We were struck recently by a paper in Scientometrics that proposed a unique way to fund scientists: Distribute money equally, but require that each scientist donate a portion to others – turning the federal funding system into a crowd-sourcing venture that funds people instead of projects. The proposal could save the inordinate amount of time scientists currently spend writing (and re-writing) grants, but would it actually work? First author Johan Bollen, of Indiana University, explains.
Retraction Watch: You propose something quite unique: Fund everyone equally, but ask them to give a fraction of their funding to someone else. Is the idea that scientists most respected by their peers will “earn” a higher percentage of funding, and everyone is just acting as reviewers?
Johan Bollen: To some degree, yes. Every scientist in this system becomes a small funding agency themselves. The size of the fund at your disposition, however, depends on how much you receive from others. Everybody receives the same base amount, but also funding from other scientists. Since you distribute a percentage of everything that you receive, some scientists will have more funding to distribute than others. So if you are a well-respected, trusted scientist, people will be more likely to give you funding, and you will become a more important “funder” in the system.
This makes sense because scientists are supposed to know who is doing the most interesting or promising work. The system aggregates this information by directing the flow of funding directly and indirectly towards those scientists who then become more important funders in the system.
Retraction Watch: You actually simulated how this might work. Can you explain your procedure, and why it serves as a proxy for this funding system? What are its limitations?
JB: We simulated the system to see whether it produces sensible results and under which conditions. Since we don’t know beforehand who a particular scientist would direct funding to, we relied on citation data from the Web of Science. Our simple assumption was that a scientist would be more likely to direct funding at someone whose work they had cited frequently, because a citation indicates that you found someone’s work influential. We counted the number of citations that pointed from a given scientist’s publications to those of others. Using that data we tried to estimate how much funding they would direct to other scientists. For example, let’s say in 2004 I cited two people. One person I cited 5 times, the other 15 times. The assumption would then be that I would distribute 5/20 of my funding to the first, and 15/20 to the second. We ran our funding system in this simulation for 20 years. The result indicated that under certain reasonable conditions (donation fraction of 50%) the distribution of funding would not be very dissimilar to that of the present system. We know this because we also had funding data for the U.S. National Science Foundation and National Institutes of Health for the same period that we ran the simulation for.
The limitations of this simulation are, of course, that we don’t know whether citation data is a good proxy for funding preferences. I might have cited scientist A frequently because they happen to have done a lot of research in my area, but I might not think it is very interesting. I could instead direct my funding to scientist B whose work is very interesting but only just getting started so there is little to cite. That’s the beauty of this system. It’s easy to reward innovation.
We therefore caution in the paper that our use of citation data is not a recommendation; we do not propose funding be distributed on the basis of citation data. We are simply using it to estimate in a simulation that which is essentially the discretion of each individual scientist.
RW: Since this is a fairly radical change from the current system, I think readers would raise many questions about its practicality. For instance, how could we ensure that early career researchers, who aren’t well known by many outside researchers, will get enough money to get a robust research program going?
JB: The present system certainly also presents challenges to early career researchers. When you are just starting out it is difficult to compete for funding. The writing and submitting of proposals takes a lot of effort away from building a robust research program, and success rates are quite low in certain programs.
In our system everybody receives the same base amount. This assures every beginning researcher a yearly unconditional source of funding to work with. They don’t have to compete for this funding, they don’t have to write and submit proposals, and their work does not have to meet with the approval of anybody to receive this base amount. That’s a significant advantage for an early career scientist.
It is however true that this base amount might not be sufficient for everyone. So early career scientists should be encouraged to get out and make their community aware of their work, vision and plans, so that they can receive additional funding. Special programs might still be needed, but one of the advantages of our system is that biases against early career scientists and underrepresented groups can be detected and mitigated, e.g. by changing donation percentages, base amounts, and encouraging the community to give more to those groups — e.g., there can be a specific donation option for “women in science” or “early career scientists”, almost like straight-ticket voting.
RW: How can we ensure that scientists don’t just fund their colleagues and friends, and distribute the funds based on merit alone?
JB: First, donations should be completely anonymous to prevent people from colluding. Second, the system would also need strict Conflict of Interest rules and restrictions such as not being allowed to donate to colleagues in the same group or institution, etc. These rules can be automatically enforced. For example, when you log in to the donation system and start entering names, the system can automatically detect that you are giving to a colleague, someone that you have co-authored a paper with, donated a lot to before, etc. and “gray out” those names. In addition, the system generates data on who donates to whom allowing the management of the system to detect collusion and infractions. We can’t completely rule out any problems, but I think they are certainly manageable, and perhaps to a greater degree than the present system where small review boards make sometimes inscrutable decisions on who either wins or loses in a given competition.
RW: You note that scientists would have to somewhat “pitch” their research to other scientists, through communications and conferences, etc. Are you at all concerned that the time and effort spend campaigning for peer funding could start to approach what scientists currently spend in the current system?
JB: The numbers seem to suggest otherwise. We cite an analysis in our paper showing that Australian researchers spend five (!) centuries worth of research time on proposal preparation. Assuming the U.S. population is 14x larger, you can imagine the enormous costs of the present system. Imagine what could be done with that amount of research time and effort instead.
Since scientists already attend conferences, participate in workshops, visit with colleagues, post to blogs, etc. I don’t think they need to do any extra work to make their peers fully aware of their vision, plans, and present work. What might change however is how scientists approach their participation in conferences and workshops, and how the role of publications might change. That will be difficult to anticipate, but it is very unlikely all of this will require the amount of time and effort presently going into proposal writing, submission, reviewing, project admin, reporting, etc.
RW: In lieu of overhauling the entire federal funding system, is there a more practical way to put this proposal into practice, perhaps on a smaller scale?
JB: We’re certainly not proposing to overhaul the entire federal funding system. A lot of really good work is being done by our funding agencies, and we should be grateful for the hard work that their many scientists and administrators are putting into supporting excellent science.
The system can very easily be implemented on a small scale, either for a specific population or group of scientists, or with a smaller overall budget to distribute. If results are good and we learn more about making these systems better, we can gradually extend them to greater populations and/or increase the overall budget. It’s actually very flexible that way. One part of the funding system could remain competition, project proposal-based, and the other could function according to our proposal.
I should mention that the Dutch parliament recently approved a motion directing the Dutch science foundation to start experiments with this system. The idea is that you can gradually scale things up over time as you gain experience with what works and what doesn’t.
Like Retraction Watch? Consider making a tax-deductible contribution to support our growth. You can also follow us on Twitter, like us on Facebook, add us to your RSS reader, sign up on our homepage for an email every time there’s a new post, or subscribe to our daily digest. Click here to review our Comments Policy. For a sneak peek at what we’re working on, click here.