Allan Lichtman must frick off

found this post https://old.reddit.com/r/fivethirtyeight/comments/1ecdc51/lichtmans_current_standing_of_the_13_keys_harris/lezif3d/?context=8

I don't normally throw around my CV, because I think it is cringy as frick when people start flaunting their degrees, but... I have a masters in political science. From Canada, mind you, but I've done my share of United States civics over the course of that degree and am familiar with statistical models that come out of American quantitative research firms.

If your argument is just birthed out of the sake of "the Thirteen Keys is interesting for poll junkies, and that makes it relevant to this subreddit," then fine, whatever. I do take some issue with you saying this though:

it is interesting to follow the tension between the keys and other models.

It seems to insinuate that 'The Keys' qualifies as a model, and I really, really don't think it does. To elucidate that claim, I want to cite a previous comment I made on this subreddit not too long ago:

Look, what I'm trying to get at is that the 13 Keys is interesting, but political scientists often use it as an example of poor heuristic analysis. That's actually where I first heard of it, about a decade ago, was when I took a US civics course in university.

The professor handed out a printout of the 13 keys with a short description of what each key was and made us all come up with our own predictions. The short of it was that everyone had different ideas on which keys were locked or unlocked, and what it really showed was the bias of the person who did the analysis.

Lichtman's success comes down to the fact that most of the time, it is pretty evident who is going to win, sometimes he fudges the definition of what a successful guess was, and, like Paul the Octopus, some of it is just pure luck.

The reason this is the case is because many of the Keys lack distinctive qualifiers that determine whether or not the keys are "locked" or "unlocked" this is further demonstrated by the fact that we don't know what the model predicts until Lichtman announces what the model says. That is, the model's output is entirely contingent on one man's interpretation of the categories he has arbitrarily declared.

SHOW MORE

You might be inclined to say that 538 or Nate Silver have similar issues in that we must go to their website(s) to see the results of their models, but in principle, if their models were not proprietary, but rather, open source, we would be able to plug in the polling data and extrapolate the same results they came up with. We cannot say the same for the "official" results of the Thirteen Keys.

In short, 538, Nate Silver, the Economist, and RCP have something we can call a model because it is impartial and produces consistent outputs based on the inputs. The Thirteen Keys is more akin to an essay that one history professor puts out every four years that has the same categories of analysis every time. They really aren't that comparable.

That said, this is all just of my gripes with the Thirteen Keys, and if it is in the interest of the subreddit to have Thirteen Keys content, then so be it. I just wanted to throw my opinion out there.

So basically he's saying what you are, but IMO Lichtman's own answers to the keys were just flat wrong for some of them (if mine differed i crossed out his)