It's not even clear that the premise is true. There's lots of 'research' done in the big tech companies.
The biggest reason why companies don't seek to emulate "Dupont, Bell Labs, IBM, AT&T, Xerox, Kodak, GE", is probably that it reads like a list of textbox examples of "companies that failed to execute on their research findings", so clearly there was something wrong with this approach.
GE (under Jack Welch specifically) is a textbook example of how financialization and focusing on numbers at the expense of products destroys companies.
Kodak is a textbook example of disruption. Yes they failed to capitalize on digital cameras specifically, but their research in all other areas was very much acted upon.
My mental model as an outsider, is the vibe out of Google is that they push the most talented folks out via process / politics. Not intentionally, just the reality of squeezing the creative type employee / work. Replacing creative smarts which is difficult or impossible to measure, with operational smarts, more easily measured. Those creative smart people mostly go on to start up other companies.
Its worked out ok for Google and others, because there's little teeth to anti monopoly, so all the big tech players can just buy the successes, which is safer than trying to grow them (esp. once the talent left). I really have no idea if this is an accurate take as its mostly vibes, sans for a few of said smart Google folks I've met in startup land(s). Yet Google is so big, they could bleed all kinds of employees telling all kinds of stories and it could all be simply random. Yet at the same time I can't help but think about every aging tech companies biggest / best products being via acquisition.
While I think monopoly is bad, I don't know if ^ otherwise is so bad. Maybe its just creative type folks _should_ avoid big tech, and build their own labs. Capital and compute are readily available to people who can demonstrate success, and its easier than ever to build and experiment in some fields. i.e. if we had stricter capital accumulation associated taxes, maybe the ills of this process wouldn't be so bad.
Bureaucrat-ification isn't a phenomenon unique to Google - it happens at every company eventually.
It's really hard to describe why it's inevitable (there are a lot of factors).
But it's self-evident really. All of the major tech players started with a single innovation that afforded them enough revenue to acquire almost everything else in their portfolio.
Aside from search, the only major product Alphabet built-in house that meaningfully moves the needle revenue-wise is their cloud segment. Youtube was acquired - and it's effectively an extension of search.
Meta had to acquire Instagram and WhatsApp. Without those acquisitions, I have strong doubts they'd still be a major player today.
You can run through this exercise with Microsoft, Apple, Amazon, NVDA etc.
The common theme is they did 1 or 2 things really well, and got big enough to acquire/copy/bully smaller players out of the market.
What's crazy is most of them still rely on that one original thing they did well for >50% of revenue.
> It's really hard to describe why it's inevitable (there are a lot of factors).
I think there's a lot of small factors, but of those the biggest on (IMO) is the fallacy that throwing people at a problem gets it done faster. For some situations: yes, for all situations: no. And you need experience or some kind of sharp intuition to know when to expand and when not to expand.
Add more and more people to a job and they'll find ways to justify their value at the expense of efficiency. And there's a snowball effect from there as an org adds people who believe adding people is always good.
Then the corpo runs into layoffs and everyone throws their hands up and says "How could we have avoided this?" By not overhiring in the first place.
The story with 3M and PostIt Notes is that the idea was originally rejected my management. The inventors created a batch and distributed them to all the executive admin assistants. When they went back a second time, they had the assistants speak up otherwise there would not be any more.
I didn't bring up 3M because of the Post-it story, but because they're being a "general research" company. From open reel tapes to sticky tapes and everything in between.
Würth is also similar. They make seemingly everything in a segment (lubrication, fuel additives, cleaning, restoration, protection, etc. etc.).
The bigger problem today is that there is simply nothing more left to research. Everything that is being worked on are at most optimizations, which allways have a dollar spent vs dollar returned amount on them.
“While it is never safe to affirm that the future of Physical Science has no marvels in store even more astonishing than those of the past, it seems probable that most of the grand underlying principles have been firmly established and that further advances are to be sought chiefly in the rigorous application of these principles to all the phenomena which come under our notice.” Albert A. Michelson (yes, that Michelson, one half of Michelson-Morley), 1894
If it feels like there’s nothing for us engineers to research, that’s probably a sign we need more basic research from the scientists!
LLMs are just better google. In the past, you used to google shit, and copy paste from stack overflow, now you just skip the middle man and go directly to Chat GPT. Anyone that has been programming for a while can attest to that the answers aren't any better, its just more efficient to iterate on them now.
AI hasn't even begun to be solved yet. Everyone is focused on feedforward transformer architecture that is never going to replace the imperative processing of actual intelligence.
Smartphones are pretty much solved, as they have replaced a lot of the need for in person interaction (which by extension means transportation). The last decade has been all about monetizing smartphones.
Wearables aren't transforming society at all.
3d printing and home fab is still too niche and expensive for most people, and you can't really make it cheaper and more accessible.
Electric vehicles largely suck. Self driving is mediocre.
We literally went through a pandemic and people got richer because they had to stay at home and not spend money on things like daycare or gas or car maintenance, without losing any productivity.
Hell, the state the US is in currently is largely explained by the fact that most all the problems in society have been solved to the extent that people have to invent bogeymen and elect a demented felon into office on the promise of solving those problems.
This is a very surface level analysis like saying that the automobile was just an iterative improvement over a horse. Or a computer is just a better abacus. Fundamental research is all about diving into the weeds and finding new problems to solve. It's true that some of the "low hanging fruit" no longer exists (you won't see someone like Euler or Newton who's names pop up all over the place), but I can promise you that real gains are being made on a lower level. These small gains in fundamental research snowball into bigger advancements. As an example, the transformer architecture used by LLMs was first published in 2017.
Automobile was improvement over the horse because things needed to get places. To improve on current automobile will require either massive government investment and regulation in the sense of flying cars, or full electrification with paradigm shifts in transportation, like induction charging roads or battery hot swaps or whatever else. The modern Corolla Hybrid is pretry much the peak optimal point of transportation.
What do humans need right now to improve their lives substantially?
high temperature superconducting would cause a big leap. cheap energy would also help. cheap compute-in-the-walls. machines doing all the dangerous jobs.
Cheap energy is possible now with solar. There is a reason why it hasn't been done yet. Nobody has a need for it. Remember, you may think it would be nice to have an electric car you can charge for microcents a mile, but most people dgaf about putting gas in their car.
Machines doing dangerous jobs also is a thing these days.
High temperature superconducting can potentially be useful in a few applications that involve high current, which mostly deal with transportation. The only real advantage of this is drone delivery service becoming cheaper, but that has big hurdles to cross.
There is a reason why being a streamer is the top choice of "what I wanna be" when you ask kids. Everything is about the internet now in terms of motivation. And unfortunately there, we already hit a hard limit of the speed of light.
oh, I was thinking about science. material science is doing some pretty cool things. quantum is getting interesting. we're just starting to really get a handle on reverse engineering the cell. battery chemistry. whether or not we're going to see practical fusion it seems likely that we'll see knockoffs. I just saw an ad yesterday that Avalanche is planning on selling waste (I mean useful quasi-stable elements). not just that but the non-sexy science (I met a guy yesterday and we talked about how a lot of his colleagues got the axe. he's working on characterizing the response of skin tissue to uv damage. that doesn't sound that sexy, but wouldn't it be nice to know?)
yeah, mostly forget about computers, we're still just coming to grips with the fact that we stopped doing largely innovative work decades ago. my bet is its going to go back to being interesting pretty soon. we are having a lot of interesting discussion about cognition though :)
All of this is research from the 90s, with a few decades of polish.
Now maybe we could start looking at what research labs have come up with since then.
> Hell, the state the US is in currently is largely explained by the fact that most all the problems in society have been solved to the extent that people have to invent bogeymen and elect a demented felon into office on the promise of solving those problems.
That's... an interesting point. I don't really buy it, though. The same could have been said of the fascist movement in Italy, or the royalists in France in 1905.
The biggest reason why companies don't seek to emulate "Dupont, Bell Labs, IBM, AT&T, Xerox, Kodak, GE", is probably that it reads like a list of textbox examples of "companies that failed to execute on their research findings", so clearly there was something wrong with this approach.