Reblogged:A Quick Meta-Audit of a 'Climate Crisis' Audit

[Gus Van Horn blog]

Of course I realize that there is no such thing as a quick audit of anything, however...

The following provocative title bubbled up from this morning's blog feed: "Let's Audit Alex Epstein." Within the post, Bryan Caplan, whom I take to be receptive to -- if not generally sympathetic -- with the arguments energy expert Alex Epstein lays out in Fossil Future, challenges his readers:

Image by Campaign Creators, via Unsplash, license.

Alex Epstein's Fossil Future addresses dozens of complex empirical literatures. It would take years for me to read them all, and I don't feel like I have these years to spare. I am however willing to run a random audit.

Two audits, actually.

Here's what I propose: In the comments, nominate narrow empirical issues that you suspect Epstein covers in a biased manner. Tell me exactly what I should read on the other side, but try to keep the reading recommendations short.

This proposition is interesting in more ways than one. If Epstein emerges relatively unscathed, it will increase his credibility among persuadable readers. If Caplan uncovers major errors, which would surprise me, how Epstein chooses to address them will become important.

The above sets aside subterfuge (for or against Epstein) by Caplan (which I do not regard as a real possibility). Mistakes by Caplan in his analysis or the small chance that he and his readers find a major error would be exposed by any give-and take.

So far, so good.

But there is a hazard: What if this audit uncovers one thing that, while it does not damage his overall case, Epstein has completely whiffed? Forget the people who will take just any error as an excuse to (continue) dismiss(ing) Epstein as a "climate denier" and as license to ignore everything else he says. (We'll get to them in a moment.)

Lots of people have a less-than-firm grasp of induction, and mistake deduction (such as in a mathematical proof) as the general model of rational argumentation. One error in a mathmatical proof invalidates the proof, but that is not always true for an indictive proof. A mistake in an inductive argument may or may not invalidate it.

For example, the theory of evolution is based on a huge preponderance of evidence from several areas of biology and other sciences. The fact that we're constantly correcting or learning new things about, say, human evolution does not call the theory of evolution into question at all.

Of course, that hazard is one I'd imagine Epstein, a philosopher by training, would be able to discern and address, so ... maybe that's more of a bonus than a hazard...

Like Caplan, I'm going to turn to a second datum: Alex Epstein recently "audited" ChatGPT on the issue of fossil fuels by seeing how its responses to Make a 4 paragraph case against the anthropogenic climate change crisis changed over time. He reports:

ChatGPT by @OpenAI now *expressly prohibits arguments for fossil fuels*. (It used to offer them.) Not only that, it excludes nuclear energy from its counter-suggestions.

Wow. What are the programmers afraid of here? An intelligent user of this AI will know that it's an AI and not take everything it spits out as gospel. And there's no helping the kind of person who fails to evaluate knowledge claims with anything less than an active mind, anyway.

So we have two non-crackpot data points, one from each side of the "climate" debate. One side is trying to find flaws in the arguments of its greatest champion, and one side refuses to consider an opposing case, even though it could be great for "steel manning," that is, exploring one's own position with the view of being able to have a firmer grasp of it or argue for it more persuasively.

I end my audit with a question: If you ran into a debate between two factions, and one of them encouraged debate, while the other dismissed even the need to examine premises, which side would you guess was more likely to possess the truth or at least have a better chance of finding it?

-- CAV

Link to Original