There's no darwinian / evolutionary process at play amongst scientific funding because we haven't nailed down a way to measure outputs we care about. Federal grants are collectively incentivized to avoid "risk" and private capital is incentivized to create near term (2-8yr) commercial value, prob why all bio investments ultimately become tx, diagnostics, or die
If we could measure outputs of scientific funding we care about (semantic diffusion throughout literature, commercial investment in identified IP, derivative job creation, data reusage, reproducibility, or sth) you could create darwinian competition _between funding models_ and figure out how we can produce more useful knowledge with the same public cash
The problem is that now, basically all federal funding perhaps with the exception of DARPA is deployed under the same incentive structure, so it's not at all surprising to me that they produce similar results.
Bottom line, we should measure scientific outputs we care about and then test many, many more funding models against each other instead of staying historically constrained / ossified. You probably wouldn't even have to increase science funding to do it, just carve out 20% of total funding for dramatic new models.
I wonder if canalisation isn’t just a failure of imagination, but a rational response to misaligned downside risk. When the penalty for deviation is career death, convergence becomes the smart move. Institutional sameness might look like conservatism, but it’s often self-preservation dressed up as structure.
Are there good resources to identify how these new research organizations support themselves, sources of funding, and different goals / incentives? Trying to get a sense of how all of these map together.
What would a tax code that didn’t drive canalization and discourage new institutional forms look like? Is there a way to get there by “pushing the rope sideways”, making the effect of the shift revenue neutral and not partisan coded so that it could be done without minimal conflict?
One force pushing this convergence may sit upstream of incentives: legibility.
Even when funding models, missions, or hiring differ, institutions are judged through the same external lenses, publication norms, career portability, funder reporting, reputational signals. Under those conditions, variation collapses. Anything that can’t be made legible to those evaluators becomes brittle.
What converges isn’t just form, but behavior, regardless of intent.
I used to be the director of an interdisciplinary water institute. The canalization effect is real! Of the suggestions, I actually am most intrigued by the concept of 'hiring more weirdos'! Interesting piece.
There's no darwinian / evolutionary process at play amongst scientific funding because we haven't nailed down a way to measure outputs we care about. Federal grants are collectively incentivized to avoid "risk" and private capital is incentivized to create near term (2-8yr) commercial value, prob why all bio investments ultimately become tx, diagnostics, or die
If we could measure outputs of scientific funding we care about (semantic diffusion throughout literature, commercial investment in identified IP, derivative job creation, data reusage, reproducibility, or sth) you could create darwinian competition _between funding models_ and figure out how we can produce more useful knowledge with the same public cash
The problem is that now, basically all federal funding perhaps with the exception of DARPA is deployed under the same incentive structure, so it's not at all surprising to me that they produce similar results.
Bottom line, we should measure scientific outputs we care about and then test many, many more funding models against each other instead of staying historically constrained / ossified. You probably wouldn't even have to increase science funding to do it, just carve out 20% of total funding for dramatic new models.
Great piece - the regression to the mean is real everywhere.
I wonder if canalisation isn’t just a failure of imagination, but a rational response to misaligned downside risk. When the penalty for deviation is career death, convergence becomes the smart move. Institutional sameness might look like conservatism, but it’s often self-preservation dressed up as structure.
Are there good resources to identify how these new research organizations support themselves, sources of funding, and different goals / incentives? Trying to get a sense of how all of these map together.
What would a tax code that didn’t drive canalization and discourage new institutional forms look like? Is there a way to get there by “pushing the rope sideways”, making the effect of the shift revenue neutral and not partisan coded so that it could be done without minimal conflict?
One force pushing this convergence may sit upstream of incentives: legibility.
Even when funding models, missions, or hiring differ, institutions are judged through the same external lenses, publication norms, career portability, funder reporting, reputational signals. Under those conditions, variation collapses. Anything that can’t be made legible to those evaluators becomes brittle.
What converges isn’t just form, but behavior, regardless of intent.
Nice piece. Sam
I used to be the director of an interdisciplinary water institute. The canalization effect is real! Of the suggestions, I actually am most intrigued by the concept of 'hiring more weirdos'! Interesting piece.