Hacker News Reader

Comment by bmitc | Hacker News Reader

> Which projects to provide with funding?

There's a somewhat old adage (?) that says: "Don't fund projects. Fund people."

The gist of it is that by funding projects, you're funding a somewhat myopic view in that anything discovered adjacent to the project is now out of scope of the funding. By funding people with interesting ideas, you get a much more creative and free-flowing research program.

View on HN

I've had a pretty successful research career, but it's very rare I've had a project that actually delivered what I said it would on the funding proposal. Something more exciting has always come along while we were working on the original topic, and I've always gone with it. We've ended up significantly delivering better research than the original project would have.

And in 30 years, no project officer from a funding agency has ever complained that we'd delivered really good results but they weren't the results we'd promised. They're generally happy something good came out of the funding, because frankly, most funded projects don't produce much of value. Very occasionally (mostly on EU projects) we've had to write up the results in a project report with the original section titles that no longer match the new content to tick some boxes and keep the funder beancounters happy.

The amount of shifty money practices I've seen so far is astonishing, all in the pursuit of more science per proposal. Ideally you use some of the funding from the previous project to do some exploratory research somewhere very different and then write up a proposal around it, but since you already did some of the work there'll be money for more research and instruments and things and maybe fun ideas that students have.. creative accounting all around. I'd be appalled if it wasn't in the name of science.

I think many people do that - it's very hard to write up a compelling proposal without having spent some time working on the idea to see if it's viable, but there's often no funded way to do preliminary work on an idea. If you don't do something like what you describe, it's nearly impossible to branch into new areas, and that makes it nearly impossible to do good research. This is another problem with funding project proposals.

But what I was describing was slightly different - once the proposal is funded, many PIs feel compelled to deliver what they said they would, whereas in the two years since you wrote the proposal, the world has moved on, you've learned things, both from your early work on the project and from others, and more promising avenues are now possible. If you feel compelled by the funder to continue in the original direction, you'll very rarely deliver good research.

You shouldn't be appalled: grants that force you to work on a precise area with no flexibility are what's appalling. Research just doesn't work like that! The current grant proposal process is the creation of administrators and politicians, not scientists.

Many of the most critical research results came out of some random hallway conversation, as a response to some new result that didn't exist when the research plan was formulated, or because someone had a brain fart. If at each point someone had put up their hand and said "sorry, scientist, that's not on your current grant roadmap" we'd probably still be dying of preventable diseases and lighting our homes with gas fixtures.

Preferring people over projects creates a situation where you are carried by your status rather than your work. If someone had a successful project, it doesn't necessarily mean their next project is going to work. A total newcomer could potentially solve a very important problem (but get their funding denied because they are still not the right "people" to invest in). This also makes it harder for new ideas to propagate.

Funding should be a mix of track record with project consideration, not either exclusively.

No, we said fund _people_ not status. Someone’s status is their job title etc. that should be irrelevant in funding decisions.

Funding projects is a terrible idea that you should keep at a minimum. If you fund projects you will end up with the BS we have now - marketing in the form of grant proposals.

I think a really helpful analogy for how science should be funded is to think of how angels and VCs fund early stage startups - I mean, how they really fund them. Which is: is this a promising area to startup within? Is this a great team? And that is pretty much it.

The downside of this approach I think is that personal biases can creep in and create a lack of diversity. You can counter that with pots of funding that directly target diversity.

Fund people, your assessment of their ability, their team, and their area of interest.

The question then is, how do select the people. In VCs, like you said, you'd look for a great team. What does that mean? It means mostly one thing: track record. This is what I meant by status. It's not your title. In academia, very similarly to tech, some people are rising stars, and they can get funding for anything they like. I'm just saying, consider the project and not just the person.

I don't think that's accurate. One of the things that make incubation and choosing which startups to fund so difficult is precisely the lack of track record. According to Graham himself, when everyone involved with a startup are new to the scenes they're invading, you instead end up relying on such nebulous concepts as "imagination", "naughtyness"[1], or "earnestness"[2].

[1] http://www.paulgraham.com/founders.html

[2] http://www.paulgraham.com/earnest.html

Einstein completely transformed physics with some stand-out papers - most written as a patent clerk - and then spent the rest of his career producing almost nothing of interest. He commented on other work at conferences and in letters, but there were no more huge breakthroughs.

In modern academia that would be a terrible record.

But giving him tenure was a smart bet, because he might have produced more.

It's the Sabine Hossenfelder problem. Should physics fund incremental research which is a fairly safe bet in employment terms? Or should the money go on talented and creative researchers who may waste most of it - but one or two may produce something transformative?

Academia is heavily slanted towards the former approach. Because academia is now a business and has the same bureaucratic and corporate values as other bureaucratic corporations.

This is terrible for original research, because smart people need to given a free hand to follow their intuitions and interests.

Being able to afford to explore, play, and potentially fail would transform physics.

Because in reality academia is producing a lot of failure anyway, without the upsides of transformative new insights.

> almost nothing of interest

> terrible record

> might have produced more

You do not seem to have much knowledge of his publication record.

Here is a list of 272 journal articles by Einstein and coauthors, the majority of which would certainly be considered peer reviewed by modern standards.

https://en.wikipedia.org/wiki/List_of_scientific_publication...

From just these papers, among the things you put into the bucket of "not huge breakthroughs" were (in small length scales) Bose-Einstein statistics and Bose-Einstein condensates (notably his 1924 and 1925 papers), his atomic emission and absorption theory (the "stimulated emission" in laSEr), and his paper with Podolsky & Rosen and (in large length scales) General Relativity, Einstein gravitational lensing, the Friedmann-Einstein and Einstein-de Sitter expanding universe cosmological models.

Actually, Einstein's 1935 paper along with Boris Podolsky and Nathan Rosen basically highlighted the existence of entanglement in Quantum Mechanics. Of course Einstein was trying to show that QM was wrong via the EPR paradox but although reality has since proven him wrong, it was still a major paper written decades after his more famous work.

I don’t disagree with your main point, but I’ve actually seen Einstein cited as someone who had a long scientific peak. He first transformed physics during the 1905 miracle year and then again in 1915 after completing the theory of general relativity. From what I understand, special relativity was built upon the work of Lorenz and Poincaré, and likely would have been discovered only a little later without Einstein, but general relativity was a staggeringly original achievement. The University of Zurich’s bet paid off more than well when they granted Einstein tenure in 1909. His best work was still in the future.

I agree with you. Einstein published general relativity when he was 38-ish around 1917 or so, over a decade after his miracle year in which he revolutionized or kickstarted several domains in relativity, Brownian motion, and quantum mechanics. Even in the year of publishing general relativity, he basically kickstarted the principle behind lasers.

I mean general relativity is one of the most tested and verified physical theories, and it was practically done all by Einstein himself. It’s a hallmark of a theory. That alone is worth everything, and yet he was also the genesis for several other fields, both before and after general relativity.

Einstein is not a great example for this discussion. Even his “failures” are useful contributions. Many here are missing the point in that it is failures that we should not be afraid of funding, within reason of course. I.e., it’s okay if a promising idea “fails”. It is input to the broader research community and questions.

Maybe the trick is to do the y-combinator of research?

In the startup world it is widely believed that a bit of hardship is needed to show dedication.

1. Make the application process much simpler

2. Support with smaller amounts, individual subsistence salary

3. Provide the right placing in the right environment.

What you're describing there is called grad school

Einstein is really a great example for when your past success does not translate to future endeavors and I even considered mentioning him in another comment.

I know a few institutions where exploring and playing is a standard procedure, however they aren't many. And even this has a limit because science is expensive. So you need a way to know which silly ideas to pursue and how to best utilize your resources and budget.

Nah. The big 4 were written in 1905 . It was like in the 1920s or something that he worked on general relativity and wormholes.

Everything else I’m not going to read because it’s just bleeding heart nonsense.

Economy is always finite, sadly. Sometimes it may be just bad administration (inefficient, low funds, etc), but then you have to deal with politics, and thats another whole show...

There’s a subtlety there. It isn’t suppose to be about who you are but what you’re about.

Stopping caring about whether projects work or not is practically the point. Research isn’t about only tackling things we know will work. That’s effectively engineering and not research.

A lot of the funding today is already about status. There just needs to be less funding given out to people based on status and funders’ pet projects and more funding based upon the people and proposals. Solid people and proposals should not be turned down via lack of status or not fitting into some box of “this will work”.

A lot of topics are lumped into “that will never work”, but it’s often the case that they were never tried or even explored.

Agreed.

Just before I read your comment I came up with that idea "there should be a select few people based on their academic merits (whatever that means/however good that will go) that will have carte blanche."

Then I read your comment. Needless to say, I wholeheartedly agree.

Reply on news.ycombinator.com