[zmibes]

Just musing don't be angry with me, but seems rather circular that R&D needs public money in order to benefit the economy. Over-simplistic I suppose, but I'd think if it were economically productive it would be able to make a self-sustaining feedback loop.

edit: ideologues already attaK! sad to see ><

[bglazer]

The problem with private R&D is that it's quite difficult for any individual company to fully capture the benefits of research. There are lots of ways this happens. Company does innovative research then gets copied by other firms. Or, they don't recognize an idea's potential but another company does. Or, the research is useful twenty or fifty years in the future. Or, the research is very useful both for them and for many, many other fields and they're only able to capture a tiny sliver of the total value.

For a couple concrete examples: Xerox-PARC did incredibly innovative computing research that turned Apple into a trillion dollar company. For a more modern take: DeepSeek literally used ChatGPT to build their own cheaper competitor.

So, R&D is incredibly societally useful and it's in the collective interest of companies to have access to research results to keep them innovative and competitive. But, it doesn't make sense for any one company to actually do R&D. It really only works as a public good.

[zmibes]

In this scenario would it not be attractive for investors to fund multiple (all, even) companies conducting the research into a given field/advancement knowing they will loose money on most but one will capture the benefit? (which, per the premise, is greater than the cost)

[bglazer]

I think this happens to some degree with venture firms taking multiple bets on multiple companies working in the same area. The individual companies are still hamstrung, though. Why would ten companies all take the same bet on a speculative technology, especially if they know that other companies are already working on the same ideas?

Also, not to be glib, but it sounds like you’re describing a very wealthy investor willing to spend a lot of money to advance social good by broadly funding an individually unprofitable research goal. That's just a government right?

[zmibes]

thank you for this thoughtful response

[tonyarkles]

I'm not angry and I totally get your point of view. I don't think you're wrong.

My belief is that the underlying issue is that most companies and their drive for quarterly results means that they won't front a bunch of the "we're not sure if this will result in anything but it's an interesting thing to look into" style of research on their own. The Bell Labs of the olden days are gone and publicly-funded R&D has essentially replaced it.

It's not all bad though, having all of that research published instead of tucked away in a private research facility can be beneficial.

[zmibes]

Which makes sense and also prompts the question: Why is research funded by public money published in journals that are not freely available to the public?

[roughly]

There’s a couple factors - one is the payoff time and risk levels make basic research difficult for even businesses with long time horizons to engage with, and American businesses are notoriously short term. The other is that there’s a large public benefit to making the fruits of that research widely available by way of risk mitigation - it’s commercial companies that take the research over the line, but making the research public allows multiple companies to take a shot at commercializing the results,

In general, considering the government and public money to not be part of “the economy” will make your internal models less performant - the public is an actor in the economy and so is the government, and both make decisions on the basis of their needs, values, and resourcing. Those entities seeing higher rates of return for their investment than other actors like private companies is absolutely consistent with, for example, different companies seeing different rates of return for the same investment dependent on their needs, resourcing, and constraints.

[teraflop]

It's no more circular than spending money on education to get a high-paying job, or spending money on changing your A/C filters so that your air conditioner doesn't prematurely fail.

> but I'd think if it were economically productive it would be able to make a self-sustaining feedback loop.

A "self-sustaining" feedback loop still has humans in the loop, deciding to reinvest money in future improvements. If those humans decide to shut the loop down then obviously the benefits will stop being realized.

[zmibes]

why do we need the state in the loop?

[connor4312]

For the same reason government in general (e.g. the US military) isn't funded by one big GoFundMe. The marginal value any individual actor gains from their investment in public research or services is almost zero. It only works when it's prescriptive on a large scale. See: the tragedy of the commons.

[myrmidon]

Because collective well-being/improvements to the nation as a whole is literally its purpose.

Unlike companies, whose purpose is to earn profits for owners and shareholders.

[Joel_Mckay]

Some projects are too expensive, risky, and or potentially dangerous for a private firm to handle.

Higher education institutions in North America often already have close financial relationships with private sector firms.

Many startups were founded in a lab on some campus. =3

[melenaboija]

Yeah, that’s an oversimplification and short-sighted. Do you really think even a small fraction of the AI research that led to LLMs and trillions in corporate market capitalization was funded purely by private money? No, it was largely driven by public funding. Even more, not just from the US, but from governments and academic institutions around the world over several decades.

[zmibes]

sorry to hear it was short-sighted. How did you come to those figures on comparison of public and private money into LLM research?