The US returns to the endless frontier?

The organization and funding of science is a key geopolitical competitive advantage – and badly underrated across most economies. The European Union has failed at organizing tightly around scientific challenges, relying on large flagship programs that fragment into systems for distributing money across member states and the US has lost track of the post-world war II investments and strategies that through Vannevar Bush gave it a singular position in the world of science and technology. That could now be changing through an interesting new bill that could be re.introduced in the US congress – The Endless Frontiers Act. The act is an attempt to re-focus the US on organization and funding of science in a creative way, and there are several really interesting ideas in the legislation. A good review and critique is available in this paper.

One of the things that is proposed is a return to the focus on problems. In the bill there is a suggestion for a Technology Directorate that would deal with “use-inspired research”:

The bill would establish within the National Science Foundation (NSF) a new Technology Directorate, alongside its several existing science and engineering directorates, to increase investment in basic sciences that are motivated by specific problems and needs. (This is
sometimes referred to as use-inspired basic research. The new directorate would be authorized to receive a very large budget, as much as $100 billion over its first five years, more than the NSF’s other directorates combined and a substantial increase over its FY2021 budget
of approximately $8.5 billion.

This is exciting because it suggests that research would be forced to think through and rank what the core problems we need to think about are. Something that I think has long been missing in the organization of science and technology funding is a sense of where the rich problems are.

The idea of a rich problem largely corresponds with how Hilbert thought about interesting mathematical problems when he refocused the entire field of mathematics by pointing to a number of 23 key problems that needed to be solved in 1900. His problems became a structuring principle for a lot of mathematical research and still retain relevance for what are seen as rich problems. Hilbert describes how he choose his problems in an interesting way, and stresses clearness, accessibility (we should have some idea of where to start looking) and assessing how generative a solution would be, the fields and applications it would unlock (admitting that this is hard, but could be guided by experience – what the act calls use-based research).

This is also what the authors of the article referenced above are chafing against. They want to ensure that the funding of basic research remains intact – and with that they are essentially thinking about basic, undirected research. Let the mind go where it will and find what it can learn – that is the general sentiment underpinning this view of basic research and it is important to really try to understand why that is being put forward as an alternative to the use-inspired research.

The idea seems to be this: basic research is valuable because we cannot do what Hilbert did – we cannot identify important problems among the possible problems to work on. And furthermore, if we get stuck in thinking about applied science we risk becoming myopic and developing a better horse rather than finding new ways of transport. Oftentimes we do not know what research insights will be valuable, so we should refrain entirely from trying to direct our research efforts, lest we miss some really important insights that lurk in what looks like ephemeral ivory tower research programs.

That is a strong claim. But it is also one that is hard to prove. Do we really have no way of assigning importance over the set of possible scientific problems in basic research? Is there no way to categorize problems into more and less valuable to work on? If so the individual researcher would be at a loss to decide where to start, it seems – and if we think about how problems are actually selected today we see that there is indeed a valuation going on.

The real challenge then is to decide if the way we select problems today is better than focusing on the notion use-inspired research or Hilbertian problem sets.

So how are problems chosen today? It would seem that problems are primarily chosen through social mechanisms – working on what no one works on is hard, and the more collaborative science gets, the harder it becomes to break ranks with the on-going research programs. This is something that retard a field massively over many years. An instructive case study may be how a social network of scientists held back the growth of artificial intelligence by dismissing neural networks and “perceptrons” as unimportant research paths.

Soliciting problems from a broader group, and without the implied peer pressure of academia, might be a more interesting path to take – and here we should not underestimate the value of involving the private sector much more. There is a lot of value in thinking through how research and development now has shifted out in many fields to private actors and building bridges to that research in ways that could create a more comprehensive problem set to focus on.

I think the basic outline for a research policy can be sketched out fairly quickly, and it is surprising that not more is being done here. The core elements that I would like to see are:

  1. More public funding and a better organization of that funding. Fast grants (ala Tyler Cowen).
  2. Focus on clear, accessible and generative problems – and solicit input for building long term problem sets that can be made the focus on decades of research.
  3. Increasing citizen participation in science and ensure that the scientific field really benefits from the openness and collaboration tools now available.
  4. Do not accept that basic research / applied research false dichotomy. Focus on how problems are chosen and improve that mechanism.
  5. Integrate private R&D in public efforts and remove any firewalls between corporations and public sector here.

There is more to be done, but these coarse grained ideas would at least get us started on a better path, and one that may ensure that we can meet the complexity of the challenges we are facing.

Leave a Reply