5 Comments
User's avatar
Gio's avatar

Excellent post, as always, Brian.

Your, correct, observation on regulatory capture reminded me of this line from Matt Ridley's The Rational Optimist:

Each empire was the product of trading wealth and was itself the eventual cause of that wealth’s destruction. Merchants and craftsmen make prosperity; chiefs, priests and thieves fritter it away.

It really is a tragedy that Popper's philosophy is so unknown these day. I myself, despite being subject to three years of philosophy classes in high school, never heard of him till last year, thanks to Brett Hall's work in popularizing David Deutsch's work, who in turn builds upon Popper.

I suspect one reason for Popper being out of the main stream is that, at its core, fallibilism is hard work. One has to come to terms with one's omnipresent ignorance. And the whole point of falsification is that one has to come up with experiments and counterfactuals.

It's much easier to run observation studies, collect the data, and say it "suggests X" and that "more research is needed." As a matter of fact, and I only realize it now, this is the same problem of incentives that you mention in the context of homelessness. When the game is publish or perish, what better way to keep publishing than to never make progress, only make more observations?

Expand full comment
Brian Doran's avatar

Thanks Gio! Yes, I really think it all ties back to incentives. One path is hard, unrewarding, and disagreeable. The other is easy, rewarding, and agreeable. I think thematically it makes us lazier, more herd-like, and less intellectually honest with ourselves if we forget even the existence of falsification. A lot could be said about this in the context of our consumption of Media, and the things we choose to believe without evidence.

But also - Popper seems to be having a bit of resurgence lately. First with Nassim Taleb, then with Naval, David Deutsch, and Balaji. Maybe as things we thought were true become exposed, and with AI increasing that inauthenticity, our brains will become falsification machines by default just for survival..

Expand full comment
Michael A Alexander's avatar

I found it pretty incoherent. It is a series of non sequiturs. Like going from the idea of saying what something Iis by saying what it is not to why stuff like cold fusion doesn't get a lot of funding. That's quite a leap

Expand full comment
Brian Doran's avatar

Appreciate your comment. The leap is that "via negativa" thinking, or empirical falsification, is how to actually prove things in science. However, no one wins awards for disproving the most theories, and the cost burden of having to disprove 1000x things to prove 1x things is prohibitive.

Expand full comment
Michael A Alexander's avatar

I don't think you can *prove* a positive. You can provide a negative, but demonstrating a pile of negative results doesn't get you any closer to a positive result. It seems it would cause one to give up. I'm a Ph.D. chemical engineer and spent a career in chemical & fermentation process support at a major bulk drug manufacturing site for a major Pharma company. I had this crystallization step that we moved from one crystallizer to another. We started to have occasional problems with small crystals. The normal cause of this is premature crystallization. You can study it by duplicating the problem in the lab and then trying a variety of fixes to find one that works. Premature crystallization usually happens because the crystallization conditions were imposed too quickly. The standard process involves adding the counter solvent (the thing that induces crystallization) slowly over five hours. Turbulence can also induce crystallization, so we use low speed mixing.

Back in the lab we could not get this crystallization to fail Every time we did it we got beautiful large crystals. We had the agitator spinning so fast that the whole crystallizer was rocking and rolling. Big crystals. We tried adding all the counter solvent as fast as we could pour it in (1-2 seconds) Big crystals. We did both. Big crystals. Vary temperature and other things I don't recall now, big crystals. Lots of negative results. We *knew* a positive result was possible since it was happening in the plant. We just could not make that happen in the lab. We threw in the towel. The problem wasn't critical, only some batches had small crystals, which were reprocessed, and the product was small volume. And then a couple of years later we discontinued the product altogether due to tiny sales volume. The problem was solved.

Finding a bunch of negatives doesn't get you closer to a useful truth. Finding a serendipitous positive result can you closer to a practical, "good enough" truth. We had another crystallization we were working with for a new product. We were trying to reduce a particular impurity and had gotten a good result with one in which the produce was dissolved in a solvent and crystallized out by adding a counter solvent. Subsequent experiments trying to optimize the process were giving us all sorts of contradictory results. One day the technician was getting ready to run a new experiment, he had dissolved the product in the solvent and had it stirring. He then went to lunch planning to add the counter solvent when he got back. When he got back, he found precipitate in the reactor. He brought it to Omar's attention who told me, and I looked at it. I said it must have converted into a solvate with a lower solubility (we had seen something like this happen in another project). The tech collected the crystals, dried and analyzed them. The impurity level in them was greatly reduced. Problem was solved. We did a straightforward crystallization with an embedded a stir period to simulate the tech going to lunch. And that's what we do in production.

What I am getting at is a bunch of negative results won't get you anywhere. You still need a positive result to build on. For example, you cannot *ever* show what something is by showing all the things it is not. This is because there are things you did not eliminate by empirical testing because they never occurred to you as something to test like my small crystal example. Or, it could simply not be possible to do certain negative tests. For example. it is often pointed out that Newton's theory of gravity is not the truth. This is wrong. It could not be falsified using the technology of Newton's time and so it was the truth for his time. When technology advanced to the point where anomalies were observed, Einstein developed a more comprehensive theory that explained them. Newton's theory falls out of Einstein's theory as a limiting case. Newton's laws still work in the realm for which they were invented. When one does ballistics calculations, treating Newtonian gravity as a true fact gives results that correspond to empirical reality (i.e. are true) every time. It's still the truth that Newton claimed.

Expand full comment