Comment by Yoric | Hacker News Reader

Comment by Yoric | Hacker News Reader

Parent Full thread

Historically, scientific progress has extremely rarely been made by companies. Most of the time, it's the result of academic research. Most researchers typically have neither time nor inclination to build a company/a product based on their own research, and having great research is no guarantee that the company/product will succeed. So the way research results become visible is typically through publications and teaching.

When companies hire graduates who have learnt from researchers, some of these graduates end up in position to "innovate", quite a few years after the actual research has been done.

For instance, I'm going to talk about the field that I know best: programming languages (to keep it simple, I'm not talking of VMs or compilers, just the languages). Pretty much everything you see in Java, C#, Python or Go dates back to the 70s (with broader testing and quality of life improvements, of course), Swift gets a few improvements imported from the 80s, but not that many. The only industrial languages that seem to have real innovations are F#, Rust and Scala, which are three cases in which the actual researchers managed to convince (or found) a company to support the language.

Anyway, it's really, really hard to measure scientific progress, and if you look at companies to try and gauge it, you're looking at stuff that is typically quite old.

Apple hired people who did academic work on LLVM and static analysis, and this turned into real compilers that people use every day (as well as Swift, which enabled SwiftUI.) There are many other examples at other companies.

I would also note: in spite of "novelty" requirements for publication, we tend to see many of the same ideas recycled over and over. Which makes sense, because they tend to be good and useful ideas that may be fundamental to the discipline.

I didn't want to talk about compilers, because examples are different, and despite the fact that I work on compilers at the moment, I don't have as much in-depth knowledge as on programming languages. In particular, while I'm sure that there are a number of novel things in LLVM, I have no clue which ones. In Swift, though? I see quality of life improvements, but nothing remotely scientifically novel in terms.

Agreed about ideas being recycled over and over in academia.

Regardless, I believe that my point holds: it's very hard to judge "scientific progress" by looking at industry, because most of the time, you're looking at stuff that was discovered decades earlier.

Aside from the transistor, negative feedback, Unix, troff, radio astronomy, the charge coupled device, cryptography, information theory, what did Bell Labs ever do for us?

Is that a Life Of Brian reference? :D

Its hard to argue Bell Labs was much of a purely private company, AT&T was bound by law to invest a certain amount of money in hard scientific research, in return for its monopoly position in telecom.

So at the very least, bell labs was the direct result of regulation, and pretty arguably just an alternate form of taxation (AT&T was also legally bared from commercializing Bell Lab's research outside of telecom)

Lasers.

Good point!

1. I will admit that there are exceptions :)

2. Clearly, I should have written "industry" instead of "companies".

Barring any error in my memory Bell Labs was a pure research lab at the time, wasn't it, i.e. not an industry research branch, right?

The issue is that academic research doesn't always translate into a product that can be useful. There are many things that can be done at small scale that look awesome, but once you try to apply it at a larger scale or trying to account for all the corner cases, it starts to fall apart. While this may not be as true for tech, it tends to happen a lot in the physical sciences.

A big recent tech-focused one that started in academia and is now in the hands of large companies is self driving cars. It looks like it may work, but then every started hitting those 1% cases and realized they can't actually ship a fully self driving car (and driver-assist features can lead to worse issues as people think they can trust the tech more than they should).

I'm not sure how that's an issue. If you privatize profit from the innovation of others, it's only fair you privatize at least some risk as well.

The only potential maybe issue is that may limit private investment - but that wasn't how it was intended to be funded anyway.

It has been asserted that the best innovations come from baser instincts - the ability to go to war. Radar came from defense research. I hate this.

As did lots of awesome materials. The material science advancements that happened with WW2 and the space race (fueled by the cold war) are pretty amazing. Nylon among many others came out of those drives.

Edit: someone corrected me about nylon. It was invented between 1927 and 1938 my Dupont. But it was still a commercial Enterprise doing the r&d for it.

Looks like synthetic rubber and synthetic oil two major ones to come out of world war II https://www.warhistoryonline.com/instant-articles/inventions...

nylon did not. you’re probably thinking of teflon, which also did not, but is at least a common misconception, unlike nylon, which is widely known to have first been used for women’s ‘nylons’ in the 30s.

Also during war patents are shared between all companies in one country or ignored all together. I think this is an under appreciated fact of why wartime creates all sorts of innovation.

That's a very interesting point! I wonder if allowing free use of information could result in the same benefits during peacetime.

Well that's how we got emacs so who knows.

Woah, I never thought about that. Do you know where I might be able to read more about copyrights / patent enforcement (or lack thereof) during wartime?

Here is one article discussing the US government pressuring aircraft engine manufacturers to cross license all their patents[1].

[1]https://history.nasa.gov/SP-4103/ch2.htm

That’s an interesting theory.

The best innovations come without profit-driven motives. The thing about defense research is that it's more popular with the public than, let's call it, nerd research.

It's having money without trying to optimize solely on future money.

It's probably just as easy to say that funding comes from the war economy rather than research. However, I am also not entirely surprised that it's the high levels of funding combined with the adversarial nature of war that pushes the envelope of things getting bigger, smaller, faster, higher, etc.

However, while we do see this trickle back into civilian life, I imagine that if funding were directed towards solving civil needs instead (housing, food, water, environmental contamination, energy, community centers, schools, etc) we'd see very different things produced, and often more well suited for peaceful purposes.

What about the fact that some of the most prosperous times for average people come after epidemics and such which result in massive population declines?

My pet theory is that the best innovations come from contexts where loss of human life is acceptable - war being one such case.

The flip side of greater consumer safety is that there's less room to get things wrong, which in turn means it's harder to meaningfully innovate.

If my theory is correct, I expect the next paradigm shift will come from a country with less consumer/worker/civil protection. China (and to a lesser extent, India) are on my watch-list.

When I say this, people sometimes pigeonhole me as an anti-regulation nutcase, but that's not me. I don't know how I feel about this. There's a real Kantian dilemma, here.

I would say certain types of problem domains will be more likely to find solutions in academia vs companies.

For example, it's going to be pretty hard to get much innovation from academia in the area of operating planet scale networks/datacenters simply because they don't have access to such environments and the money to support it nor the day to day challenges of actually running something like this.

That said, much of innovation happening in companies is likely part of the secret sauce that makes the company successful so we're not going to see much of it make it out at the time the innovation takes place but over time companies and people working for them do open up about their solutions.

Haskell is definitely another language that comes to mind when discussing research and innovation

Definitely. And definitely not industrial :)

Reply on news.ycombinator.com