Comment by txcwpalpha | Hacker News Reader

Comment by txcwpalpha | Hacker News Reader

Parent Full thread

> The public sector builds the infrastructure, often following decades of investment and work.

It does? I don't think I follow. In your list of components, every single one of those is, at least in the US, largely or almost entirely handled by private companies.

The big semiconductor companies are private. I actually don't think there are any notable public entities that make their own chips. Hardware companies (I'm assuming you're talking about things like motherboards, routers, switches, etc) are private. Fiber optics/communications/networks are laid almost entirely by private telecom companies (and there's actually a big push to take this away from private companies and make ISPs be government entities). The article that you're commenting on is all about a private entity investing money into laying fiber and improving the protocols that communicate over it.

>I have rarely seen a start up on improving optical fiber or electronic chips.

There are a lot of SMBs working on chip design. I'm less familiar with fiber, but a quick google shows at least a couple, all private.

Oh no!

Consider high speed optical communication. The invention of transistor, laser, optical fiber, optical amplifiers, modulators, communication and information theory, DSP, and many other important technologies, was funded by taxpayers for a long time. You may say, some of it happened in old bell labs, but bell labs was a state-subsidized monopoly, rather an exception, and even things like information theory were really developed at MIT and other public research labs (Shannon had an office at Bell Labs for some time but had limited contact there).

Consider machine learning. From 1940s to 2010, neural networks were funded by governments grants, and developed at universities. The governments even funded building practical applications out of deep neural networks, for example, in the area of speech recognition, unsuccessfully. The periods are now called AI winters. It’s really in the past 15 years that the subject has become practical and companies are building products based on it. FANG is capturing the profits of this decades-long research and development.

Today there are millions of graduate students and postdocs and professors at universities performing risky research and experiments. A lot of it doesn’t pan out. The public bears the risk. Once an idea begins to work, it’s handed over to private sector for further development and bringing the work to market.

Most of the research papers published at conferences and journals, Nobel prizes, Fields Medalists etc come out of universities, not startups or FANG.

Yes, Google installed fiber cables (perhaps even bought it from Corning or the like, and also contracted out the installment). But that’s not infrastructure R&D.

I think there is no denying that basic long term research is done by and large in public sector. Private sector is still focused on short term product research.

Reply on news.ycombinator.com