Comment by linguae

I believe this depends on the type of research that is being done. There are certain types of research that benefit from our current research grant system and from VC funding. The former is good when the research has a clear impact (whether it is social or business), and the latter is good when there is a good chance the research could be part of a successful business venture. There are also plenty of applied research labs where the research agenda is tightly aligned with business needs. We have seen the fruits of applied research in all sorts of areas, such as self-driving vehicles, Web-scale software infrastructure (MapReduce, Spark, BigTable, Spanner, etc.), deep learning, large language models, and more.

As big of a fan I am of Xerox PARC and Bell Labs, I don't want to come across as saying that the Bell Labs and Xerox PARC models of research are the only ways to do research. Indeed, Bell Labs couldn't convert many of its research ideas to products due to the agreement AT&T made with the federal government not to expand into other businesses, and Xerox PARC infamously failed to successfully monetize many of its inventions, and many of these researchers left Xerox for other companies who saw the business potential in their work, such as Apple, Adobe, and Microsoft, to name a few.

However, the problem with our current system of grants and VC funding is that they are not a good fit for riskier avenues of research where the impacts cannot be immediately seen, or the impact will take many years to develop. I am reminded of Alan Kay's comments (https://worrydream.com/2017-12-30-alan/) on how NSF grants require an explanation of how the researchers plan to solve the problem, which precludes exploratory research where one doesn't know how to attack the problem. Now, once again, this question from the NSF is not inappropriate; there are different phases of research, and coming up with an "attack plan" that is reasonable and is backed by a command of the prior art and a track record of solving other problems is part of research; all PhD programs have some sort of thesis proposal that requires answering the same question the NSF asks in its proposals. With that said, there is still the early phase of research where researchers are formulating the question, and where researchers are trying to figure out how they'd go about solving the problem. This early phase of research is part of research, too.

I think the funding situation for research depends on the type of research being done. For more applied research that has more obvious impact, especially business impact, then I believe there are plenty of opportunities out there that are more appropriate than old-school industrial research labs. However, for more speculative work where impacts are harder to see or where they are not immediate, the funding situation is much more difficult today compared to in the past where industrial research labs were less driven by the bottom line, and when academics had fewer "publish-or-perish" pressures.