out, a virtual one. This cheesy model predicted Biden would win. Biden, you understand, before he was booted. Then they swapped in some other candidate.

Subscribe or donate to support this site and its wholly independent host using credit card click here. Or use the paid subscription at Substack. Cash App: \$WilliamMBriggs. For Zelle, use my email: matt@wmbriggs.com, and please include yours so I know who to thank. Buy me a coffee

1 Comment

  1. McChuck

    Of course, with Leftists, the bad model is everything inside their head. And then when reality doesn’t match up, they either deny reality, or attack us. They really don’t need much of a prompt to attack us.

Leave a Reply

Your email address will not be published. Required fields are marked *