Posts in: science

I was also given some language changes to consider, so I might sound less like chatGPT to reviewers.

Being accused of using ChatGPT to help write a manuscript was not a second-order effect of LLMs that ever came to mind, but of course it would happen. Yikes.

February lectures of note

A sentence to ponder:

Like a disease, pregnancy is caused by a pathogen, an external organism invading the host’s body.

Can you guess if it comes from a dystopian science fiction novel or a peer-reviewed research article?

Adam Mastroianni writes about the declining trust in scientific institutions:

I, too, would like to beat the charlatans and the terrorists, which is why I want to do better than, “Don’t trust those guys—they lack the proper accreditation!” If that’s all you got, people shouldn’t trust you. Instead of arguing from expertise, you should use your expertise to make better arguments.

As much as I would like to proclaim that this was my thinking all along, and that institutional decay was why I went for medical residency before completing a PhD, well, I can’t — it was a wholly different set of reasons. But it seems to have been a good decision, so I’ll allow myself a pat on the back.

Today I learned — from my 4-year-old, no less! — that house flies are among the top 10 non-bee pollinator families. Numbers 1 and 2 are also in the fly family. Is it time to retire our swatter?

A proposal to reform the NIH intramural program that makes a whole lot of sense. (ᔥTyler Cowen who recently linked to another new-to-me science blog that is quite good. A Blogroll update is long overdue.)

Five somewhat esoteric mental models I have found useful

In anticipation of the new edition of Poor Charlie’s Almanack arriving by mail — alas, the new delivery date is February 15 — I have been mulling over the more unusual mental models I’ve adopted since first reading about the latticework. The latticework is a mental model of its own — a meta-mental model, if you will — and you would do well to adopt some all of Mungers. The five I list here aren’t the models you will find in the Almanack, but I would not have identified them as such and remembered them were it not for Munger’s wisdom. The links are to Wikipedia and journal articles, for now, but I hope to write a detailed account of my own for each, you know, once I get around to it.

  1. C.H. Waddington’s probability landscape, which has applications far beyond developmental biology, where it was first introduced;
  2. Activation energy, especially as it relates to motivation and administrative inertia;
  3. Epiphenomena as real and tangible things and not nuisances to be brushed aside, which I realized via both Hofstadter and Girard;
  4. Phenotypic plasticity, which we take for granted at the level of an organism (in an environment of caloric abundance mammals will get fat) but not so much at higher and lower scales, i.e., cells and societies;
  5. Carcinization, the nature’s attempt to evolve every crustacean into a crab five separate times, or, as I like to see it, nature’s way of telling us that hyper-optimization is seductive but ultimately a dead end — and again, mind the scale.

These five are interconnected in interesting ways, and if you arrange the arrows just right they do for a mini-lattice. Kudos to Munger for finding the right term. Munger discussed his own mental models in detail in the Almanack and they form a larger, but still loose, network.

The web of mental models is, of course, Nassim Taleb’s Incerto, built so densely and interconnected so profusely that Branko Milanović was right to laud is as a new type of writing. And in fact of the five models I listed, the first one — Waddington’s probability landscape — is a neat bridge between Taleb’s investigations and the other four. But that is a discussion for a future time.

January lectures of note

The calendar is full again. Rejoice?

I tend to add “to all who celebrate” because the Serbian Orthodox Church is still on the Julian calendar, which is off by 13 days. Most orthodox churches are aligned with the Gregorian calendar and have been so for the last 100 years thanks to a revision proposed by — drumroll please — the Serbian Orthodox Church. Having a good idea and not following through because Russia isn’t on board is a typical Serbian thing to do, so points for consistency I guess.

Today’s Washington Post has a good write-up on how genetic engineering of the near-extinct American chestnut tree to make it more resistant to infection went wrong:

After he enlisted the help of Ek Han Tan, a geneticist at the University of Maine, to analyze the chestnut’s genome, they made their discovery this fall: The plants they were working on were, in fact, not Darling 58 trees.

Instead, they found they were working with a different chestnut line — called the Darling 54 — where the gene was inserted in another chromosome entirely, potentially corrupting one of the tree’s existing genes.

In a phone interview, Newhouse, the SUNY ESF director, acknowledged the mix-up but said he wasn’t sure what transpired.

“As far as exactly how it happened, we don’t know,” he said. “It must have been a label swap between these two trees that we were working with at the same time” in or around 2016.

The brilliant minds who think engineering mosquitos is a good idea can’t foresee that even a seemingly innocuous clerical error can lead to disaster, never mind the second-order effects to nature if your project succeeds. Whoever’s read Taleb’s Incerto (or some of his tweets) knows better.