• Adam Cifu: Goodharts Law and Medical School Admissions. A significant contributor to scientific slop is “research output” as requirement for fellowship, residency, and even medical school admissions. And yes, I am also a fan of the law.
• Paul Teirstein for WSJ: The Gatekeeper Driving Doctors From Medicine. On board certification, which now seems to have some powerful enemies so consider me conflicted.
• Jonathan D. Cohen and Isaac Rose-Berman for NYT: Gambling. Investing. Gaming. There’s No Difference Anymore. “There should be a line of separation,” Alexander Hamilton [wrote] in 1792, “between respectable stockholders … and mere unprincipled Gamblers.” As a side note, it is incredible how much sports betting has exploded in the 15 years I’ve been in the US. One of many ways America is looking more and more like Serbia, where seedy gambling outposts are now more common than grocery stores.
• Brian Potter: More on US Pedestrian Deaths. An update of a previously linked-to post, which raises more questions than answers. Allow me to be somewhat conspiratorial: could worse hospital care combined with increasing organ donation contribute to this increase? There have been 3,200 more pedestrian deaths in 2023 compared to 2009 and 8,393 more deceased donor donations according to OPTN. And then there is this.

• Raghuveer Parthasarathy: Reading Like It’s 1937. The experiment was to choose only books published in a single year and read these in a short period of time. As expected, there were more duds than classics, and with not that many books coming out in 1937 it amount to very few good books indeed. This kind of experimentation does support the claim that we live in a literary golden age, and I’d say it’s because of a combination of high absolute numbers and us being better at detecting edge cases, not necessarily that the relative percentage of classics per year has increased. If anything, it is probably much lower because of all the AI-generated slop.
• Mike Taylor: If you’d built a “tool” that stupid, why would you advertise the fact? The AI tools described are indeed stupid, but remember that the marginal cost for advertising them is 0, and if you find them stupid you were never the intended audience. In that way they very much resemble phone scams and emails from a Nigerian prince.
• David Shaywitz: The Startling History Behind Merck’s New Cancer Blockbuster. A story from 8 years ago, but it was news to me that pembrolizumab had a very roundabout way of reaching patients. GLP-1 inhibitors had a slightly less circuitous route, but still took too long to get to where they are. Is this truly the best way to develop drugs? (↬Derek Lowe)
• Cal Newport: Is Sora the Beginning of the End for OpenAI? Probably not, and this is not really an edge case, but as someone who has trouble concluding blog posts (which is why you’re reading a list) I just wanted to pause and admire that last sentence.

The British crime drama Broadchurch came out in 2013. John Favreau’s food porn vanity project Chef was released in 2014. Despite both now being more than a decade old, in my mind they are still filed under “new things that came out that we missed because we had an infant in the house while also being medical residents”. It was therefore jarring to see how dated they both were, and for similar reasons.

Broadchurch deals with the murder of an 11-year-old boy in a small coastal community. Twitter is mentioned a handful of times, only in the context of breaking news. There is no Instagram or messaging apps: pre-teens email each other. The boy’s family is at a loss for how to attract national attention to the killing and finds the answer in a tabloid journalist. It all feels quaint, though admittedly I don’t know if that was intentional even in 2013 (from the edgy music and the oh-so serious tone of the show, I suspect not). I won’t mention a recent British show by name for fear of spoiling other, but if you’ve seen both you will now what is the clear parallel and how much things have changed.

Chef, on the other hand, is completely Twitter-dependent, and is arguably one of the first movies to use Twitter #MainCharacter dynamics as a plot point (Justine Sacco had landed a few months before the movie was released, and probably wasn’t even on Favreau’s radar). Twitter is shown in a completely positive light, and I can’t think of any other movie that has done that. It is also a good time capsule of the food trucks on Twitter craze. The early 2010s were the peak time for both, before culture wars killed one and covid the other.

So now I am inclined to see what else came out in that 2010–2015 period. Is it too early to be nostalgic for those times?

• Rusty Guinn: Anatomy of a Narrative Virus. A case study of our collective disregard for truth.
• Branko Milanović: Defeated by reality. A post-mortem to neoliberalism.
• Doug Muir: Five technological achievements! (That we won’t see any time soon.) A healthy dose of techno-pessimism.
• Helen Andrews: The Great Feminization. A just-so story of our current condition. Note that “only a woman could have written this” does not necessarily make it true.
• NYT: Nick Offerman Woodworked His Way to Playing the President. A portrait.

• Finally, someone has solved a real problem with AI! Whatever you think of the idea, please please please try using it.
• Cost of Convenience in Cancer Care. Indeed.
• Prostate screening. Indeed again.
• “A probabilist is to a statistician what a car designer is to a mechanic”. A Conversation with Nassim Taleb on the Risks of Cancer and Heart Disease.
• “You have such a stunningly superficial knowledge of what went on that it’s almost embarrassing to listen to you."

Always on the lookout for new blogs, I was happy to see a former leader at the National Science Foundation, Jim Olson, start one (↬Tyler Cowen). Based on the formal and didactic style I would say I was not the target audience for it, but it is better then nothing. It may also provide a convenient catalyst for my own thoughts.

For example: Olson’s most recent post is about the replication crisis. He points the finger on verification not being sexy enough for grant funders and academic journals, which is true. But if anything, having more people verify more and more claims in the ever-growing steaming pile of academese would make things seemingly worse, at least in the short term. This is the same kind of thinking that wanted to end medical reversals. You don’t want to end them, you want to make them unnecessary in the first place!

Now, fear of your claim being verified may frighten some researches from shooting from their hip, but unless paired with some sort of immediate punishment it would hardly make for a good stick. And what is preventing the person who made the original claim from demanding verification of the verifiers, and so on, and so forth, ad infinitum?

Olson also recommends more detailed methods, so that replication would be possible in the first place. This has already been implemented as anyone who had to fill out Cell’s never-ending STAR Methods can attest. Nature and Science have similar requirements, and some of them don’t even have a word count limit for the Methods section. Granted, many other journals aren’t as rigorous, but that should help you figure out which journals to follow.

So, instead of asking why we don’t have more people verifying claims, I would ask why we needed verification in the first place. Olson touches upon the core issue, mentioning “the time horizon problem”:

NSF grants run 3-5 years. Tenure clocks run 6-7 years. But scientific truth emerges over decades. We’re optimizing for the wrong timescale.

During my time at NSF, I saw brilliant researchers make pragmatic choices: publish something surprising now (even if it might not hold up) rather than spend two years carefully verifying it. That’s not a moral failing—it’s responding rationally to the incentives we created.

Of course it is about incentives. No amount of verifying will change that. People are chasing after tenure and accolades, not truth, and many a tenured professors shrugged their shoulders at the mansions of straw they had built over the decades. At best, they provided an easy target for a successor in the field to refute, unless of course there is a whole cabal of like-minded researchers protecting the dubious claims. But the default position is that these mansions of straw stay there, moulding and festering, side-tracking post-docs and spamming PubMed searches.

I have no clue what the solution may be. Maybe there is none and this is the equilibrium — let reality provide the final vote. But the status quo feels far from optimal.

• Derek Lowe: Where Do We Stand With “Liver-on-a-Chip” Technology?. It helps, but we are nowhere near replacing animal toxicology models with in vitro, let alone in silico assays. Now, that’s toxicology. I would have a much lower threshold for throwing out the window all those animal models of disease, particularly when testing any treatment that works wholly or in part via the host immune system. Those quickly turn into Rube Goldberg machines — good for intellectual stimulation, useless for inference.
• Brian Potter: Why Are So Many Pedestrians Killed by Cars in the US? A marvelous exploration of data that doesn’t point to a single cause but does more or less absolve phones while painting a big red “X” onto SUVs and (pedestrians taking) drugs. (ᔥTyler Cowen)
• Dan Snow: Secretive vendors are exploiting a free money glitch in the U.S. healthcare system. As much as it makes my blood boil to see this kind of profiteering in health care, ultimately Americans spend so much because they get so much, at least in terms of volume and technology if not outcomes. It may not be evenly distributed or even effective, but you get what you ask for. If anyone wants a different profile of health care spending they should look at some other people.
• Boone Ashworth and Kylie Robison for The Wire: I Hate My Friend. I saw these adds on the New York Subway a few weeks ago and wondered who on Earth would be crazy enough to wear an always-on microphone, and who would be delusional enough to think it could be a successful product. (ᔥJohn Naughton)
• Andrew Gelman: Uncanny academic valley: Brian Wansink as proto-chatbot. An article headline from last month asked what kind of an Age we lived in. “Despair” was my suggestion but I now take it back — the answer is clearly “bullshit”.

• Science: A cGAS-mediated mechanism in naked mole-rats potentiates DNA repair and delays aging. Or: why are naked mole-rats the longest living rodents, and not by a small margin? Seems important.
• FT: The quest to make young blood into a drug. Vampiric.
• NIH director embarasses himself on X. I am all for promoting healthy lifestyles, but one can do so without posting dubious charts from suspicious sources.
• All Derek Sivers’s e-books are free until October 18, prints are $4 each. All are highly recommended, particularly Useful Not True.
• The Drifter looks like a cool game for those who liked old-school point & click adventures. It’s in my Steam pile (yes, that’s how I’ll call it… ᔥRobin Rendle).

• Aidan Walker: now is not the time to ban phones. The Gen Z case against Jonathan Haidt, with whom I generally agree. Walker makes some good, if muddled points. In any debate about children, school and phones it is important to distinguish the extent of a ban (during class — absolutely, at recess — we can discuss the pros and cons), which kind of phones are to be banned (smartphones versus others, and nowdays they all come on a spectrum), and maturity of a child (there are some 12-year-olds who are mature enough to babysit and some who cannot be left alone by themselves, and in the same family!) So it is an issue with some subtetly and since both social media and public policy ale places where subtetly goes to die I do not expect these questions to be answered any time soon.
• Ted Gioia introduces Jared Henderson: The Honest Broker Launches an Interview Series with Our First Guest Cory Doctorow. Henderson being the host of the interview, posted on Gioia’s Substack page. Doctorow’s book Enshittification has just been published and even though I suspect it is one of those that could have been a blog post it is indeed on the pile. Doctorow’s closing recommendation, Letters from an Imaginary Country, will also end up there when it comes out.
• Casey Handmer: Career Development Guide for Job Seekers. Advice for academics and government employees who want to switch to industry, from the perspective of an engineer. Fortunately biotech is more forgiving or else I would never have made the transition but I imagine things are even more competitive now than four years ago so all this is helpful. More boradly, academia and industry have so many “false friends” — conventions and processes that have the same name but are different and in some cases diametrically opposed — and the interview stage is just the tip of the iceberg.
• Scott Sumner: Fat tails. It is more about slippery slopes than fat tails, but Sumner has never been one to write a good headline.

• José Luis Ricón: Systems Biology: understanding beyond genes. With an eye towards aging, but broadly applicable to much of medicine. As ever, there is a relevant xkcd comic on the topic (ᔥDerek Lowe).
• L. M. Sacasas: The One Best Way Is a Trap. “So, once again I invite us to ask a simple question: Is there, in fact, “one best way” in every realm of experience? And even if there were, at what cost would we discover it? And what would we gain? Might it be that in the course of pursuing the “one best way,” we would lose our way in a more profound sense?”
• Anthony DiGiorgio on X: Imagine we mandated everyone have car insurance that covers new tires, oil changes, and wiper fluid. I think you can guess where this is going, and he is not wrong. The cost of health care is skyrocketing because we are always finding new ways to spend money, and isn’t America a society of maximizers?
• Kyla Scanlon: Who’s Getting Rich Off Your Attention? Robert Reich responds, on Bluesky.
• Bonus: Tyler Cowen doubles down on min-maxing the society in his brief comment on Sora 2. “Still, from my distance it seems quite possible that the “slop” side of the equation is a simple way to fund AI “world-modeling” (and other) skills in a manner that is cross-subsidized by the consumers of the slop.” That way may be simple, but also quite diabolical.