Jake Seliger has died

Due to lack of evidence for chemo having any benefit for his cancer with devastating impact on quality of life he was instead put into several months of proton radiation therapy, a Keytruda clinical trial, and then surgery all within about 6 months. After about 7 hours into surgery we got a call from the lead surgeon who told us they were unable to get clean margins due to proximity to the bone but spent an extra several hours removing all visible cancerous which was the best they could do before they had to hand it over to the reconstructive surgeon. That was a horrible gut punch and I can still picture the expressions on my brother and mother's face as we processed what we were hearing.

He survived the surgery, recovered impressively quickly, but we were all terrified waiting for the first 3 month follow up scan expecting the worse given the lack of margins and potential of being extremely fast growing. But it was clean, and so was the next 3 month, and he's now through his 7th or so clean scan I believe.

I had read so much about survival rates but the data for his specific cancer was all so poor plus confounded by his tumor being huge (data says that's very bad) but no systemic spread or to adjacent organs (data says that's very good), no clean margins (bad) but no remaining tumor tissue (less bad).

Eventually I realized all the time I spent digging through PubMed for any scrap of a data point hadn't actually given me any insight, reassurance or had any predictive value whatsoever. Which was tough to accept given how my brain is wired. So all I can really do now is be grateful for the extra time he's already had and the possibility that seemed so unlikely pre-surgery that he may even have many more, but the future will always be unknowable either way and not worth obsessing about.

(I'm wanting to do a proper write-up of that but am still organising my own thoughts. Meantime I can just strongly urge people to read the piece.)

The absolute opposite of my experience. Every one of the fields I've gained expertise in over the years indicates to me that more human effort would pay not just dividends but increasing returns. Particularly so with research. The biggest thing I learn when I gain expertise is just exactly how much we don't know.

I independently reinvented my entire field of research by accident, because I didn't know it existed; and I'm pretty sure I did a better job. There are a few puzzle pieces and key insights it probably would've taken me a few years (or decades) to discover, without which I couldn't demonstrate (or, if I'm being honest, know) the immense superiority of my approach, but I'm pretty sure the majority of the field is obsoleted by the ideas I've had.

Oh, look at me, I'm so clever… right? Good guess, but actually no. When I ask the right questions to my peers (outside the field), they usually propose a similar approach. (The main differences can be attributed to the fact I've thought about this for years, and they've thought about it for minutes: a few of the obvious things don't work, but then things get elegant when you replace them with ones that do.)

That must mean the time's right, then? I did suspect it was this… until I found a few dozen publications by a widely-respected expert, from half a century ago, talking about a (slightly underdeveloped, idiosyncratic) version of the approach like it was common sense, remarking that people were a lot more receptive to this idea than they used to be, but still it was not being adopted.

A minor application of this approach to a different field would completely revolutionise it (even moreso than it does my field). I briefly fantasised about doing that, before dismissing as "something to investigate later" (i.e. "a childish fantasy I haven't found the holes in yet"). But not only does that application work, it was implemented and trialled. The results were published 26 years ago, in a paper that concludes by confusedly asking why nobody was doing this. After reading the paper (which I found completely by accident while looking for something else), I felt much the same way. (Still nobody is doing it, in case you're wondering. It's been cited six times – once in its field, two years after publication, and five times in other fields in papers where it isn't really relevant.)

In physics, parts of chemistry, and parts of mathematics, there is room for revolution. I'm not sure any of the softer fields (counting mathematics qua philosophy as soft) are particularly good at being revolutionised. (Economics is particularly bad: nearly every economist knows that economics is wrong, and yet it persists.) And it's not because new ideas are slow to be adopted: at least 10% of my field's practitioners dropped everything to try to solve its Hard Problems with LLMs. (Those I know, I talked out of this, by describing – with non-rigorous theoretical arguments – what they would find. After years of work, the best result I've seen is only slightly better than the bound I predicted.)

I'm sure there's some way to solve this problem, but I don't know it. I wouldn't be surprised if the solution has been identified, written down, published, forgotten, independently reinvented, tested, found to work, published again, ignored…

I don't say this to cast aspersions on science, or the academic system; nor to blame people or institutions for their failings. This problem is well-known. It's not a matter of people being stuck in their ways: if anything, academics are too credulous, too willing to believe that things work which don't. But I don't think adjusting the "credulity" dial, or the "speed" dial, or the "try novel approaches" dial, is a solution. Science, as it stands, cannot be done faster, because the system doesn't know how.

All I'm saying is, watch out for this. Support people who think they have a better way, check the literature for interesting things (and then confirm them, if you can). Question not whether your current approaches are good, but what they actually do – and ask whether that is really what you want to be doing, or a proxy goal that everybody's lost sight of.

And if you find yourself along with your peers, dismissing something out-of-hand without having tried it, maybe don't lend your voice to the snubbing crowd. Maybe it's bogus, maybe its proponents are wrong about its merits, but it might have merits. We can't afford not to know.

Seinfeld asked something like 'How do you deal with constant annoyances, all the time, when you're president?'

And Obama replied 'I expect it's similar to what you do -- you fall in love with the work. Sure, sometimes it's painful, annoying, backwards, foolish, etc. But in the end you fall in love with the work, and it saves you.'

That defined purpose in a way I'd never thought about it, and probably undergirds every religion.

It also made me try to open my heart more to people dealing with chronic pain or mental health issues, in terms of their subjective effort. Objectively, it may look like they're just doing {normal thing}, but subjectively that may be requiring 10x or 100x effort from them. And that effort (the work) deserves its own respect, independent of outome.

I'm not sure how trust would be eroded by the failure of an experimental drug to treat a terminal case where no other drugs are working.

Desperation can cloud even the most rational of minds.

Yet euthanasia is a very fraught subject in the land of the free. The country may be coming less religious, but the puritanical values remain; every life is precious, even one that's in unbearable, irreversible agony, and those who suffer probably deserve it.

Just because one is terminal, it does not make one an expert in medicine or pharmaceuticals. So it shouldn't be used an exception to the processes in place.