Funding Models and Progress
rootsofprogress.org
rootsofprogress.org
Speaking from the standpoint of computer science, I believe the golden age of scientific funding was between WWII through the early 1990s. Bell Labs and Xerox PARC invented a lot of the technologies that we rely on today, and there is still an untapped reservoir of research that has yet to be fully applied (I still believe that Smalltalk80 is quite futuristic in many ways despite its being 40 years old). DARPA and its predecessor ARPA was key for funding a lot of fundamental computer science work. Back in the 1960s ARPA had a funding model that emphasized "funding people, not projects." Alan Kay discusses this more at http://worrydream.com/2017-12-30-alan/.
Unfortunately, since the 1990s (and possibly before then), both industry and government have shifted to the model of "funding projects, not people," and there's a greater emphasis on delivering short-term results instead of funding medium- and long-term research efforts. There is also the PhD production problem, where more PhDs are being produced than there are research jobs available for them, whether they are in academia, government, or industry. This results in plenty of computer science PhDs working as software engineers in non-research environments. For a few years I've been considering becoming a professor at a teaching-oriented university; computer science professors have been in high demand at teaching universities for the past few years, and there are fewer "publish-or-perish" and funding pressures for teaching-oriented professors than for professors at research universities. But now with the COVID-19 pandemic and the sudden shift to online education, I am concerned that this shift to online education, as well as the economic fallout of COVID-19, may dry up opportunities at teaching universities.
Because I still want to pursue a career in research even if traditional opportunities dry up, I am interested in alternative funding models that will ensure that researchers could still make a living in a world that demands instant results.
All these while, investment in science and research have had returns that blows up any other investments you can possibly imagine. So why are companies so relunctant at even putting in 1% of their revenues in research? Why we as nations put less than 1% of GDP in research? Why true exploratory research is only expected to happen in academia?
These are obvious and massive gaps in our economic story and progress. But anytime you have gaps, you have opportunity! One of my favorite funding models to play around is a slight variation of Y-Combinator model. Instead of giving $120K to teens who can whip out websites that you can later offload to trillion dollar stupid gorillas in 5 years, imagine you do this: You hire researchers who are passionate and give them $0.5M grant to have 3 year runway. They need not build companies, do pitches or generate revenues. They just do their own curiocity driven exploration, they set their own agenda. Bunch of these folks even perhaps work in same buildings so they can collaborate and may be spend some money to hire students/post-docs/engineers. After 3 years, VC looks at their ideas, prototypes, demos and funds a teen with massive enterprenuerial spirit to build products off of those ideas. In other words, we create pool of innovations and pool of people who can build companies. We decouple short term determinisitc execution from long term randomized innovations. It's like having brain with parts that think fast and slow.
How will this work out? I would expect 99% of funded researchers producing nothing useful in terms of business. From the rest 1%, possibly 9 out of 10 are minor things and may be 1 is significant breakthrough leading to $10B company. But more importantly, there is 1 in 10,000 chance that breakthrough leads to something of the order of $100B. This would result in expected return of 20X. More importantly, these companies will be cumulative, so your expected return grows over time with same continued investment! If done right, I think this version of YC could become first $10 trillion company because it will be sort of a meta-company whoes product is long lasting companies built on big breakthroughts.
There are obviously several issues in this idea that I haven't worked out yet: (1) why VC should fund academia like labs when government is already doing it? (2) researchers want to open up their ideas so what moat do you really have? (3) how does collaborations happen? (4) why any enterprenuer wants to execute on someone else's ideas? etc.
PS: Loonshots is a must read anyone interested in this area.
"[T]his syndrome [that the customer is always right] has been killing NSF funding in computing for many decades. They require in their proposals an explanation of how one is going to 'solve the problem', and this has effectively filtered out all but engineering type proposals (some of them have been good -- but most of the stuff that needs to be done requires funding for 'problem finding' and NSF -- for fear of Congress -- has been unwilling to do this)."
I suppose a full professor at a research university who already has tenure can perform non-funded research, perhaps advising grad students on teaching assistantships or external fellowships. But this would be a dangerous career move for an assistant professor, or even for an associate professor who already has tenure but who wants to be promoted to full professor. No funding means no grad students and postdocs to help carry out a professor's research agenda unless those students and postdocs are externally funded. No funding means less overhead money going to the university. No funding often leads to no tenure. Getting funding means "getting with the program" and doing the type of work that funders want in the way that they want it.
This, combined with my love for teaching, is why I want to be at a teaching university where there are fewer pressures regarding fundraising and publication (those pressures don't go away, but they are not as big as at research universities). I can control the quality of my teaching. I don't have full control over whether my grant applications and papers get accepted or rejected.
We need to convince those funding science, whether they are in industry or government, that science doesn't work on a schedule. Funders can't just decree, "I want a cure for cancer in 18 months." Science is inherently risky; experiments fail, hypotheses are wrong, setbacks happen, and sometimes knowledge accumulation is slow. Funders who don't allow for failure are stifling science by pushing scientists into incremental, evolutionary projects that are "safe" but don't necessarily lead to breakthroughs.
But I believe the reason why such short-term thinking is encouraged by funders of science is because of a desire for accountability. Taxpayers funding academic research would be more likely to vote for politicians who encourage spending on science if public funds went to research that many taxpayers would deem "useful." Companies funding industrial research want to fund work that is relevant to the company. A lot of funders have this unfounded fear that scientists will just take the money and blow it on "thinking" at beachside resorts.
There needs to be accountability to make sure that the money doesn't get wasted on non-science activities, but at the same time funding should not stifle scientists' ability to do science.
Regarding your idea about a new type of Y-Combinator-style model for funding science, I like the idea. In a world without places like the 1970s Bell Labs or Xerox PARC, this would be a great incubator for the next generation of these types of products. We got people like Alan Kay and Dan Ingalls who led the scientific efforts, but we also have people like Steve Jobs and Bill Gates who were able to recognize the business value of this work, and we also have people like Charles Simonyi who successfully crossed over from research to entrepreneurship; his work on Xerox PARC's Bravo word processor led to Microsoft Word, which he made a fortune on.
One big challenge would be the accountability challenge from the funders: how will the funders hold the scientists accountable without resorting to micromanagement? How will the scientists retain their independence and not be pressured to pursue "sure bets" that are almost guaranteed to be productized rather than following their own interests, even if they run the risk of not leading to products that could be picked up by an entrepreneur?
It seems to me that incubators like Y-Combinator are more enmeshed in / perpetuate the short-termism that has so drastically affected basic research. I mean, from what I understand the funders for HARC pulled out only a little after a year.
> academia isn't immune from these short-term pressures.
This is 100% true. It's true because the "short-term pressures" are wider changes to economic culture that have emerged since the late 70s. Some fairly call this neoliberalism, but it's plain enough to just link it to the maxim of shareholder primacy that really took flight from that period. The expectation that markets are so fundamental that typically non-economic activity could even be governed by market "laws" became a kind of wisdom incubated in the high prestige MBA programs and leaked out as "management theory" about how organizations should be run. In short, Universities are now run more like businesses than they used to be (a glut of extra "managers" everywhere in the hierarchy, a "customer first" approach). That pressure extends down to the actual academics doing the work. They have less support, less funding, and must seek grants from increasingly short-term thinking institutions.
This is a far cry from the ARPA network of university researchers that gave us the largest qualitative leaps in computing. Universities and even the conditions of the grant funding were much more open and flexible back then. It should not be surprising that we are still largely using systems and methods designed in the 60s/70s (looking at you, Unix) because that's the end of this important period.
I too am interested in alternative funding models for research. I've looked at the teaching faculty jobs you mentioned, but there aren't too many of them in my field as-is, so I don't consider those to be a good option for most.
The model I'm currently looking at uses ideas from the financial independence/early retirement community. Basically, I'll do research independently and live as cheaply as possible so that what money I do get goes farther. (It helps that grad school has made me live very cheaply already...) I don't require expensive equipment to do my research, so I think this is doable. As for how to make money, I'm looking at multiple options including working a full time job for a while to build capital, starting a small business, grants, getting money from patrons, etc. While adjunct pay isn't good, working as an adjunct every once in a while might not be a bad idea to find new collaborators and get library access. I'm also told that the UT Austin engineering schools pays about $10K per class for an adjunct, which again, isn't fantastic, but would actually cover about half of my estimated yearly costs.
Ultimately I think I can spend much less time raising funds than tenured professors do, which I've seen estimated at around 40% of their time!
https://www.johndcook.com/blog/2011/04/25/chasing-grants/
For starting a small business, I'd expect many to require a lot of time up front, but I'm willing to do that if it means that longer-term I can make reliable income with minimal investment on my part. I'm starting with consulting, which I don't expect will be a huge source of income for me, but I'll take what I can get. Ultimately I'll probably have to have multiple income sources, which isn't entirely bad. If any one source dries up then I shouldn't be much worse off.
This seems to be the largest online meeting place for independent researchers: https://forum.igdore.org/
While there is no guaranteed way of choosing the right research to fund, an allocation of those funds among a set of common projects with similarities and dissimilarities would yield the most progress -- or at least, the highest probability of achieving progress. But how do you go about calculating the allocation?
There is a good set of papers that describes a framework for performing such an analysis.
M. Weitzman et al., 1992 M. Weitzman et al., 1998 H. Simianer et al., 2003
We built an implementation of the framework described in those papers, and though we are targeting the agricultural sector, we'd be curious to see it applied for other purposes such as measuring research progress, and allocating funds for it.
You can check it out here: https://mostdiverse.com
Reach out if you want to run some of your own data, we are happy to see this applied to areas outside our expertise.
P.S. We are trying to get funding too! Ha!
In the decision analysis class I took 3 years ago they proposed the framework in the following paper:
Matheson, James E., Michael M. Menke, and Stephen L. Derby. "Managing R&D portfolios for improved profitability and productivity." The Journal of Science Policy and Research Management 4, no. 4 (1989): 400-412.
https://scholar.google.com/scholar?cluster=62055342011699436...
The papers are:
1. On Diversity (1992) - M. Weitzman 2. The Noah's Ark Problem (1998) - M. Weitzman 3. An approach to the optimal allocation of conservation funds tominimize loss of genetic diversity between livestock breeds (2003) - H. Simianer
https://academic.oup.com/qje/article-abstract/107/2/363/1838...
https://www.jstor.org/stable/2999617
https://www.sciencedirect.com/science/article/abs/pii/S09218...
Most logical funding available for my case was denied by the authority "because that would have set a precedend... Look your idea is nice, useful and all, but we never actually activated this (fully defined by law) funding and we fear other might come and ask for it if we do it for you"...
Project is still in a corner of my mind but in standby (also for personals reasons beyond funding).
So yeah I can pretty much relate!
So, I would like to know: who funded Alexander Fleming work? who funded the Oxford lab that followed the work? Why they were underfunded?
This is among the reasons why, for all I love free software, it tends not to be a driver of innovation.