[anu7df]

Yes, the university taking 60-70% is almost highway robbery, but it is hard to completely disassociate from them. For one, the university can through economies of scale reduce the cost of setting up and running experimental facilities. Moreover, one usually finds motivated researchers in academic settings (I am anticipating dissent here, since that could be because of the funding flowing only through Universities). However talented and motivated researchers usually find other work situations which while to fully aligning to their erstwhile research interests, pay for a good life. Those who really have the passion that exceeds the mundane normalisation of drive ( also called life per some definition) usually seek refuge in the academic setting. So, yes if we can set up labs like the old Bell labs or find a way to replicate the tenure model of universities, taking funding away from the univ mechanism and giving it to individual unaffiliated researchers won't be practical. But even if that can be achieved, you have just replaced universities with a slightly different structure. Better option is to trim the fat and realign priorities at universities, which is also easier said than done. Cushy administrator jobs, student housing mansions with 5 star facilities, large stadium colosseum all should go.. But systemic reasons, societal perception, need to justify legacy admission all factor in here as strong headwinds.

[j7ake]

Does anybody really know where the overhead money of top places end up ?

Some institutions (eg famous ones) have near 100% overhead. Overhead sounds reasonable for example to maintain IT and core facilities, but I have a suspicion that a lot of it is wasted in a black hole of unknown function generically labelled “admin”

[godelski]

Most of the overhead goes towards admins. But it is also important to remember that many universities have multi-billion dollar endowments. It's been explained to me that "I don't know what endowments are for" when I've pointed out that my (not top 10) university could indefinitely pay for all US student tuition (and 90% of foreign) on the interest of our endowment at a 5% yearly growth. So I'm still not sure what endowments are for considering that there are 15 schools with endowments that are over 1m/student. So yeah, I still have no idea what endowments are for.

https://www.collegeraptor.com/college-rankings/details/Endow...

[gumby]

In the us the size of endowment affects US News college rankings. So it’s not to be touched!

[kmmlng]

I'm not in the US, but from what it sounds like, the situation isn't much different where I live.

A surprising amount of overhead goes to employees in accounting who perform some black magic so that everyone on staff is payed by some combination of projects. Whether someone actually works on said projects is irrelevant of course.

[dmurray]

100% seems reasonable for other businesses.

If I had a factory with 100 people assembling my product, or a sales team with 100 people selling it, I'd be quite happy to only budget for the cost of another 100 people to pay my rent and run my IT, payroll, accounting, janitorial services, maintaining my machinery...

I get the impression that modern universities have five or ten other employees for each teacher/researcher, which assuming everyone gets the same pay implies 500-1000% overhead, or subsidy from sources other than grants.

[ok123456]

The problem is that they've created a whole hierarchy of useless deans and other admins to eat up all that overhead to justify it.

[mattkrause]

100% is a bit of an exaggeration. I’ve heard of one place with a 100% overhead rate and it was for something very specialized (marine research vessels, which probably do eat money). Everywhere else I know of is near 50 percent, and this includes “famous” places (Yale, McGill, MIT, Georgetown, etc).

The overhead rates are supposed to reflect the University’s actual costs and are negotiated with the government, so I suppose you could try to make a FOIA request for them.

[WSchneller]

Most universities have their negotiated Facilities and Administrative (F&A) rates posted online via their Sponsored Research Office. For example, Harvard charges a 68% indirect rate and Stanford about 58%. The public R1 institutions I’ve done business with charge 40-50%.

This doesn’t paint the whole picture though as sometimes additional overhead costs (eg. healthcare benefits) can be budgeted as Fringe Benefits for personnel working directly on the project. In that situation, the F&A rate would stack on top of those costs.

[logisticseh]

If this is true then the universities will win the grants anyways, so why not adopt my proposal?

(It's not, especially for disciplines like CS and math.)

[anu7df]

If your proposal is to make all grant applications blind with respect to grant writer information, it is an idea that has a lot of unintended side effects. Firstly, not every field, in fact, most fields are not Math or CS by definition. It is certainly the most popular in this crowd, especially these days, but without proper facilities most of science cannot be done. So, making an application for grant anonymous, you have dramatically increased the risk of nothing useful coming out of the proposal. A talented independent grant writer (who may or may not be a great scientist) now has a much higher chance of getting the award. But if he doesn't have the ability to execute that great idea the project will fail. Most often idea is aspirational with no detail, the breakthrough comes in only during execution. Execution requires a team and facilities. This is why things like MURI (multi-univeristy-research-initiative) were created. Also, a lot of the grants are earmarked for early career researchers, without a lot of experience in proposal writing, which also will go away if the awards are anonymous. I am sure there are other points I am missing and also there may be ways to remove the shortcomings, but I don't see a clean and easy path forward with this.

As for math and CS, I am quite positive at least in Math most work is done by individual professors/post-docs/students without much involvement of grants. In fact, a lot of universities do reduce the overhead charged to purely intellectual pursuit with no major facilities usage kind of departments. Not sure all math or cs research will qualify for that because of the HPC requirements, but the option is out there. Universities as signals for funding and research success really appears to me like the best of a bad option set.

[friendzis]

> So, making an application for grant anonymous, you have dramatically increased the risk of nothing useful coming out of the proposal

To me this sounds not as a critique of the proposed system, but rather defense of the current system.

Science is underpinned by the scientific method which essentially is "form hypothesis -> create controlled environment -> test hypothesis". I do not want to start debate around expected outcome rations, but negative result (hypothesis being false) is perfectly normal outcome of the scientific method. Current system tries to defend against rent seeking by being heavily biased towards positive outcomes. This encourages heavily incremental research, which a bit paradoxically can be also seen as rent seeking. Furthermore, it also encourages low quality science (like p hacking and similar stuff) and discourages publication of negative outcomes, possibly leading to hitting the same dead ends multiple times.

Blind grant proposals could reduce bias towards "past performance" decreasing ratio of positive results, and encourage "riskier", more brave hypotheses to be formed and tested.

Past performance as a metric in grant application encourages research optimization. Some may see this as a good thing, but it discourages early-to-mid career researchers (and even late in their careers) from pursuing something non-obvious. Safe science is more often than not optimization and edge case/condition testing.

[thfuran]

I think you're missing the point. Negative results is not the issue. If your hypothesis involves NMR, you need access to an NMR machine to test it. Most people outside of a university don't have that and so would fail to execute.

[friendzis]

Ok, apparently I did miss your point a bit.

However, your argument is still orthogonal to grant blindness itself. Mere affiliation with with research institution does not guarantee (although highly increases the likelihood, I can agree on that) access to certain research instruments.

This is the thing I have tried to explore in my previous comment. Current model practically revolves around researcher "fame" and grant board "knowing" their [lab's] ability to execute technical part of the research. This creates some perverse incentives. Naturally, grant boards become more inclined (consciously or not) to favor certain institutions and researchers, giving more grants, further increasing "competitiveness". Researchers become disincentivized to publish negative results and are incentivized to resort to data hacking and irreproducible science. This sure does give rise to the OP's mentioned Twitter prof or "trust me bro" science. There is much more politics in science than we publicly admit.

Current process does not actually evaluate grant proposal on technical ability to execute, this is left for institutions "underwriting" proposal. We could start evaluating technical ability regardless of whether grant applications are blind or not. If peer review meant "peers got roughly similar basic results in basic reproduction" instead of "someone in the field looked at the final draft" I guess we could have blind proposals with advantages of current system without its major political disadvantages.

I can agree that open process does help filter out researchers underqualified to do the research in question, but it does come at a certain cost. Ability to weed out underqualified teams does not necessarily require current process, i.e. this can be achieved by other means.

[texaslonghorn5]

Just a slight clarification/correction/question: as far as I can tell, M in MURI is for multidisciplinary. That means the DoD is seeking interdisciplinary research (and maybe intra-university collaboration), not explicitly inter-university work -- is this correct?

[anu7df]

There may be separate multidisciplinary initiatives, but "Multi" in MURI definitely stands for inter university initiatives. All the MURI grants I have seen (not many and long ago to be honest) were multiple universities and one area of research.

[rsfern]

It’s “multidisciplinary”, but funded programs are almost always large multi-institutional teams https://www.defense.gov/News/Releases/Release/Article/295323...