Thanks for the thought-provoking piece, Ben. I'm familiar with many of these issues.
A few suggestions:
1. If your thesis is right, I'd expect soft-money research professors to out-perform standard professors who get much of their salary paid by teaching and service. Do we see that?
2. TTO's are generally dysfunctional. Is there a "standard deal" that could be put in place for university research out-licensing?
Future Feed in Australia is a rather functional example. They hold the patent for anti-methane red seaweed and their open licensing of it has spawned about a dozen companies.
Or could we imagine a company that controls all IP generated from a University and has the mission of licensing it out efficiently and for financial benefit of its stakeholders? You could think of this as a JV between the university and private capital. It'd be easy to imagine it building an incubator as well.
The Hardcore Institute of Technology proposal does not make total sense to me. It just sounds like an R1 without the degrees.
At Caltech where I got my Ph.D., nobody cared about my grades there. They care about my production of science and engineering, i.e. my portfolio. For engineering disciplines, the focus on high-impact papers is muuuuch lower compared to practical invention. Yes, I got a degree, but basically as an official stamp for completing my apprentice project.
Even at MIT, my undergrad, my professors told me "get good grades or grad schools will think you're dumb, but that won't get you into a Ph.D. program. What will get you into a PhD program is impressive work as an undergrad at a research lab." And this extended to admissions - I got into MIT because of being an Intel Semifinalist. They'd rejected my early admission and deferred me to main admission. So I sent them a packet with my Intel Semifinalist win and tehy sent me an admission letter a week later.
1. The experiment unfortunately has too many confounders: soft-money professorships are lower status so there's a selection effect and also (I think) they get far less startup funds, among other things.
2. People have tried the standard deal and failed. Before Bayh-Dole there was a company that did all the licensing by default.
Caltech is basically the closest thing to the Hardcore Institute of Technology there is. I assume still that professors need their grad students to publish to get tenure though.
Thanks Ben. Caltech is really a hardcore science school - literally half of the school is the chemistry department. So it’s very much a “get nature/science publication or bust” place rather than a tech transfer & startup place. Engineering is secondary to science, and much of the engineering connects to JPL - which Caltech operates. The Caltech TTO is interesting because of that, because it’s mostly handling IP and technology from JPL.
I do think Caltech is a bit different than how I imagine your model. Students and postdocs are siloed into Professors' labs per the typical academia model. When they tried to do something less siloed, the Joint Center for Artificial Photosynthesis, this structure was a serious burden to efficacy.
Thank you for sharing part of Ben's essay. I look forward to the next installments when they are posted here. I agree that universities have shifted from having specialized, targeted roles in the development of students and science to becoming more of a "jack of all trades," especially as great industrial research monoliths like Bell Labs, as mentioned in the article, have shrunk or disappeared. Now, teaching, fundamental research, applied research, patenting, and commercialization of technology are all housed within the academic (non-)industrial complex.
There have been some government efforts to patch the inefficiencies of university technology transfer, such as the SBIR and STTR programs (https://www.sbir.gov/). However, these programs do not fully address the underlying structural issues. They often operate as add-ons rather than true alternatives, and the incentives and cultures within universities are misaligned with the needs of effective technology transfer and commercialization. I’m interested to see potential solutions or new institutional models in the upcoming parts of this series. Thanks again for sparking such an important conversation!
There are lots of problems at Universities (having been a professor at an R1 for more than 20 years, and about to start my 10th year directing a University research center). As a result, I'm always interested in takes on exactly what they are and what can be done to address them.
A few questions to consider after my first read through:
1. Even the Universities with the highest overhead rates have much lower overhead rates than "unbundled" private, government, and industry research labs. Does the author believe that this is *not* this *because* of the economies of scale that come with bundling?
2. Does the author have any plans for reducing research overheads in his "unbundled" alternative, or does he believe that the increased tech transfer will justify the increased overhead rates?
3. The author appears to have a number of concerns with Universities extracting profit from their innovations and using that to subsifize future innovation (e.g. University tech transfer arms). I have lots of concerns about tech transfer arms of Universities, too. However, how does the author expect "unbundled" fundamental research institutions from going the way of Bell Labs, whose inablility to profit from its innovations was part of the reason for its demise?
4. How does the author expect people to be trained in to performing fundamental research if he unbundles fundamental research from the mentorship and experiences that come with that
I'm interested to see where this series goes, but I'm concerned that it may be taking a "why we should break up the economies of scale that have out-competed my preferred ideal" direction that would be unfortunate.
1. There's certainly cross-subsidization that goes on in universities. The question is whether the hits to research effectiveness is worth it.
2. Insofar as new institutions should have business models that don't just rely on grants, sure. But there's a longer discussion about why the idea of overhead is silly to be had. The rest of the world just calls it "margin"
3. I have no problem at all with organizations extracting profit from their innovations. I thing tech transfer as it exists is terrible, and have weird feelings about privatizing innovations funded by public money.
4. See "Hardcore Institute of Technology" also I am not advocating for getting rid of universities, I think training is one of the things they do very well.
Cross-subsidization and economies of scale aren't always the same thing, though. To pick an example, I direct the HPC center at my University, and am occasionally asked "why shouldn't our University researchers just use [pick your favorite cloud service] instead?"
It's a reasonable question, and a big part of the answer is simply economies of scale. Larger institutions have enough aggregate demand that it's cheaper (in many cases by a factor of 2 or more) to host the resources locally, while small institutions whose demand is bursty pay more per compute unit because they don't have enough demand to amortize the fixed costs of running on-premise large-scale computing resources.
Your answers get to two of the larger questions I have:
1. What research eco-system supports these unbundled research institutions? The reality is that modern research is infrastructure-expensive (computing, instrumentation, laboratory, etc.) and generally not something that can be efficiently done in a 1800s-era basement. Without economies of scale backing them these unbubdled institutions are going to be less efficient than larger organizations. I'm guessing the vision that these institutions will make up for that inefficiency with higher-quality results than come from current Universities and lower costs (and risk?) than come from industry labs or startups?
2. What is the financial model for these institutions? Are they public (like the FRO proposal) or private (like Bell Labs/BBN)? For profit or non-profit? Who gives them funds for what reason and on what timeline and term?
This vision of unbundled research institutions seems like it's trying to the a Goldilocks solution between large universities and small startups. Goldilocks solutions are attractive when they get the best of both worlds, but can also fail spectacularly when they end up with the worst of both worlds.
Why exactly are the unbundled research institutions being propose going to be the former and thrive and not the later and suffer the fate of the Bell Labs and Xerox PARC that inspired them?
Thanks for the thought-provoking piece, Ben. I'm familiar with many of these issues.
A few suggestions:
1. If your thesis is right, I'd expect soft-money research professors to out-perform standard professors who get much of their salary paid by teaching and service. Do we see that?
2. TTO's are generally dysfunctional. Is there a "standard deal" that could be put in place for university research out-licensing?
Future Feed in Australia is a rather functional example. They hold the patent for anti-methane red seaweed and their open licensing of it has spawned about a dozen companies.
Or could we imagine a company that controls all IP generated from a University and has the mission of licensing it out efficiently and for financial benefit of its stakeholders? You could think of this as a JV between the university and private capital. It'd be easy to imagine it building an incubator as well.
The Hardcore Institute of Technology proposal does not make total sense to me. It just sounds like an R1 without the degrees.
At Caltech where I got my Ph.D., nobody cared about my grades there. They care about my production of science and engineering, i.e. my portfolio. For engineering disciplines, the focus on high-impact papers is muuuuch lower compared to practical invention. Yes, I got a degree, but basically as an official stamp for completing my apprentice project.
Even at MIT, my undergrad, my professors told me "get good grades or grad schools will think you're dumb, but that won't get you into a Ph.D. program. What will get you into a PhD program is impressive work as an undergrad at a research lab." And this extended to admissions - I got into MIT because of being an Intel Semifinalist. They'd rejected my early admission and deferred me to main admission. So I sent them a packet with my Intel Semifinalist win and tehy sent me an admission letter a week later.
1. The experiment unfortunately has too many confounders: soft-money professorships are lower status so there's a selection effect and also (I think) they get far less startup funds, among other things.
2. People have tried the standard deal and failed. Before Bayh-Dole there was a company that did all the licensing by default.
Caltech is basically the closest thing to the Hardcore Institute of Technology there is. I assume still that professors need their grad students to publish to get tenure though.
Thanks Ben. Caltech is really a hardcore science school - literally half of the school is the chemistry department. So it’s very much a “get nature/science publication or bust” place rather than a tech transfer & startup place. Engineering is secondary to science, and much of the engineering connects to JPL - which Caltech operates. The Caltech TTO is interesting because of that, because it’s mostly handling IP and technology from JPL.
Sure -- I just meant in terms of serious training. (I'm not sure if you're aware but I went there for undergrad so of course I'm a bit biased)
ah :-) another techer :-)
I do think Caltech is a bit different than how I imagine your model. Students and postdocs are siloed into Professors' labs per the typical academia model. When they tried to do something less siloed, the Joint Center for Artificial Photosynthesis, this structure was a serious burden to efficacy.
Thank you for a very interesting take on accomplishing an important objective.
You're welcome!
Thank you for sharing part of Ben's essay. I look forward to the next installments when they are posted here. I agree that universities have shifted from having specialized, targeted roles in the development of students and science to becoming more of a "jack of all trades," especially as great industrial research monoliths like Bell Labs, as mentioned in the article, have shrunk or disappeared. Now, teaching, fundamental research, applied research, patenting, and commercialization of technology are all housed within the academic (non-)industrial complex.
There have been some government efforts to patch the inefficiencies of university technology transfer, such as the SBIR and STTR programs (https://www.sbir.gov/). However, these programs do not fully address the underlying structural issues. They often operate as add-ons rather than true alternatives, and the incentives and cultures within universities are misaligned with the needs of effective technology transfer and commercialization. I’m interested to see potential solutions or new institutional models in the upcoming parts of this series. Thanks again for sparking such an important conversation!
There are lots of problems at Universities (having been a professor at an R1 for more than 20 years, and about to start my 10th year directing a University research center). As a result, I'm always interested in takes on exactly what they are and what can be done to address them.
A few questions to consider after my first read through:
1. Even the Universities with the highest overhead rates have much lower overhead rates than "unbundled" private, government, and industry research labs. Does the author believe that this is *not* this *because* of the economies of scale that come with bundling?
2. Does the author have any plans for reducing research overheads in his "unbundled" alternative, or does he believe that the increased tech transfer will justify the increased overhead rates?
3. The author appears to have a number of concerns with Universities extracting profit from their innovations and using that to subsifize future innovation (e.g. University tech transfer arms). I have lots of concerns about tech transfer arms of Universities, too. However, how does the author expect "unbundled" fundamental research institutions from going the way of Bell Labs, whose inablility to profit from its innovations was part of the reason for its demise?
4. How does the author expect people to be trained in to performing fundamental research if he unbundles fundamental research from the mentorship and experiences that come with that
I'm interested to see where this series goes, but I'm concerned that it may be taking a "why we should break up the economies of scale that have out-competed my preferred ideal" direction that would be unfortunate.
Thanks for reading!
1. There's certainly cross-subsidization that goes on in universities. The question is whether the hits to research effectiveness is worth it.
2. Insofar as new institutions should have business models that don't just rely on grants, sure. But there's a longer discussion about why the idea of overhead is silly to be had. The rest of the world just calls it "margin"
3. I have no problem at all with organizations extracting profit from their innovations. I thing tech transfer as it exists is terrible, and have weird feelings about privatizing innovations funded by public money.
4. See "Hardcore Institute of Technology" also I am not advocating for getting rid of universities, I think training is one of the things they do very well.
Cross-subsidization and economies of scale aren't always the same thing, though. To pick an example, I direct the HPC center at my University, and am occasionally asked "why shouldn't our University researchers just use [pick your favorite cloud service] instead?"
It's a reasonable question, and a big part of the answer is simply economies of scale. Larger institutions have enough aggregate demand that it's cheaper (in many cases by a factor of 2 or more) to host the resources locally, while small institutions whose demand is bursty pay more per compute unit because they don't have enough demand to amortize the fixed costs of running on-premise large-scale computing resources.
Your answers get to two of the larger questions I have:
1. What research eco-system supports these unbundled research institutions? The reality is that modern research is infrastructure-expensive (computing, instrumentation, laboratory, etc.) and generally not something that can be efficiently done in a 1800s-era basement. Without economies of scale backing them these unbubdled institutions are going to be less efficient than larger organizations. I'm guessing the vision that these institutions will make up for that inefficiency with higher-quality results than come from current Universities and lower costs (and risk?) than come from industry labs or startups?
2. What is the financial model for these institutions? Are they public (like the FRO proposal) or private (like Bell Labs/BBN)? For profit or non-profit? Who gives them funds for what reason and on what timeline and term?
This vision of unbundled research institutions seems like it's trying to the a Goldilocks solution between large universities and small startups. Goldilocks solutions are attractive when they get the best of both worlds, but can also fail spectacularly when they end up with the worst of both worlds.
Why exactly are the unbundled research institutions being propose going to be the former and thrive and not the later and suffer the fate of the Bell Labs and Xerox PARC that inspired them?