Hacker News Reader

Comment by laughing_man | Hacker News Reader

1987, literally the peak cold war military spending, is an interesting year for comparison. Much of that high tech manufacturing (and employment) was underwritten by the taxpayers through the Pentagon, and that military tech eventually made its way into the civilian market.

Very soon after the US took a decade-long "procurement holiday", and we lost an enormous amount of manufacturing expertise.

Can we bring those jobs back? Sure, with a lot of tax money. Do we want to? I do - the value of "service economy" jobs is in free fall as companies replace white collar employees with LLMs.

View on HN

Is this really about military spending? The US used to make so many products and appliances domestically - everything from steel to uranium to doorknobs to refrigerators to chemicals to CRT displays to roofing shingles, and everything in between.

The military benefited from the massive industrial base that supported this production - but it didn’t create it.

And now the loss of that domestic manufacturing base is largely why military production - and indeed, any kind of large-scale endeavor in the US, including construction - is slow and expensive.

An interesting rabbit-hole I fell into the other day, was researching the French company Vantiva, the maker of my ISP's fiber ONT (and many previous cable modems I've had from other ISPs, under their previous names Technicolor / Thomson.)

If you trace out why Vantiva makes modems, it turns out that it's because all the patents for cable modems / coaxial carrier-network signal modulation and amplification were filed by RCA; GE bought RCA; and then GE divested its own + RCA's consumer-electronics businesses (including these patents!), selling that unit to Vantiva.

Presumably, cable modems were the kind of industry that only support a few big players, because there's not enough margin there after licensing costs are paid to patent-holders. The default winner of such a market would be a vertically-integrated player who holds those patents and can therefore make cable modems without licensing them.

That player was RCA (American); then GE (American); but is now Vantiva (French).

Vantiva released the cable-modem patents into a free IP sharing consortium kind of thing 15+ years ago now; but only once they already had an extremely dominant position in the space, with existing contracts with pretty much every ISP, such that they could be assured continued dominance even without the weight of licensing pressing down on all their competitors' backs.

Whenever I read about tech IP, I run into similar stories to this one. Some American company owned some tech innovation, but sold it to an overseas buyer some time in the late 80s / early 90s. And now it's not worth it to make that thing in America any more, because doing so would require licensing that IP from its current overseas owner.

(Perhaps, if America wants to be competitive, the government should encourage American firms with lots of free cash to [re-]acquire foreign companies that hold especially-valuable IP. Then, at least the IP incentives would lean in favor of vertical integration of manufacturing within the US.)

The article identifies three areas as "high tech manufacturing": Computer and electronic products, Pharmaceuticals & medicine, and Aerospace products and parts.

It's likely microchip development would have happened somewhere else without large US defense contracts. The industry was literally created by defense contracts for things like spy satellites, combat aircraft, and ICBMs.

Same with aerospace. The entire commercial aerospace industry exists because of bomber development in WW II, and all the engine tech since was initially developed and deployed in military aircraft. Carbon fiber, radar, fly-by-wire, plus the entire manufacturing process that produces aircraft that work the first time were all technologies developed or made practical to manufacture for the military. There's a reason passenger aircraft aren't manufactured in countries without a lot of military spending - companies like Boeing, Airbus, UAC, and Comac all depend on military contracts to pay development costs.

What constitutes high tech manufacturing is probably a bit different today. I would include sectors like exotic materials (superconductors, boutique alloys, and almost-here stuff like graphene and CNTs) and battery tech.

its not specifically military spending, just good government. If they pivoted after the cold war back into the space race or something, we would still have a high tech sector and it need not necessarily be military focussed

Could also have invested in public education and healthcare but noooo

The US spends twice what we spent in 1970 on education per student, in constant dollars. I don't know how to make the schools better, but money sure didn't do it.

And we're going broke paying for health care. I don't know how much more we spend, but it's multiples.

you're a decade too early. Compare it to 1980 and you see how it flattens out very quickly. Compare it to 1990 and you see the cost per student has basically flatlined. Spending diverted from government funding of schools to student loans, so even that's misleading.

>And we're going broke paying for health care.

Signing a bill giving trillions to billionaires certainly does make it hard to fund healthcare. Especially when the wealth concentration these days has the top 10% making 50% of the money. I wonder how we solve that...

A decade too early for what? Sure, it flattens out, but the fact remains we're spending a whole lot more money on schools than we did when the students were objectively learning more.

And what bill was it that gave trillions to billionaires?

The US is I the top ten of spring per pupil in the world. We also have more public school "administrators" per pupil than anywhere in the world.

Spending isn't the problem.

We have an enormous K-12 administration level partly because we have so many local school districts, each of which typically needs to separately buy textbooks, serve food, apply for state and federal funding and handle compliance, run HR and physical plant operations and janitorial and IT, manage transportation, respond to inquiries from the public, etc.

It’s very hard politically to merge school districts because even beyond labor considerations, people have a sense that their district is superior to the one next door and think a merger will create immediate chaos and long term harm to their kids’ educations and their property values.

Spending isn't the problem, when you ignore how the spending is composed.

We flattened out decades ago, and that's because we went from subsidizing schools to funding loans. The money makeup per studnet won't look different... until the student graduates and can't pay it off.

So our solution was obvious: make it so they can't bankrupt and stay in debt forever. Great way to build an educated citizenship.

Where did you see that white collar jobs are in a free fall? The employment rate is still tightly correlated with level of education as it always has been.

People who graduated in June are having a real struggle finding a job. Companies haven't laid many people off (outside of the ones that over-hired recently, like Meta and IBM) but they're not hiring. That's just the breeze that's heralding the storm. LLMs are going to make "any degree" jobs an increasingly rare thing as time goes on.

It seems a little early to be drawing trend lines from the experiences of June graduates.

Out of curiosity, If that storm comes, what do you predict will happen? I've been very curious what folks think is going to happen if the above becomes very true.

That's a really good question. I don't know.

It seems like jobs that require you to do something with your hands will be relatively more valuable than they are today. There will always be white collar jobs, but I suspect the pay and status associated will go down.

We go the road of the UK. At some point the billionaires finish their ransacking and all flee the US. GDP plummets, we hit a depression. We recover but pretty much lose the de facto superpower status.

Things keep creeping on but the struggle is real. And then we go the Germany route (again). We have the second coming of Donald Trump voted in promising to bring prosperity, but this one isn't just satisified with stroking his ego. So we initiate WWIII.

Depending on how that escalates or not is well beyond my abilities to guess. We could end up with a mass extinction that ends this era of civilization, we could be taken over by the rest of the world and have land diverted to various parts of the EU and Asia.

> People who graduated in June are having a real struggle finding a job.

So were people who graduated in 2007-2009. I guess it heralded the end of white collar jobs. Oh wait...no it didnt.

> LLMs are going to make "any degree" jobs an increasingly rare thing as time goes on.

If you actually believe this, it makes you one of the rare few who likely could actually be replaced with an LLM.

As I am retired, I can assure you beyond any doubt I cannot be replaced by an LLM in the workplace. If you don't realize that's going to happen you don't realize just how many people come to work shuffle proverbial paper from one side of their desk to another, making sure forms are filled out correctly or this month's TPS cover sheets are attached. If you're in programming or engineering you tend to think office jobs require a lot of context and invention. But most of them don't.

Ten years ago if you graduated with a law degree and went to work for one of the big NYC law firms, the first place they would put you is in a big room with dozens of other new lawyers doing discovery. You'd get a stack of documents to read, looking for relevance to the particular case that firm was working on. Those jobs are already gone, for the most part.

And didn't someone high up in Microsoft say 95% of the programming in the future will be done by LLMs? Do you expect Microsoft intends to keep the 95% of programmers it doesn't need anymore?

> As I am retired, I can assure you beyond any doubt I cannot be replaced by an LLM in the workplace.

That makes sense. Being removed from today’s workforce makes it easier to believe the narrative that AI is causing these shifts, when in fact they’re really the result of the end of historically low interest rates and other broader economic conditions.

> If you don't realize that's going to happen you don't realize just how many people come to work shuffle proverbial paper from one side of their desk to another, making sure forms are filled out correctly or this month's TPS cover sheets are attached

These people are not the entirety of the white collar market.

> And didn't someone high up in Microsoft say 95% of the programming in the future will be done by LLMs?

Moronic Microsoft executives have said a lot of things. It doesn't make them right.

> 'Microsoft's CEO Steve Ballmer scoffed at Apple's glass-and-metal gadget, telling USA Today that "there's no chance that the iPhone is going to get any significant market share. No chance."'

> Do you expect Microsoft intends to keep the 95% of programmers it doesn't need anymore?

I expect like most things that come out of their mouth, it's obvious bullshit because they are involved in peddling AI and would desparately love you to buy some.

> guess it heralded the end of white collar jobs.

There is still studies that show that graduating in 2007-2009 permantly stunted your career earnings. Those gradutes were very much screwed, statistically speaking.

Also there wasn't a narrative in 2008 about trying to automate your white collar work away. The powers that be do now want the white collar jobs to come back. It may not succeed, but it's going to create a sub-generation of people who completely lost faith in the country the reside in. That sort of population is dangerous, to say the least.

>If you actually believe this, it makes you one of the rare few who likely could actually be replaced with an LLM.

What's your interpretation then? LLM crashes and burns and Google just goes back to using humans again? Everything lives happily ever after?

> There is still studies that show that graduating in 2007-2009 permantly stunted your career earnings. Those gradutes were very much screwed, statistically speaking.

Sure, but that’s not the claim being made here, so it’s irrelevant. Also, my earnings are just fine. So again, seems like some made out ok, and others didnt. It certainly wasnt a permanent white collar apocalypse.

> Also there wasn't a narrative in 2008 about trying to automate your white collar work away.

That depends on where you were. Plenty of people I started with were automated out of their original roles and ended up in entirely different careers. In my industry, one person today can do what used to take ten.

> What's your interpretation then?

The same as it's always been. I'll live in reality and not a fantasty land sold by executives trying to sell me on their AI turd, or hide behind AI as a smokescreen for employment activities they were going to carry out either way.

> LLM crashes and burns and Google just goes back to using humans again?

It’ll settle into its place. Some jobs will get knocked out, the same way automation reshaped mine. But this idea that AI is about to replace the entire white collar workforce?

Come back to me when this stuff isn’t confidently spitting out nonsense half the time because until then as far as I can see, this stuff at best lets people do more, it doesnt magically replace a workforce.

LLMs don’t seem to be destroying jobs en masse—though they do seem to be eliminating some jobs in copywriting and graphic design —but they also don’t seem to be creating jobs the way previous levels of investment in technology have.

Reply on news.ycombinator.com