Comment by teleforce
1 year ago
There's side effect benefit of big kahuna companies mainly on the significant breakthrough and game changing research output because these excellent researchers are paid handsome money compared to conventional universities or research institutions.
We saw this with AT&T Bell research labs with their inventions of transistor and Unix, among others. The same thing happened with Google research with (arguably) deep learning and transformer.
Split them up at your own (US) perils, not unlike killing own Golden Goose.
Drawing from own experience working with Harvard, MIT and Google researchers, I could not disagree more.
When you talk to a researcher, do they strike you as someone who chases handsome amounts of money, or someone who chases ideas?
You bring up research labs. I listened to Alan Kay's numerous talks over the years (as an example of a prominent CS researcher), not once does he mention that he joined for the money at Xerox PARC. Yes, he was paid, but the main advantage was being given free reign to conduct research with the best experts in their fields, i.e. to invent and pursue ideas.
The important part from a financial perspective, is to be able to have finances to back a research division, where you can spend billions on building a new type of technology, if need be, that may not pan out. You don't need a monopoly to accomplish that.
You know who does chase handsome amounts of money? Day traders and everyone gambling on the stock market.
Because you specifically mention Alan Kay, I just finished reading “Dealers of Lightning”, which is about PARC, and says that the researchers there were very handsomely paid. IIRC, they were paid 20% more than their counterparts in ‘regular’ Xerox R&D. Xerox was also a big company, making a lot of money when it started PARC; arguably a monopoly (depending on how you define the term).
That still doesn't prove that Kay worked there because of the financial incentives. I don't think that 20% is nearly enough for most top talent to bail to an otherwise less attractive company if they deeply care about what they're working on so long as they aren't wildly underpaid. There needs to be a combination of incentives to drive movement.
2 replies →
> is to be able to have finances to back a research division, where you can spend billions on building a new type of technology, if need be, that may not pan out. You don't need a monopoly to accomplish that
A company in an industry with very tight margins has much less money to invest in fundamental research. All the recent growth in generative AI has been driven by companies with very high margins; Google, Facebook, Amazon. If all those FANG were in tightly competitive markets and hence had low margins, they wouldn't have had billions of dollars to spend on the GPU compute necessary to develop modern language models. Which is evidenced by the fact that no companies in more competitive sectors have produced any large language models.
> company in an industry with very tight margins has much less money to invest in fundamental research
OpenAI has raised almost $18bn to date [1]. That puts it in the top 10 corporate R&D spenders globally, ahead of Intel and the entirety of big pharma [2]. (And OpenAI's gross margins for its API business are estimated around 40% [3]. Standard fare for tech. If anything, OpenAI subsidising its business with Apple and ChatGPT is behaving more like a tech giant than a start-up.)
The top of that list are the big 5 American tech companies, spending about $200bn annually on R&D. By coincidence, that's roughly the pace of U.S. VC spend [4]. The depth of American private capital markets make a solid case against favouring housing these long-shot bets inside tech giants. (Particularly absent non-compete and IP reform.)
[1] https://techcrunch.com/2024/10/02/openai-raises-6-6b-and-is-...
[2] https://en.wikipedia.org/wiki/List_of_companies_by_research_...
[3] https://www.theinformation.com/articles/a-peek-behind-openai...
[4] https://www.reuters.com/business/finance/ai-deals-lift-us-ve...
I think you're getting it backwards. The research operations are a desperate attempt stave off regulation to keep the sweet, sweet monopoly profits coming in (and those profits are so big that the bean-counters allow it). I believe that was the explicit strategy at AT&T. We collectively pay way, way more.
It'd be way more efficient and cost effective to just set up a well-funded government labs to do that research.
It's an interesting argument for monopolies possibly being a net-good, but I don't think regulators really look at it. Companies do R&D because they like their monopoly status and don't want to be caught flat-footed by something new.
Yes. If a large company didn't employ those researchers, someone else would, and if they were for someone else maybe they'll come up with something which could damage the large company.
It's not about building and owning the next best thing, it's about preventing someone else building and owning the next best thing.
Well then the monopoly still isn't bad, because they can still be brought down solely by merit (just that no one have done it yet). Unlike government-enforced monopolies that'll stay no matter what they don't even have to do R&D.
I am not at all confident that a government lab will be either well-funded or efficient in any sense of the words.
It’s this kind of baseless negativity that is toxic to mental health. I’m done with this site for a while. It’s just too exhausting reading lies about how much everything sucks.
https://en.wikipedia.org/wiki/ARPANET
TCP/IP?
I love it when the top comments on a hacker news thread is justifying monopolies. Fuck economics right?
Do you think it's impossible to have a nuanced discussion about monopolies? Their net effect may be wholly negative while having some interesting aspects
Not impossible, but mostly impossible. You can discuss the interesting aspects of large corporations, but you can't really discuss them in a vacuum. The top level post about "big kahuna" companies comes across as an unambiguous defense of monopolies, not an attempt at nuanced conversation.
2 replies →
You should know 80% of Hacker News works for either Google or Facebook.
Honestly, it's disgusting.
Have these people even read the white papers that Google releases? They are mostly marketing pieces.
When systems and technologies are not publicly reproducible, why should scientists and (most) engineers care? I will not take Google at its word and would not recommend it to others.
> side effect benefit of big kahuna companies mainly on the significant breakthrough and game changing research output
“Given that production could be carried on without any organization, Coase asks, 'Why and under what conditions should we expect firms to emerge?' Since modern firms can only emerge when an entrepreneur of some sort begins to hire people, Coase's analysis proceeds by considering the conditions under which it makes sense for an entrepreneur to seek hired help instead of contracting out for some particular task. The traditional economic theory of the time suggested that, because the market is ‘efficient’ (that is, those who are best at providing each good or service most cheaply are already doing so), it should always be cheaper to contract out than to hire.
Coase noted, however, that there are a number of transaction costs to using the market; the cost of obtaining a good or service via the market is actually more than just the price of the good. Other costs, including search and information costs, bargaining costs, keeping trade secrets, and policing and enforcement costs, can all potentially add to the cost of procuring something via the market. This suggests that firms will arise when they can arrange to produce what they need internally, and somehow avoid these costs.
There is a natural limit to what can be produced internally, however. Coase notices ‘decreasing returns to the entrepreneur function’, including increasing overhead costs and increasing propensity for an overwhelmed manager to make mistakes in resource allocation. This is a countervailing cost to the use of the firm.
Coase argues that the size of a firm (as measured by how many contractual relations are ‘internal’ to the firm and how many ‘external’) is a result of finding an optimal balance between the competing tendencies of the costs outlined above. In general, making the firm larger will initially be advantageous, but the decreasing returns indicated above will eventually kick in, preventing the firm from growing indefinitely.”
https://en.m.wikipedia.org/wiki/The_Nature_of_the_Firm
Trying to turn a zoo into a farm, as AT&T attempted to do with Bell Labs post-divestiture, had limited success and incurred great emotional and spiritual cost on the institution. Bleah.
just skip the middle man skimming off the top and 10x the national labs funding.
I wish this were a viable option, but it is not. US national labs are horribly, horribly mismanaged. For some slower-moving fields like particle physics where institutional knowledge is key, they hold up alright, but for fast moving fields like quantum they are very behind. They are stagnant bureaucracies. I could tell stories, but better to just compare the output of national labs in many fields to those of the top universities in the States.
I think you need to better support the contention that the National Labs are “horribly, horribly mismanaged” [not even just horribly, but horribly, horribly]. I think many of us would like to hear your stories. But note that many in your audience here have decades of experience across both National Labs and leading industrial laboratories. Please, share your stories that span the contributions of tens of thousands of top-level STEM contributors across practically every area of scientific and engineering endeavor over the last, say, 15 years. Remember to stay unclassified…
1 reply →
Then separate out the basic science from the defense work which requires that bureaucratic oversight. Or direct that funding to universities. My main point is allowing monopolies because they direct their excess profits to research to hide their excess profits is just a complicated tax to fund basic research, which we would be better off spending directly on research without bloated executive salaries and distorted markets in e.g. search or browsers.
This! And where do people think open source funding, hackathons, bug bounties for software that's not even theirs, oss-fuzz, really incredible but not necessarily profitable research like Project Zero comes from? AdWords largesse.
That is like wishing a benevolent dictator, who'll be in most cases more efficient then democracy but is a huge risk to take.
The split of ATT killed Unix2, so we spent 30 years re-implementing Linux+k8s. These things that existed in Unix2 & Plan9 were re-implemented by Plan9 employees in Google Labs.
UNIX exists because ATT was split. They could not profit from software (by law because of an agreement with the government) so early versions of UNIX where made free.
This should be well known, simple google search:
https://www.linuxquestions.org/questions/linux-general-1/why...
i can't even understand what you are saying? AT&T was good, or bad?
AT&T copyrights led to linux, and linux, independent of unix, has been a huge boon for good, and for unixness.
the threat to unix now is all the people who by nature prefer Dave Cutlerness, and can't see that their way is the wrong way, now they are using linux (because it won) and trying to ruin it.
The AT&T split had nothing to do with monopoly regulation (as opposed to the Bell breakup in 1982), other than the fact that Wall Street wasn't rewarding regulated operating companies with dot-com valuations. AT&T wanted to sell hardware to other telcos and dot-coms, so spun off Lucent, which had no idea what it wanted to do with P9/Inferno (which was a fantastic piece of kit!) other than embed it into a couple of network products. Lucent bet heavily on unstable CLECs like Worldcom, generated a couple of headline-creating network crashes, and then failed to capitalize on their pole position in optical long-haul (to be fair, they also bet heavily on a very unstable Global Crossing for that). There's a lot of mismanagement and failures that can be ascribed to Lucent leadership without government or regulatory intervention being involved.
UNIX only became a success, because ATT initially wasn't allowed to charge real money for research work.
Seconding info about "unix2". I used to pour over the trade tabloids, and I've never heard of this.
Novell bought UNIX and has some grand plans for "SuperNOS", which also never shipped. It certainly wasn't anything like K8s.
Never even heard of Unix2 - was it a complete replacement ?
So wait, markets don't work, then? A free market, theoretically, promotes innovation by ensuring that businesses must advance their products in order to compete with one another. You're saying that a lack of competition promotes innovation by concentrating all of an industry's capital under one roof.
The thing that makes markets work is the struggle. A Darwinian survival of the fittest in a way. Once the struggle is over and only one contestant remains, the results are generally dystopian.
Also I believe that even when working optimally the Darwinian mechanism can't solve certain problems. Some things need to be dealt with by a group of motivated people working for other goals than profit.
Markets gave us compuserve and facebook while CERN gave us the open web, for example.
Yes I can see the dystopian consequences of google’s search monopoly profits, which they have used to do such horrible things as:
- Providing a free alternative to Microsoft’s monopolized office suite and desktop OS
- Provide a free alternative to Apple’s mobile OS, spurring a revolution in access to the internet for the world’s poor
- Provide free global maps with streetview sights
- Provide a free to access video platform with invaluable educational resources that allows millions of creators to make a living and that likely wouldn’t exist save for Google’s monumental investments and ability to sustain years of losses
- Research given away for free that ignited the current AI revolution
- Research given away for free that is revolutionizing medicine and drug development
In sum, truly a horrible thing they’ve done
3 replies →
Both can be true in varying degrees at certain points in time. They're not mutually exclusive. There are benefits to centralization and concentration of capital. Competition is the same exact process that leads to monopolistic entities in the first place.
Regalian roles are to ensure fair competition by reducing any actor bigger than the state to something smaller, and ensuring the economy works with transparent information (no lying, rule of law, etc.)
Companies getting too big are natural; Letting them get too big is what happens when your state borrows a trillion per semester: Your state is obese, intervening in every little sector of the economy (thus the opposite of liberal), and not playing its regalian role.
You should indeed reduce the size of both the state and the largest companies, to let the economy self-regulate, but then, how would the US govern the rest of the world?
Lina Khan's response to this argument. [1]
[1] https://www.youtube.com/live/L_QaZk5iJOA?si=ZkxBe1CHgagmcBcW...
What groundbreaking household name advancements has Microsoft Research yielded? Not all of these labs are equally revolutionary.
Clearly this model no longer works. Bell labs had 11 nobel prize winners. What did Google invent? Slightly better generative neural networks whose offsprings now pollute their search results?
You could interpret this the other way - Why has Jeff Dean been snubbed by the Nobel committee? Why hasn't Larry Page gotten a Nobel for inventing the search technology that half the planet now depends on? I don't know what category to put that one in, but there's some important results in lightspeed-limited communications in "The Datacenter as a Computer" that would be worth extending the Physics category for.
"search technology"
Google is an advertising company, search is a by-product and has been for a long time.
2 replies →
Spanner is one thing I'd say they invented but they built a whole bunch of really neat stuff in order to be able to run search, back in 1998. that they're this behemoth conglomerate that it's cool to hate on doesn't erase the fact that they had to build all sorts of new things when they were just starting out.
Google invented the ability to put an animated Gif inside a spreadsheet cell.
with COM and OLE, this was possible in excel on win3.11 !
Two of Google's researchers are getting the Nobel prize for AlphaFold2 this year.
DeepMind, not Google organically.
1 reply →
indexing against Nobel prizes is like trying to use patents as a proxy for innovation
Rory Sutherland has a great take on this which is (paraphrasing and my own interpretation):
Innovation is a lot easier when you have a lot of money to spend on R&D. In order to get that money, you can't compete on price b/c that's a race to the bottom. Instead, you want to focus on quality and/or customer service so that you become a monopoly and then can use monopoly profits to innovate to higher quality products and services.
Clip: https://www.tiktok.com/@rorysutherlandclips/video/7314765561...
Peter Thiel made a similar argument about monopolies years ago. As I pointed out then [1], the argument only shows that monopoly is good for the monopolist; it doesn't actually show that monopolies are good for society as a whole.
[1] http://blog.peterdonis.com/opinions/monopoly-money.html
classical MBA says that a firm can compete on price OR branding, unfair advantages (moat) notwithstanding. Competition in commodities is difficult but not impossible given a rational economic environment. Some would say that the modern expectation of returns on investment are irrational, and warp the economics around them too.
What on Earth... you do realize that antitrust regulation was the only reason we got Unix in the first place, right?
This is a theoretical benefit which is directly at odds with the benefits of competition in a healthy market. For google, my observation is the "big kahuna" benefit of google basically does not exist and competition needs to be restored. Google is famous for not innovating on anything successfully, they produce graveyards of trash. Instead what they do is buy other companies then enshittify them in an anti competitive dance towards causing more damage than productivity.
You really have to think about exactly how our modern markets work and why buyouts are such dominant strategy. It's only sometimes about taking what you buy then using it, it's mostly about taking what you buy to stifle competition these days.
Look at twitter and Vine, twitter bought then shut down vine as part of a standard operating procedure just to stifle competition, and they had so little interest in capitalizing on what they bought that it left a market gap so wide TikTok filled it instead. But usually these practices do not leave such big market gaps, usually they simply shut down competition successfully and the buyer wins. Then in many cases if the company owners refuse to be bought out, extreme anti-competitive practices begin to destroy their business, which will not be punished until long after the victims get shut down. So owners need to choose between a huge pay out, or their company getting destroyed. Owners tend to choose the former.
> Google is famous for not innovating on anything successfully, they produce graveyards of trash.
- AlphaFold (just won a Nobel prize)
- Transformers (the "T" in GPT)
- Waymo (autonomous vehicles)
- Sycamore (quantum computing)
These are just a few off the top of my head.
If your idea of innovation is a better RSS reader, then sure, I agree with you. But in terms of things that push the forefront of technology, I have a hard time thinking of another company with greater impact in recent years.
> "big kahuna" benefit of google basically does not exist
I just given you the deep learning and transformer benefits.
There's a reason why the darling of AI Renaissance namely transformer was not invented at MIT, Stanford or Berkeley.
>Google is famous for not innovating on anything successfully
PageRank
Gmail
Maps
MapReduce
Chrome
Protocol Buffers
Go
Gmail was revolutionary at the start, but stopped innovating 10 years ago - why don’t we still have a good search engine within it?
MapReduce would be invented anyway (I implemented it from scratch before learning of it’s existence).
Chrome is just a slightly upgraded Firefox (and novadays Safari is just as good if not better with ai)
PageRank was what gave Google monopoly, it’s not a result of monopoly.
Go - I can give you that. ProtoBuf - not my field, but isn’t it just a format that someone else would develop to fill a niche? (unlike say mp3 that had new compression algorithms baked in)
Maps - I can give you that. Some people might argue that it was an acquisition, but without Google’s muscle, Street View would not be feasible.
6 replies →
Maps was technically an acquisition (Where2). But like YouTube, Doubleclick, Google Docs (Writely), Translate (Word Lens), Google Flights (ITA) and many others, Google successfully grew these products into giants.
2 replies →
Comparing the innovations of Bell Labs with..... _Protobuf_ of all things makes me gag.
1 reply →
Go you can hardly call an innovation. All of the ideas existed previously, and it's a poor execution on those ideas for reasons that have been discussed on HN at length before. They created it to serve their own needs in conditioning the labor market to make their hiring process easier.
1 reply →
Didn't some of the early GPT work come out of Google?
2 replies →
So it's OK to burden the whole country with a monopoly as long as they fund a handful of Nobel prizes?
Competition is a prerequisite for healthy Capitalism. Lack of competition is the Achilles heel.
Please do tell us how Google is a burden to the whole country.
Is it the free maps? free mobile OS? free email? free cloud storage? free video service? free office suite? free desktop OS? free AI chat?
Read a book on tech entrepreneurship. The “goal” of most startups is to get purchased by a big tech company. That’s utterly fucked, and tacitly demonstrates the problem.
3 replies →