I work as a scientist. I'm not sure I agree science is a strong-link problem. Rather, it contains both strong- and weak-link problems.

A lot of science work consist of what my scientist mother calls "bread work" - validation of existing techniques, building a general corpus of knowledge, maintaining databases with useful information and so on. These is weak-link work. A bad quality database will undermine any research which is built on it and not aware of its limitations.

A good recent example may be the stem cell research field, which was (and perhaps still is?) bogged down by being built on bad fundamentals.

But yes, more broadly speaking I do agree that science funding should be built to encourage more risk of failure, more elitism, and more consolidation of large project to be able to "punch through".

I like the weak-link/strong-link distinction, but you need to be careful what problem exactly you are talking about. "Science" is not a problem. "How to distribute funding to scientists" is a problem. And here the weakest links actually matter, because you only have a finite amount of funding, and you don't want to waste it on the weak scientists, so you have more for strong scientists.

Another way of phrasing that problem is, "what actually is strong science, and what is weak science"? Over a few decades, that question usually plays out correctly, but even a few years is often not enough to tell, even for specialists, especially for truly new and original science. As an example, right now, would you say abstraction logic [1] is strong or weak science?

And of course, venture capital funding is also not a strong-link problem. You can waste money on some weak startups that you thought were actually strong, but do that too much, and your fund goes under.

[1] http://abstractionlogic.com

One purpose of science is to provide the rest of the society, who are not scientists, with reliable insights, ideally with actionable advice on how to solve a problem. If you treat science purely as a stong-link problem, the burden of quality control lays with the consumer of science. Peer review attempts to lay it with experts. That approach is nowhere near perfect, but is the best we have. And it scales much better.

For all we talk about "learning" being the purpose of a college education, the true purpose for students is accreditation: to get a degree to get a job. Take away the degree and very few people will attend college.

This article (and the few P&T committees I've sat in on) makes it sound like faculty have created the same environment for themselves. "Just publish something" so you can have numbers on the board, so you can promote. Remove publishing requirements for tenure and I'm guessing the volume will drop off precipitously. Journals are accreditors for tenured positions the same way colleges are accreditors for the labor market. (Or so it seems to me!)

Metaphors matter[1].

"Strong-link problem" is an oxymoron.

Any kind of chain is a "weak-link problem".

If an individual scientific work stands on its own, if it's not a link that is part of a chain, it can't be a strong link. If it is a link, than it can also be a weak link, making science a "weak-link problem".

In The Study of History, Lord Acton referd to "shining precepts which are the registered property of every school"[2].

> learn as much by writing as by reading; be not content with the best book; seek sidelights from the others; have no favourites; keep men and things apart; guard against the prestige of great names; see that your judgments are your own; and do not shrink from disagreement; no trusting without testing; be more severe to ideas than to actions; do not overlook the strength of the bad cause of the weakness of the good; never be surprised by the crumbling of an idol or the disclosure of a skeleton; judge talent at its best and character at its worst; suspect power more than vice, and study problems in preference to periods.

The point that this article makes doesn't seem more then this "shining precept" that was a cliché even in 1895, but with an absurdly confusing metaphor.

[1]: https://en.wikipedia.org/wiki/Metaphors_We_Live_By

[2]: https://en.wikiquote.org/wiki/John_Dalberg-Acton,_1st_Baron_...

Further discussion: https://news.ycombinator.com/item?id=41993801

It’s great to put names to these concepts but it’s a false dichotomy. It doesn't follow that just because you need to eliminate the worst, that you need to also encourage the best.

There are certainly problems where you want to eliminate (or minimize) the worst outcomes while also facilitating the hit-it-out-of-the-park game-changers. Solving poverty shouldn’t preclude providing people opportunities to get rich by changing the world for the better.

There are also problems where it doesn’t matter what happens at either end. Show up to marathon for example and no one minds how well you do - just to give it a go is great! There’s no certification as a marathon runner and there’s also gold medals for the fastest marathon runners.

All in all this article is arguing that problems can be plotted on a single axis of strong or weak problems, but it’s more like a 2x2 matrix (or a 2-dimensional area if you want gradations.)

"When you’re looking to find a doctor for a routine procedure, you’re in a weak-link problem."

The supplies to scientific research as well.

Scientific publication does not have the gatekeeping that medical industry does to ensure that routine doctors can do most routine procedures without harm.

The gatekeepers of modern "Science" are explicitly harmful to novel insights. "Science advances one gravestone at a time" is not a new observation

Previous discussion (55 comments): https://news.ycombinator.com/item?id=35712694

I read this and thought, I wonder where LLM's fall in the range...I think about the quality of LLM's more than I want.