What a $500k grant proposal looks like | Hacker News Reader

What a $500k grant proposal looks like

austinhenley.com

293 pointsazhenley130 comments

You know, the thing I always found extremely frustrating or non-transparent about the academic grant process (and it leaks over into government / NSF / etc applications by the nature of who evaluates those: academics again) is how much unofficial or unspoken criteria come into it.

You get extra credit points if you're active in visiting other universities and giving talks on the annual job circuit. You tend to do well if you advisors are well-connected. You have to sprinkle a bit of fad-ish topics into your research if even if it's a stretch to do so.

I suppose that it has always been so. And now looking back, I don't find huge fault in that you should be able to connect your research with larger ideas. Or that people in richer environments with more information are generally more successful.

But having tried to be on the faculty track before, from a middle-tier university, I never knew these things. Or the professors I was around never told me those unofficial factors. And I thought it very unfair that the judges allowed those factors to creep into their consideration without making it clear to everyone that those were important. If giving talks in person is important, tell people at least.

I naively (I guess) thought that the strength of what you wrote in the application/grant document dominated. Maybe it does for some fields or if reviewers are diligent about understanding the blinders they wear and their availability bias.

Well, anyway, I left the field all the same. No regrets, but I wish these fields would improve their explicit processes about it.

I do note with some amusement now in industry, that basically I am able to approve spending of $500k for certain things with approximately a 30 minute conversation with leadership (with appropriate justification and documentation)...

I went to a grant writing workshop sponsored by the university. When I got there the entire thing, and I mean entire thing, was about buttering up the grant review officers, how to network with people at the agencies, how to work your way into being their friends, and so forth.

Not a single thing about the writing process, or how to frame your ideas, or anything like that.

I know that's not how grants are always awarded, but as a junior faculty member it's hard to know what to make of that, when a university-sponsored workshop is structured around those issues.

It's difficult for me to wrap my head around this all most of the time, to convey to an outsider what bothers me so much about the whole thing, or to even articulate it to myself. I can make comparisons to certain things (Theranos, cryptocurrencies, or economic bubbles) but it's very abstract and fuzzy and imperfect. So much of it seems empty to me, or false.

In contrast, ours focused almost entirely on polishing and refining our specific aims page. Eight weeks on what was effectively perhaps two pages of writing. Draft after draft, peer review, etc.

Landed said grant on first submission.

This is impressive. Did others in the workshop have same experience as you of getting grant on first submission too? If so, the workshop can get credit, else you deserve the credit.

Candidly, my hit rate isn't high enough to assume it was me ;)

I didn't keep track of most of my cohort from that program, but the ones I have had been fairly successful. Mostly, I think the benefit was structured time and an available "red team" to read it (one contract research organization I've worked with had one of these, and it was an amazing experience) etc.

When I went to one of these things, it was pretty much entirely about how to scope the work and how to write the proposal. Eg, for a big ERC grant, you need to propose something big and exciting enough to be worth 2 million euros, but small enough that you and 3 students/postdocs can do it in five years, since that's what 2 million euros can pay for. But our grant office did make sure that most of the speakers were people who had actually applied for these grants in the last round.

One really key thing for me was to reach out to senior colleagues in the same area (definitely not just at the same university) to get multiple examples of what successful grant applications look like. There's a bunch of text in the grant call, but the panel are going to be senior in-area academics who may know what a good application looks like, but won't necessarily be able to articulate all the rules.

Everyone I know finds writing proposals hard. So much of writing papers is being careful about citing prior work and not over-claiming for your results, but for grant proposals we are basically forced to make big claims about things that we have not done. However, it's worth noting that no one on the panels or at the grant agencies see research proposals as binding promises to do certain work: they mostly all want you to follow the science and change course if you spot something better.

Those two messages are very different from my experience writing (and generally getting) grants in France. Grant reviewers are anonymous, how could you even butter them up ?

Not the reviewers, the agency officers overseeing the reviews and awards.

Not just Theranos, virtually all early stage investment sees warm introductions as more important than a perfectly structured deck

Part of it's perfectly rational: faced with two competently written but not directly comparable proposals, knowing that one of the proposers is less of a mystery than the other is additional information about their ability to deliver, and part of it's humans being humans.

Ironically I'd see much of crypto and other investment bubbles as the opposite cognitive bias (who cares about the gaps in our knowledge, this number is going up faster than that number?)

I think the last part of what you are saying is actually maybe closer to why the comparison seems apt to me. Trivial patents are another comparison. There's an Emperor has no clothes aspect of it to me, a focus on appearance of scientific prestige or rigor over actual substance. Ioannidis' "Why Most Scientific Findings are False" is maybe the best-known reference point, although I think that's maybe slightly different from, if related to, the underlying phenomena I have in mind.

HN maybe isn't the best forum to explain all of this but I'm quite jaded at this point with academics.

I don't want to come across as too negative either, as implying nothing works or there is no progress or something like that, but there's a huge disconnect between what I see personally in scientific circles in actuality vs the nominal narrative in terms of how ideas are taken up, how consensus is reached, what happens to accurate but dissenting perspectives, and how credit accrues. Some of it is common to our time, but some of it I think is amplified or something in the academic community.

I also realize at some level my experiences are not universal, but that there are a lot of people who feel the same (apparently) maybe just underscores my point.

When someone like Peter Higgs is saying he wouldn't be employable today (https://www.theguardian.com/science/2013/dec/06/peter-higgs-...), scientific institutions should be paying closer attention and doing something about it.

I believe that's the problem with most not for profits. Their bottom line is very arbitrary, so objective decisions are difficult.

Compared to, say if I was a paper clip salesman. What makes money and what loses is far more obvious. In addition, if the plan lost money, its immediately obvious and can be stopped

Academia, just like venture capital is made up of people

I used to think that academia was up to merit, during my teaching days

The only and I mean only reason I ever got a grant is because I called the asshole and asked him to craft it

for me

Having been both a grant writer and a reviewer for NIH and NIH-alike grants, there's an element of both.

Established groups definitely get a bit of a pass with "Well, surely Professor So-and-So will sort this out" that an early career person won't. And environment does come into play in terms of whether or not a university has the ability to manage a grant, and yes, large universities basically get an auto-pass there, but it's something people talk about.

But I will say that even for established people at large universities, you do have to stick the content. A ton of recent cycles have been dominated by "Hey, we do COVID research now!" from groups that never did infectious disease research before, and a lot of those got clobbered in review.

But I also do my best to teach my students about the "soft factors" of academia - there are a lot, and they are, by and large, hidden.

Your last point isn't lost either - I occasionally get a little bitter when I translate funding rounds on Hacker News into grant-equivalents.

> You get extra credit points if you're active in visiting other universities and giving talks on the annual job circuit. You tend to do well if you advisors are well-connected. You have to sprinkle a bit of fad-ish topics into your research if even if it's a stretch to do so.

Sounds a bit like any job:

* Cross train in other departments

* Demonstrate your functionality/team’s functionality to other groups through talks, white papers, ‘internal consulting

* You do well when your boss does well, make your boss look good

* Buzzwords and trends make the world go round

Instead of buzzwords and trends, more accurate to say:

"Communicate all your ideas using narrative form. Embed your ideas in motifs, illustrate them with visual icons. Explain their worth by using existing context."

The reason buzzwords and zeitgeist terminology works is because they're a shorthand to understanding a topic, and a linker to existing context.

It's important to consciously understand how that works though.

I've gotten a fair amount of grant money from different Canadian research grants, some retrospective, some prospective. I've also written grants as a side thing. Probably all summed together I've helped pull in around $1m–$2m.

I don't know about the academic side, but for corporate grants this part:

> sprinkle a bit of fad-ish topics into your research if even if it's a stretch to do so.

Does work, but only if the core thing you're tackling isn't as meritorious as what the reviewer is used to seeing. So, for example, one grant I wrote for a hardcore AI startup that had huge data engineering issues on a shoestring budget. Former Stanford, Google, etc folks. I didn't need to pull in anything. I just straight wrote the claim (retrospective grant applications are called claims) and we got a big cheque. Most of the work was meticulously going through their pull requests and estimating hours that were research vs traditional software development.

As an aside, I make around $5k per day with this type of work because I get pulled into the most technical claims and I'm very good at it, but I don't enjoy it as much as software development so I only take stuff that I think needs my help.

Getting your talk accepted to a tech conference is exactly the same way with unwritten criteria. People are crazy animals.

How is this different than any other walk of life?

also noteworthy elements:

university overhead 53%, so that leaves $250K. Over five years that works out to basically $50K/year

That needs to cover 1 month summer salary plus the salary and the tuition for a single student, and the travel to conferences, and the experiment ...

and that is a "prestigious" CAREER award only few people get ... no wonder the author jumped ship

Actually the math doesn't work quite like that, but it's a common misconception. 53% indirect costs (overhead) is multiplied on the direct costs rather than the total amount, so it is actually 500K-500K/(1+0.53) = 173K taken by the university assuming the full amount can be charged indirect. In practice, some charges like tuition and equipment do not have indirects, so the actual amount taken by the university would be about 135K +- 35K out of the 500K. So you get about 365K of direct spendable money from the 500K if the overhead rate is 53% (as it is at the University of Tennessee where the author was).

The indirect cost rate is decided by the federal government based on data by the universities, so it's accounted for in the grant award amounts. So it's not as nefarious as it sounds.

Point is the same

And most of the time (if not always with NIH) when researchers say they got a $x grant, x is the direct cost so what they get directly. then you can often negotiate things back from your university such as time on shared equipments, students...

Also I don't think that is taxable income so basically multiply by 1.67 since most people end up paying roughly 40% of their income in state/federal taxes, social security, medicare, etc.

Where did you get 40%?

The cool trick is that some of the direct costs are for things that the university didn't pay for.

Consider the building. Some rich people donate enough to pay for its construction. The university didn't pay anything. Moreover, that's a tax-deduction for them, so the Feds are effectively paying for part of it.

Now some sponsored research is to take place in that building. You guessed it, the university charges rent to the research grant. Remember that the university didn't pay for the building....

That sort of thing is factored into the programs and policies. Efficiency isn’t really the point.

You don’t pay for the building, but a bunch of Vice Associate Provosts get paid to do something and the union janitors make a living.

Economists can map this crap to direct and indirect labor hours generated and other economic metrics. At some level we do this stuff as a type of work program. Somebody calculates how many hotel nights and cheeseburgers are consumed for each construction dollar spent, etc.

The janitors are paid for by the rent.

The "efficiency" here is that the university gets paid for an asset that it didn't pay for.

Is that not allowed? If someone gives you a house or a car, is it wrong to profit off it by renting it out or offering it on rideshare?

The idea that universities or other public services should be profitable in any way is bewildering to me

They get paid for the whole institution. It’s not a contracting model in the traditional sense, it a model designed to support the educational mission of the university.

> it a model designed to support the educational mission of the university.

Except for the small problem that research contracts with orgs, some for profit, that don't have an educational mission, such as SRI, BBN, etc, follow the same model (absent the donated buildings).

Sure, but what are new grad salaries in industry these days? 100k-200k/yr base?

At my institute, that's not how overhead is calculated. Overhead is a percentage of what the PI is budgeted for.

So for a $500K grant, 53% overhead really means about 34.6% of $500K. The university's cut is $173K, which is 53% of the PI's budget of $327K.

It's quite similar to the Australian system. The most prestigious, hard -to-get award with ~10% success rate, which took me two months to write last time (I was rejected) awards on average AUSD 400k over 3 years. That's a Discovery Project from the Australian Research Council.

A postdoc with overheads (postdocs make decent salary of around 100k + overheads similar to OP) budgets at about 150k per year. So with the most prestigious, biggest award the non-human researchers can get, you can barely fund one guy for 3 years. There is an ongoing exodus because this is insanity.

53% is pretty good! I believe Stanford and MIT have O/H over 100%

MIT is 55% Stanford is 58%

The only place I've ever encountered with an overhead rate above 100% was Scripps, and that's because they have to maintain research vessels.

US National Labs (e.g. LBNL) can easily top 200%, based on what my colleagues told me at the time. I never applied for funding myself while there, so I cannot directly confirm the actual figures.

How national labs are funded is its own completely bonkers system.

> university overhead 53%, so that leaves $250K.

At many universities large capital expenditures over some threshold ($5K for my institution) have no overhead. Also a lot of that money is supposed to go toward supporting students and that money is well spent.

How? From what are your admins paid then?

A couple notes:

- As mentioned by several people, your math is a little off, and indirects are not a flat tax, there's a calculation for what is eligible for being charged indirect costs and at what rate.

- The CAREER award is prestigious because it is a signaling mechanism - it's directly funding a particular PI more than it is a project. It's...sort of like being picked for a YC round in that respect.

- NSF funding is designed, philosophically, around smaller awards by and large, compared to say the NIH. The average "Oh, this'll make sure you get tenure" award for NIH-funded fields is an R01, which has a direct budget of about $1.25 million.

I'd be curious what the impact of fewer but more voluminous grants would be on academia. As other commenters have noted this grant turns into roughly 1 student for 5 years working on the project for a professor. Even for this grant, you would expect that a professor would need 3-5 such ongoing grants to really keep the lab functional.

It took 2 weeks to craft this grant proposal. Assuming the average academic has a 30% hit rate and the average grant provides funding for ~1 year... then this translates to roughly 30 weeks per year. A somewhat extreme example, but I wouldn't be surprised if these numbers get worse as you move to lower tier universities - and better for higher tier universities.

Would it be better to simply grant lab's to researchers with a general research plan on a 3-5 year cadence?

More voluminous grants would be riskier and harder to evaluate, AIUI. As of today, the idea is that a grant functions both as a prize for having a competitive proposal in the first place, and as a wager on behalf of broader society on the actual research impact. So yes, grant proposals take time to craft but that's just part of the game, and arguably of the actual research that you'll be doing.

possible, and an employee who doesn't know if they'll have a job next week might also work harder - at least for a time.

We are already taking the societal risk of employing a professor and giving them grant money - at least we can get the productive side of the money we spend.

There are indeed larger grant programs. Those tend to be collaborative affairs. You have multiple PIs across different institutes going in on the same submission. Each person contributes a piece of a larger (hopefully cohesive) whole. It's generally less work per person compared to a solo submission for a smaller pot, but it's one of those cases where it's crucial to have established working relationships in your research community.

In many fields, students do not need extra grants for funding. Startup funds, teaching assistantship (TA) and other sources fund PhD students. You can advise 3-5 students who TA at the same time, without writing grants. Some fields on the other hand need more external grants to fund students. It really is field-dependent.

This.

A mentor of mine is able to fund most of her students, save for maybe their final year, on TA-shps.

I'm in a purely research department - if I don't have funding for the full time, I don't have students.

Most of the federal grants I am aware of have duration of 3+ year (the typical duration of a PhD after having completed the course work).

That part about how some people have strong opinions about what absolutely needs to be in a proposal, but if you look at a handful of successful ones, there's no clear winning recipe, really resonated, having recently gone through this for a similar ERC competition.

The part about finding >10 typos is pretty much my nightmare, and I appreciate the candor with which you share that and other parts of your experience.

I am not in academia but the 53% overhead to an outsider is absurd! Probably the same reason tuition has gotten so high, someone has to pay for all those deans, vice deans and vice vice deans!

Compared to industry? That's very low! Universities are cheap.

53% overhead works out to be 34% of the total money.

I've written grants with many industrial partners (companies you would recognize). All have had much higher overhead rates. Normally these rates are kept secret, but here's a fun memo from the Federal Transit Administration talking about 127% overhead rates! https://www.transit.dot.gov/funding/procurement/third-party-...

The DoD gives as an example a rate of 125%!

> For example, in recovering the indirect costs associated with particular contracts during the year, each dollar of engineering direct labor worked on a contract is burdened with an engineering overhead of 125.95 percent.

https://www.dhg.com/portals/0/headers/industries/government-...

This puts universities in some context: https://www.healthra.org/wp-content/uploads/2018/08/Nature_K...

You'll see that non-profits and hospitals tend to have far higher rates. This doesn't include industry though. Which would be far higher.

The DoD gives as an example a rate of 125%!

For DOT engineering design firms and environmental firms, a typical overhead rate is 165%-ish, I've seen ranges from 125% to 250%. Over 180% is considered "high".

I've heard Harvard's is/was 96%? Or at least that was for a NIH R01-seeking person there in ~biochemistry in 2015. Where I'm at it's probably around 100% but I really don't want to know. We're all expensive.

It's 69%, so you were...close I guess?

https://osp.finance.harvard.edu/files/univ_july_2021_signed_...

That overhead honestly doesn't seem that high to me. If I think about private companies with low need for physical equipment (e.g. software companies) I suspect the overhead would be comparable if not greater. Usually I hear ~100% as an average figure (keeping in mind that overhead is defined as a percentage of non-overhead costs so 100% means half of your money spent is on overhead).

But I'm not that confident. Would love to hear from business owners.

I’m at a small software company that does lots of smallish (500k-1.5M) contracts like this. The government does audits to determine our overhead rate. Our is 2.8. That is my billing rate to the government is 2.8 times my hourly rate as determined by my salary. That pays for rent, benefits, and for things like the employees who aren’t on direct labor (e.g. accounting, HR, etc)

That means your overhead rate would be around 65%.

No it does mean the overhead is 280% if defined the same way as the university (overhead is calculated as percentage of non-overhead costs, not overall costs).

But I'm not sure whether the company would count just an employee's salary as non-overhead costs. I'm fuzzy on the details of how overhead is usually calculated.

> No it does mean the overhead is 280% if defined the same way as the university (overhead is calculated as percentage of non-overhead costs, not overall costs).

But he said:

> That is my billing rate to the government is 2.8 times my hourly rate as determined by my salary.

Doesn't that mean that the government pays a total of 2.8x his salary, so overhead would be 180% (+100% salary = 280% total)? Or what mistake did I make here?

Yeah, I don't know exactly what is called "overhead" if it's the full 2.8 or 1.8 above my costs. All I know is the accountants told me to write down 2.8 and that the bill that we send the government is 2.8x what I make "hourly" for the hours work on that contract.

Yeah. So for every hour you bill, about 35 cents of every dollar goes to you and 65 goes to the company.

And this is where it's weird, because 65%, 180%, and 280% are all valid, but different, ways of looking at it.

Because you get 35% of the money, the company gets 65%. i.e. 65% of your billable hours are overhead.

Or, the company gets 180% of what you get. Or, you could say your overhead rate is 180%.

Or, you charge the government 280% of what you get paid. Which is the effective end result.

>That is my billing rate to the government is 2.8 times my hourly rate as determined by my salary.

Why does your salary define the overhead rate for the entire company?

Sorry, that was me being lazy because I was typing on my phone. The government tallies up all qualified expenses the company has (salaries, benefits, rent, utilities, etc). And then tallies up all how many hours employees worked on "direct labor" so things directly related to a client (in lawyer speak, think "billable hours"). And then they come up with a ratio which is your overhead rate and that's the rate you're allowed to charge the government.

    all_expenses / total_hours_worked_by_all_employees = average_rate
    all_expenses / direct_hours_worked_by_all_employees = direct_rate
    direct_rate / average_rate = overhead_rate
    my_hourly_rate * overhead_rate = my_billing_rate

There's more nuance on what are qualified expenses and what are direct hours and things like that, but the government has formulae, and they determine your billing rate - or at least the overhead that turns into the billing rate.

Now, of course, my company can't pay me $1000/hour which turns into $2800/hour invoices to the government because then we'd be too expensive and never win the contract. So there is downward pressure on salaries because of that, and it prevents us from hiring superstars because their hourly rate will be too high. So that's one reason the government doesn't really get the A team working for them.

And if you look at the budget of most R1 state universities, you'll see that tuition is really far from covering the cost of education.

Keep in mind that many researchers are given a huge startup package at the beginning of their career to jumpstart a research agenda. They also don't pay for any rent for lab space, electricity, and other campus resources. All of those expenses are basically paid for by the overhead tax. It's the lifeblood of any academic research institution.

While "huge startup package" varies wildly - epidemiologists are cheap, for example, you are by and large correct.

The data center where my nodes for HPC are kept? Indirects.

The salary of the folks who review and submit the proposals, manage the budgets, and make sure we're complying with neat things like human subjects protections? Indirects.

Journal subscriptions? Indirects.

Direct costs cover remarkably few things.

53% is both not particularly that high and not unusual. Keep in mind that in private industry, that stuff is built into the price. I've also seen the indirect rates for several private research firms - they're higher.

I think the telling thing is that, even at that rate, companies often find working with us to be cheaper than working with another industry partner. Our pricing just tends to spook people because it comes in a big chunk just labeled "Indirects" rather than snuck into all the other budget lines.

> My university took a whooping 53% for overhead.

Highway robbery.

I love these kinds of breakdowns. It's very useful to help communicate successful models in people's crafts.

I wonder how the university gets to the 53% overhead figure. Do they have a fully costed internal accounting of things like "rent" for lab space and share of administrators' time/salary?

They have a shockingly detailed model that they negotiate with the federal government ! https://cfo.berkeley.edu/sites/default/files/facilities_and_...

Wow! Shockingly detailed is not an overstatement.

It seems theirs is even higher (57%).

Yes. Cost of space, cost of power, cost of business office staff to process the paychecks, cost of a bunch of CPA and auditors to make sure you're following all the federal regulations about how the money can be spend, cost of IT to provide WiFi and network jacks and bandwidth, cost of water in the water fountain, etc, etc.... At institutions where a large fraction of the work going on is grant funded if you didn't do it this way you can't really make your institutional budget balance. And 53% is actually pretty low compared to many institutions. Source: I've been 100% grant funded the last 20 years of my career and have been involved in writing 100's of federal research grants.

> Cost of space, cost of power, cost of IT to provide WiFi and network jacks and bandwidth, cost of water in the water fountain

Time to move to a WeWork? WeWork All Access - $3600/year. Or at home for "free" if no real lab is needed.

> cost of business office staff to process the paychecks

Time to move to Direct Deposit? For a grand total of $0!

> cost of a bunch of CPA and auditors to make sure you're following all the federal regulations about how the money can be spend

Let them spend it however for free? or hire a CPA for a couple days. No need for $173k worth of CPA time per grant.

> At institutions where a large fraction of the work going on is grant funded if you didn't do it this way you can't really make your institutional budget balance.

Based on the math, they either need to invest into automation or it seems like a dying industry.

I think it’s your expectations that are out of whack. There is sticker shock, but pretty much everywhere else is higher.

Also, do you know what a clusterfuck it is if someone screws up accounting, payroll, or legal? Those systems are Byzantine and you really do need an expert to navigate them legally.

Institutions don’t take the risk that an individual might take.

My institution employs about 7000 people.

I'm interested if there's a single company that employs that many people you can point to that pays $0 to handle payroll.

It's an open secret that the indirect costs are budgeted to make profit for a university. That is, the stated costs exceed actual costs. This has been publicly discussed by former federal division heads in major outlets.

The problem with this is the universities rely on this funding to stay afloat, so there's a tail wagging the dog problem. Basically the evaluation of scientific merit then derives from a decision as to whether or not it can procure money from the federal government.

That in turn is driven by all sorts of factors:

https://journals.plos.org/plosone/article?id=10.1371/journal...

I don't want to argue that grants are bad but I do think the system is broken. Everyone passes the buck to the federal government in this kind of shell game, and then the federal grant system becomes the [near] monopoly over scientific decision making.

The analogy I tend to use in my head is that it's like if your state gave out grants to fund plumbing projects, and people realized that these grants paid more than the actual costs for the project. You have a plumbing project, get the plumber to get a grant for you, and pocket the difference. So people stop hiring plumbers on the basis of the work they do per se, and instead hire them on the basis of their ability to procure state grants. That's kind of what the system works like now.

Do industry grants have larger overhead? It doesn't matter one way or another. That's a different issue. Two wrongs don't make a right either.

I'm very curious if this number is actually correct; in my experience "53%" overhead means the university adds a $0.53 "tax" on each dollar you spend, which works out to the university "taking" ~35% of the total amount. Still high, but not as high as I used to think.

As faculty in the UK on the receiving end of the current pension cuts that little voice telling me to leave for industry while I still can only gets louder when I hear about people jumping ship like the OP...

[Tangent] Educators just aren't as valued as they used to be. This is a significant issue in my family, as I come from a clan full of teachers of many types.

> The CAREER Award is a special 5-year grant that can only be applied for 3 times by an individual researcher in the first 6ish years of their career. It is deemed "prestigious" and at many schools it effectively means you'll get tenure.

> I was offered the award, but I declined it. Then I resigned from my faculty position and joined industry.

OK, we all want to hear more about this right? OP spent a lot of time preparing an application for a very prestigious academic-career-making grant, succeeded in getting it... and then quit!

I might not be very rational :)

Nevertheless! Care to share your story? What turned you away from academia?

I wasn’t happy.

I shared a bit more detail here: https://austinhenley.com/blog/leavingacademia.html

How do you cope with the feeling that you're missing out on future prestige?

Same way that I (who left industry to pursue an academic career because I wasn't happy) cope with the feeling I am missing out on a giant FAANG salary: mostly, I just don't think about it.

Like, I have enough money to afford a house and take my kids on vacation from time to time. So having lots of money is something I idly wonder about sometimes, but no critical need is going unmet by its lack.

Conversely, if you have a prestigious position, most likely everyone around you does too. For example, I'm a professor, and yes, it makes my mom proud. But all of my colleagues and many of my friends are professors, too. So in terms of day-to-day life, it just doesn't matter much.

I think everyone wonders about what might have been. But you can really only make one choice at a time, and so even if you had chosen differently you will still wonder what might have been.

Maybe they just discovered levels.fyi :)

>OP spent a lot of time preparing an application for a very prestigious academic-career-making grant, succeeded in getting it

He looked into the abyss long enough for it to look back.

as someone who couldn't manage getting hired by an R1 university, and also had three NSF CAREER proposals rejected, it's a bit sobering to see someone get it all so easily and then walk away from it.

Not that I necessarily blame him, fairly burnt out on academia myself currently.

Not getting a CAREER award can really set one back, both in time wasted applying but then having to compete with those who have them. I often felt like the NSF asks for so much these days, including excessive (and possibly unrealistic amounts of) K-12 outreach. I am always curious if you go back after 5 years, how many of the winners actually did everything they promised in the proposal.

If there is no measurement, or limited enforcement - then I'd bet it's approximately equal to one or two visits.

Whenever someone tacks on way more requirements than could possibly be met, there is a natural prioritization. If after 5 years you didn't deliver on the core idea - then the NSF probably won't care that you completed all of the k-12 outreach. If you hit the research objectives out of the park, then they are going to demand you max out the k-12 outreach.

I found those NSF outreach requirements very frustrating, because I knew (or believed) that aside from some actually conscientious and dedicated people, most everyone was making up perfunctory or fictitious plans for how they would satisfy this requirement if they got the grant.

And the NSF would not be checking on whether people actually did those things, or take back the money if they weren't fulfilled, so basically (I felt at the time) it was incentivizing people to lie about their intentions. I didn't feel good about making up such stories. So I guess I lost out to those who were willing.

If you're going to put in requirements like this, I think you'd better follow up and make sure people don't take you for fools and say what you want to hear but not do it later.

(edit to add: my experience is based on having been through this 15 years ago, so it's probably to be taken with a grain o salt)

> I knew (or believed) that aside from some actually conscientious and dedicated people, most everyone was making up perfunctory or fictitious plans for how they would satisfy this requirement if they got the grant.

> And the NSF would not be checking on whether people actually did those things, or take back the money if they weren't fulfilled, so basically (I felt at the time) it was incentivizing people to lie about their intentions. I didn't feel good about making up such stories. So I guess I lost out to those who were willing.

On what are you basing these assertions? As part of an NSF funded research project with an outreach component, we definitely had to deliver on it and I had to write about it in our annual reviews. If we had failed to deliver on our promises in a perfunctory of fictitious way, we definitely would have been in jeopardy of having our funding pulled.

The NSF doesn't have to take back money because they don't give you all of the money up front. They can refuse to give you the rest of your funding, and then good luck on your next NSF grant proposal.

I believe you, about how it is now. It didn't seem that way 15 years ago when I was observing it.

Having written NSF grants and other grants - DARPA etc. I have won some and lost many. I compiled a few non obvious takeaways:

- Story telling is everything. It seems this is a huge lesson never taught in grad school.

- Technical details sometimes work against you. The author is absolutely right that getting the general thought process across is crucial.

- Who you get on your review committee tends to significantly skew the outcome. Can make or break your chances.

- I have known a lot of groundbreaking work funded by other money for these exact reasons. Sometimes good science is too far afield for people to understand. An anecdote i like from recent times is how Eric Betzig built the super res microscope. Here's the background.

https://arstechnica.com/science/2015/04/quitting-failures-a-...

TLDR: he won the Nobel...

The sad truth is $500k for a career grant seems like a lot when you're in the university, but when you get out and see where else money is being spent you realize how poor academia really is.

The takeaways are very true. I would go further to say that not only is the story telling and the scientific vision a very important part of the sell, the language and terminology needs to be in sync with the reviewers. Often times researchers write very technical details in the proposal that is hard to parse and backfires.

In non-blind proposals, pre-communication is the key.

> The sad truth is $500k for a career grant seems like a lot when you're in the university, but when you get out and see where else money is being spent you realize how poor academia really is.

It's not just about the $500k, it's about the freedom to do your project your way from start to finish. I can't get $500k from any investors to do my work; their number one question is what's their ROI, which is not even in the universe of my concerns. The last thing I ever want to think about is how to profit from my research.

Another choice is to join a company and then convince that company my idea is worth while. If I get the greenlight to go ahead that's great, but then I'm still always living under the threat of being fired or the project getting canned. Any University is happy to have a CAREER recipient, and your award is portable between institutions.

And sure you'd possibly be able to convince your company to sponsor a project you're interested in leading, but the for something like $500k you'd better have a good business case for it, and I can't imagine them approving something that would last 5 years unless it really aligns with their business goals. In order to get the clout to actually ask for this kind of project you'd also need to be at the company for a while, whereas something like a CAREER award is intended for early-career faculty. Imagine going up to your boss and saying: "You know that job you hired me for? I'd like you to keep paying me my regular salary, but I'd also like $500k to work on my own pet project about 70% of the time from September to May, and 100% over the summer. Oh, and I'd like to do this for the next 5 years."

Respectfully, I disagree. The point you make about financial freedom, imho is an illusion. A few thoughts on your first point.

The NSF Career grant affords you the latitude to pursue your research interests, but those are inevitably aligned with the academic process as you are in pursuit of tenure. So you need to publish and you need to produce publishable work- very unlikely you will choose only high risk moonshot projects as your likelihood of receiving tenure is directly correlated to those publications. Furthermore, your statement that the NSF or the university is not an investor, is also flawed imo. The NSF specifically asks for impact as it ties the money it allots a research to economic impact to the US- as it should since it is tax payer money. I.e your investor is the US. A similar argument can be drawn to the university that provides you startup lab funds- their long term goal is to receive publications/ patents/ prestige they can the monetize against through either selling those ideas/ receiving royalties or donations go their foundations. As you become a more famous researcher you also attract more masters students who pay a good deal of money to go to school there. They are in fact your investor only expecting a different ROI...

As for companies- how much do you think an engineer or ml researcher costs per year to a company especially the caliber in a Phd lab? 500k expense is pretty small. your assumption that a company wont give you 500k to fo work is an illusion- they do it's just not hard cash, they spend on resources that you use. An average PhD level base salary at a Fang - 200+k plus bonus/ equity closing on over 300k?

So the real question is pick your poison.

Good points!

To the first point, starting in 2017 the Career solicitation dropped the tenure track requirement. I’m not tenure track and I have no aspiration of perusing tenure.

Your point is taken about there being “investors” no matter what, but the expectations of those groups couldn’t be any more different. NSF is happy if you spend your money on grad students. The idea of building a research program that is attractive to paying masters students is a far different prospect from building a profit making venture. I’m quite comfortable talking about and demonstrating the broad impacts of my work, but those impacts don’t include making investors a boatload of money in terms of selling a product. That’s just not in the cards. And to put a finer point on it, I’m much more comfortable with my investors being the general American public and my country as opposed to principally rich people.

To your second point, I agree that it costs a lot for a company to hire an ML engineer as an employee. But I would say it is not typical for a a researcher with maybe only a couple years of experience post grad school to be offered $500k to work on any project of their own choosing for 5 years (or more, let’s just be clear that $500k is like the minimum you’re even aloud to request for a career award).

I mean, maybe that is happening, I’ve honestly never heard of that. Does anyone here have that kind of arrangement, or have you heard of anyone who has been able to pull that off? Would be very interested to know how to approach a company to convince them of that. Pay me a salary, give me $500k, let me work on whatever I want for 5 years at my own direction, and that would sound comparable to the kind of freedom you get with a career award.

`The sad truth is $500k for a career grant seems like a lot when you're in the university, but when you get out and see where else money is being spent you realize how poor academia really is.`

This hits me every now and again.

But your first point is huge, and one thing we've been working on with our program's grant writing class. Students phone in their specific aims pages to go write their approach sections, which is what they're working on, and that's a huge mistake.

Browsing the author's site, I notice they have a post about Information Foraging Theory, a framework that has been very helpful for me in my technical writing career! https://austinhenley.com/blog/informationforaging.html

I was involved in writing a grant relating to product development (Singapore has a SG$250k grant for developing product prototypes). Between 5 colleagues, I estimate there was a bare minimum of 300 hours of work... the application itself, multiple rounds of follow ups and justifications, documentation/receipts on how money was spent, etc.

The whole thing has a strong 'here's what a $1,000,000 lottery ticket looks like' vibe to it.

Don't get me wrong, I'm sure it's a great proposal and I am sure the author put in a lot of work and thinking and I am sure he deserved to win, but that's simply not enough nowadays.

I only know the European grant system but I'm still perplexed why grant submissions are not blind. It's pretty obvious that certain universities usually get X grants in each cycle simply by putting their name on a proposal. I'd much prefer if you pitched the idea + costs etc. in a blinded document and have double blind decision making which proposals get accepted.

Sure that system could still be gamed but at least it would be a bit better (imo). The downside is that you cannot have the "what did you do before that merits getting the grant section" but that could come in the second round or something. It can probably be ignored, too because it doesn't matter that much if the research design is solid (imo).

Ya I totally agree. I think the track record should be evaluated separately and independently.

What is missing here so far is maybe what happens on the other side. Some researchers meet for a few days at NSF headquarters to discuss the proposal with some NSF officers. It is called the "panel".

The researchers essentially rank the proposals and decide the outcome:: proposals at the top of the ranking will be funded until money dries out.

Researchers do not know how to evaluate grants the way the NSF wants them to do it. Researchers know how to review papers but not NSF proposals. The broader impacts expectations of the NSF are as obscure to the researchers in the panel as for everyone else. At a panel, after discussing the scientific merits of the proposal, the NSF officers asked the researchers in the panel for broader impacts in the proposal. The first day, the researchers kept pin-pointing (what they thought could be) broader impact items that the NSF officers rejected. As time went by, researchers in the panel started figuring out what kind of broader impacts the NSF officers expected, and could use it to push proposals they liked because of the science. Broader impact sections of the proposals were used to supercharge an already interesting proposal based on the science.

While 15-20 researchers may be in the panel, a few of those will be leading the discussions because of character and interest of looking at these proposals. If these researchers have integrity, the panel will do a decent job. I am sure it is highly dependent on the panel, but the ones I participated in, I thought we did a decent job, maybe with an inclination towards funding young PIs who never got funded before. But networking with NSF officers didn't help any PI; the science dominated the ranking.

Polished broader impact statements appear naturally in tier 1 universities that have many examples and templates of previously funded proposals. Most of it is window dressing; of course there are genuine initiatives but that's not the norm. Asking for diversity statements or whatnot will favor applicants from the alumni and PIs of the Stanfords and Harvards who can network with their peers to gather existing successful statements. After a few generation of sharing these statements, the polishing gets really really good and looks quite serious. It's fairly easy to write broad impact statements with 5 funded examples, much harder when you have none and nobody to talk to (does requiring diversity statements in applications actually hurt diversity candidates? Who knows, studying that question would likely not get funding).

Successful proposals read really well, so well that you wonder how much of the proposed research is already solved. That is a strategy used to write some proposal: when the NSF deadline comes, pick your 1 or 2 most significant unpublished working drafts and write a proposal on the results. Some results you know you already have, others could be interesting generalizations that are within reach, and add to the mix maybe one sexy moonshot. Because you had the paper underway, the story and notation are things you already thought about for a while, and the resulting proposal looks pretty good. If the proposal gets funded, you already have some results to show; good job.

Source: academic who also got CAREER the first time, somewhat burnt out by all these games. Don't get me started on academic awards (the ones with a famous name attached to it)--but let me just say that staying away from those will delay the burnout, a bit; those games are much more rigged than NSF panels.

Coming from NIH-land and occasionally dabbling in NSF-funding via EEID, RAPID and some collaborations with CS folks, the broader impacts section remains the thing that's most opaque to me.

I've written a couple of grant proposals over the last 2-3 years (though not in academics) and it's been a mysterious black box where I had no idea where to even start, so thank you a lot for sharing this!

Thanks for this, this is interesting.

I agree, when doing something like this it's hard to know what it should look like in order to be recognised professionally and having a template or example is a huge huge help.

If you read the proposal, the 8th section is the forced diversity statement and goal. I was expecting it to be a little bit more obvious.

The 'Broader Impacts' section the proposal is a required element for all National Science Foundation proposals (like this one). It need not be about diversity but you need to articulate how your proposed work has "potential to benefit society and contribute to the achievement of specific, desired societal outcomes." So certainly it's forced. In the same way demonstrating the Intellectual Merit of your work is also forced (that being the other criteria against which the proposal is going to be judged).

Researchers should absolutely be required to ruminate on the broader public impacts of their research when requesting vast sums of money from the public.

I once heard a researcher on NPR being interviewed about the ethical considerations of Deep Fake research and their answer was (parphrased) "I don't think about these things, I just do the science. The ethical impacts are for others to think about" and I jaw hit the floor.

Today we see the unintended consequences of various technologies finding confluence in the public sphere, and the impacts are reverberating across society. Now more than ever researchers have to be cognizant of the broader societal impacts of their work.

Is there any similar advice for getting NSF or other agency SBIR grants that provide for a greater chance of success?

I wonder what conditions need to be improved in order to make faculty positions more attractive?

0% or negative overhead?

No teaching?

Longer Sabbaticals?

Fewer admin duties?

Be left alone to do research?

Faculty positions are already some of the most coveted and competitive jobs on the planet

Universities still fight for top talent. So there is still incentive to improve conditions in order to attract and retain talent. Eg how to lure a candidate with multiple offers in hand.

It's not clear that "more attractive" is actually necessary - as mentioned, they're already hugely competed for. But to address your comments:

0% or negative overhead? - While indirect cost return, which is where I get some of my overhead back, is awesome when you can negotiate it, indirects are not the boogeyman HN likes to make them out to be. For government grants, these are known factors. For private industry, lower overhead rates can often be negotiated. Same for foundations, etc.

No teaching? - This depends entirely on the person, but generally, reduced teaching obligations help if you're interested in research. There are people who are passionate about teaching and don't want to give that up. There are people who are in purely research positions (like me) and maybe voluntarily teach a single graduate seminar once every other year. Many of both groups love their jobs.

Longer Sabbaticals? - There is some research showing that faculty who don't take sabbaticals are more likely to leave. I'd suggest instead of "longer" more frequent. An opportunity to "clear the deck" so to speak.

Fewer admin duties? - wistful, longing expression

Be left alone to do research? - This is probably our #1 request, but how to actually make that happen is something of an open problem.

What is an "education plan" in the context of a grant proposal?

I thought it would mostly be training grad students, but they did a great job weaving the program into their undergrad courses and like professional outreach. That section of the proposal is super skimmable, each paragraph has a pithy bold summary header!

I'm guessing the idea is all grant research should serve some educational purpose for the grad students- is that the idea ?

The NSF heavily pushes the idea that there should be a mentorship and teaching component to most/all of their funded proposals.

Nothing in this proposal should take 5 years. For $250k you can easily do this research in a single focused year probably.

I imagine researchers must actually be doing multiple grants in parallel or have other main jobs to earn decent money because if they only do this one thing at a time I have no idea why someone would choose this life of poverty when they could be making solid six figures building practical things.

A lot of the work involves training PhD students to do the research, it’s a big component particularly for NSF funding. So it’s not just the raw research output

And yeah, most professors try to have multiple programs going. This one would have funded one student for a few years

Folks submitting Career grants already have nominally full-time time jobs as faculty with corresponding full-time salaries. The jobs are usually termed '9-month' appointments to allow them time in the summer to get paid for research. So the 5 year timeline reflects that this is not most of what they are doing. Certainly if you have CS background and you're mostly interested in making money this is not the career path for you.

Yeah, but that's the way these things go. The people working on this project are working on 5 other ones as well. It would be nice to work on fewer research projects at once but the way these grants work (check out the low total amount), its generally unfeasible. Its sort of laughable extremely hard to get grants from the NSF, DoD etc. wouldn't even pay 2 typical FAANG SE 1 year FTE.

I do wonder why academics don't open up research collabs with industrial engineers more frequently (without the corporation involvement). I can think of a half dozen academic research problems which are bottle necked on coding.

That grant proposal should answer your question - what's going to pay for that industrial engineer in that collab? You get $50k/year for 5 years and you can't even pay that to an outside engineer since then your proposal wouldn't fulfill the education goals by having a grad student 'trained' through doing that thing. (quoting OP of what the annual budget covers - "1 month of summer salary for myself, 1 grad student at an above-average salary plus tuition, no equipment, alternating international and domestic conference trips for both myself and the grad student, and funds to pay participants in user studies".

Hire would certainly be a no go, but partnership/volunteer work could be viable. However that also would not meet the criteria of the grant.

As academic, coding is my favorite part. I would not give that up

Especially once we're talking about staff.

Keeping a $100,000 scientific software developer on staff, at 10% hit rates (not out of the ordinary for the NIH), means writing a million dollars of grants per year.

And if one of them is successful, that salary is 40% of the total budget for a non-modular R01.

NSF proposals are expressly not designed around funding 100% FTE research positions. "A single focused year" isn't what they're funding.

Reply on news.ycombinator.com