5 ms·

Our biggest problem is that research is expected to be in Academia. Not withstanding few exceptions such as DeepMind and FAIR, vast majority of industrial resea

Our biggest problem is that research is expected to be in Academia. Not withstanding few exceptions such as DeepMind and FAIR, vast majority of industrial research is by definition short term. For example, the founding document of Microsoft Research asserted that they only want to hire researchers who are willing to work on relevant problems with 2 year horizon with emphasis on product integrations. Similarly, even at Bell Labs research was expected to help phone lines (a fun story being C/Unix were underground projects, borderline "unapproved"). 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.

why_only_15 7y ago
The US puts 2.79% of its GDP towards research. It's arguable this should be increased, but ~$550b is an impressive figure in any case. It would be interesting to see what exactly is included in this figure and how much of that is 'basic' etc.
- sytelus 7y ago
  Exactly. A very little of it is basic and vast majority of it simply usual development projects that you have to do anyway. This is similer to big tech who oftne claims massive expenditure in "R&D" but R part typically gets 1% and D part gets 99%. For reference NSF is less than $8B, a chump change for US GDP of $18T.
linguae 7y ago
The trouble with this industry view that research is expected to be done in academia is that academia isn't immune from these short-term pressures. There are some five-year NSF grants and some other opportunities for long-term funding, but many grants are shorter than that, and they also seem to favor evolutionary work rather than revolutionary work. Here is a quote from Alan Kay (http://worrydream.com/2017-12-30-alan/ http://worrydream.com/2017-12-30-alan/): "[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?

why_only_15 7y ago

The US puts 2.79% of its GDP towards research. It's arguable this should be increased, but ~$550b is an impressive figure in any case. It would be interesting to see what exactly is included in this figure and how much of that is 'basic' etc.

sytelus 7y ago
Exactly. A very little of it is basic and vast majority of it simply usual development projects that you have to do anyway. This is similer to big tech who oftne claims massive expenditure in "R&D" but R part typically gets 1% and D part gets 99%. For reference NSF is less than $8B, a chump change for US GDP of $18T.

linguae 7y ago

The trouble with this industry view that research is expected to be done in academia is that academia isn't immune from these short-term pressures. There are some five-year NSF grants and some other opportunities for long-term funding, but many grants are shorter than that, and they also seem to favor evolutionary work rather than revolutionary work. Here is a quote from Alan Kay (http://worrydream.com/2017-12-30-alan/ http://worrydream.com/2017-12-30-alan/): "[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?