On a Saturday ten years ago, I went into the office to do something. I was surprised to see Sunil (not his real name), whose wife had delivered their first child the previous night.
"Oh hello, Sunil," said I, "what are you doing here? You should be home with your kid."
"I can't leave!" he said, absolutely frantic. "I have to finish this project!"
Of course, Sunil is an H-1B.
Now, maybe Sunil is just a workaholic with issues. But it was clear to me from our discussion that he legit believed that if he didn't finish THAT project on time, the company would fire him, and he and his family would be sent home.
This guy was good at his job, too. He wasn't on a performance plan or anything like that. It struck me that the H-1B program creates a kind of indentured servitude, and employers exploit that.
That’s a significant human problem with this system. He’s almost certainly better off than in his native country money wise but he really is being exploited!
Can you explain the difference between exploitation and motivation? If you framed this differently then I don't think that it would feel exploitative at all. Imagine that he was in India hoping to get a promotion which would take him to America, but there's only one spot available and he has to out-compete 20 other engineers. Would his decision to work late instead of going home to his newborn still feel exploitative? Or would he seem like a go-getter who's working hard to help his family? What if he was a student studying all night before the final? Is that exploitative?
Now, maybe Sunil is just a workaholic with issues. But it was clear to me from our discussion that he legit believed that if he didn't finish THAT project on time, the company would fire him, and he and his family would be sent home.
This guy was good at his job, too. He wasn't on a performance plan or anything like that. It struck me that the H-1B program creates a kind of indentured servitude, and employers exploit that.
Yes it's possible but he's a workaholic/go-getter but if this was a worker who was a citizen the employer would respect the person's personal life and not pressure him to work at a time in his life that literally happens just once!!! That's the exploitative part, you only have kids a few times in your life and that should be kind of sacred!
I absolutely support a high skilled visa auction. But with both a number cap and a salary floor.
If you say you want the best <whatever> on earth, you can afford to pay at least $300K for them.
And maybe it will make some companies fire their HR drones and reform their hiring process to actually find the best Americans for whatever job they are doing instead of the best at playing HR-keyword bingo.
Also, a better hiring process would incentivize universities and other training systems to actually teach people the thing they are purportedly being taught to do.
The problem is universities feel they need to award students with a degree even if those students aren't skilled enough for one. This is good for universities (increases revenue) and crap students, but bad for good ones.
Regarding whether H1B is about attracting skilled workers or exists to push down wages, I would suggest the following:
- companies that employ H1B workers have to pay extra taxes (e.g. 20% of those workers' wages) for importing and employing them; this means that employers are disincentivised from employing H1B workers to cut costs
- H1B workers are free to move to different jobs; this means that an employer can't treat them like shit as they'll simply leave.
“Reduce the size of the program and make companies pay full price for the privilege of its use, and we’ll go from there.”
Worth a try. What the commentary here does is to focus on a particular aspect/field of the H1-b VISA program, so I hate to recommend too many sweeping changes based upon such.
For example—to keep it focused upon CSc—my department had any number of *faculty* hired with such VISA’s from overseas. I can guarantee you, they were the best we could find in their subfields. Salary was never the issue and was the same for all new hired faculty anyway. It was always a matter of strengthening the department and our reputation in the field. In short, merit! This positive aspect of H1-b VISA’s of is what you’ve acknowledged in your commentary today.
On the other hand, horror stories such as with Disney abound. Disney back in the 80’s (90’s?) fired their entire IT staff and replaced them with H1-b VISA holders. Brazenly, they even had the audacity to incentivize their former (US) IT employee’s to remain on for 6 weeks to “train” their foreign replacements! That opened my eyes wrt H1-b VISA’s and their use in cost cutting to the detriment of American citizens. Such abuse must be eliminated.
workers as expense items and not part of the “core mission “ of their business. Disney needs to produce high quality films and theme park experiences. Every thing else doesn’t matter. Is that short sighted?? Maybe but that’s how many business view IT.
A factor that several of your posts have touched on but not really explored is what colleges are *supposed to be for* and whether they're the best format for teaching certain subjects and skillsets. An example I like to use is that I am frequently told that it is absolutely necessary to have a four-year degree (but really, a Master's!) to become a librarian, because there is really a lot of information you need to know to be a librarian. And yet it takes the Navy 18 months (22 if I'm counting Basic Training) to teach a high school kid how to operate a nuclear reactor. I feel pretty good assuming that there is more information in operating a nuclear reactor than in being a librarian.
I think college is good for teaching certain skillsets and knowledge but when made widespread tends to revert to theory-based concepts and are less good at practical knowledge and application, at least outside of certain disciplines. Also, related to "College English Majors Can't Read": college as a filter now is only the most coarse grain imaginable and anyway colleges have zero incentive to actually act as filters and not pumps.
I have no solution, it's turtles all the way down, I fear!
Personally, I learned assembly language in my CS studies, and it was our primary language in school. This is because I am an Old Person and that was the curriculum then. After a few years of working in assembler, pointers will never be a mystery and learning C was straightforward.
Iam not particularly smart, either, but have succeeded in making a living in software for 50 years.
In summary - your success in coding depends on how you were taught.
Two important things I think you missed: first, immigration doesn't only suppress wages via supply and demand. The semi-indentured nature of H1B harms their negotiating position, as someone else said, but there's more: legal status in the US is worth tens of thousands of dollars per year or more to some immigrants. H1B visaholders have been getting work-anywhere permits for their spouse via H4 EAD for 10 years. Their children become/became citizens automatically depending on what the courts say. And either they or their children can sponsor future family chain migration, something that can be worth a lot culturally depending on where they came from. All this is worth money, and the company hiring them knows it. This is probably less of a factor in the higher tiers.
Second, if the pool of H1B immigrants disproportionately comes from cultures that have less taboos on lying and self-promotion compared to US SWEs, this will massively muck up the hiring pipeline. The hapless HR lady is going to see page after page of perfect keyword matching, bragging, and questionable master's degrees from people who are willing to say anything to sell themselves, then throw the honest ones in the trash. In some cases, like having to leave the country with your family if you don't get a job in 90 days, I can't say I blame the visaholders.
>All this is worth money, and the company hiring them knows it.
So what's wrong with that? The visa worker is getting significant non-monetary benefits from being here. Why shouldn't those benefits be part of the negotiation? It seems economically inefficient for them not to be.
You are hiring H1B people from peasant cultures. They would do anything for a buck or a shot at the “American dream”. Americans actually want to be treated like human beings and most of them want to have a work life balance. Americans cannot compete with this mindset in general. We have higher expectations and want to be able to leave an employer if we’re treated like trash. H1Bs know this is their only shot so they are willing to eat that shxx sandwich all day long.
AI will replace or lower the skill level required for many programming and networking jobs, especially at the entry to intermediate level. The result will be hiring at reduced salaries and staffing cuts. This is especially true of any web-related programming. The types of people who are needed in IT are those with advanced math and engineering skills focused on AI, who will eventually put themselves out of work through further enhancements to AI. My advice is to enter a blue collar construction-related field, which (for now) can't be automated.
My nightmare AI scenario is that it will be used to replace juniors. Whether it can perform in that role is an orthogonal question in the short…medium run. It’s even almost irrelevant whether it’s actually used to replace juniors or whether that is just the vibe/meme.
You’ll never have more engineers that started in 2024 than you have now, and gaps in onboarding create non-contiguous cultures of practice. Cf. how hard of a time the Germans will have restarting nuclear if they ever come to their senses since nobody in their right mind over there has studied for that field in the last at least 10 years, and probably 30, the way the mood shift was going. It’s not easy to kickstart a disrupted domestic talent pipeline.
H-1B is basically the same effect as AIs here, or Greens. Each of the substance or the suspicion of being replaced/obsoleted alone is enough to make the career seem like a dodgy, and low-status, prospect.
And yeah, most candidates are terrible and so is almost all HR. A huge haystack, few needles, and a gaggle of farmhands who seem to think their job is loading a wagon with hay. It’s so bad that for all the impracticality, sometimes there’s a real benefit of “fringe” languages that have a smart ingroup like Rust or HS/F# or OCaml, or the CL diehards, etc. It is less the tech itself (no matter how good in principle – BYO library ecosystem will rarely be the trade-off of choice) than that the hiring haystack shrinks much faster than the amount of needles. They’re actually often the very same needles.
(My dream AI scenario is that AI keeps getting more productive but never becomes reliable. Basically commoditised complement for programmers, a co-worker with encyclopedic knowledge, superhuman speed, and dodgy epistemic habits that needs competent supervision.
Those AI scenarios are obviously compatible, at least in the “apres moi…” mode.)
Er no. We'll see fewer entry level jobs but not zero. But it will not reduce the demand for higher level ones so much as make the higher level ones more efficient.
I didn’t say zero entry level jobs. There will still be some. But they will no longer require guru-like levels of knowledge of JAVA, C++, Perl, C#, and so on. The new jobs will be semi-skilled computer technicians who are good users of AI.
Programming is expensive and error prone, especially when updating software. AI will make this process far less dramatic than it can be. It will be able to automatically update drivers or roll back software versions. Something which now can produce significant amounts of down time.
I don’t think the rise of AI is a good thing. But there is nothing standing in the way of its use from a business standpoint. It’s best to prepare for the inevitable. The IT workplace has remined somewhat the same since the 1990s. It’s about to change for good.
Yeah I think we're in some agreement. Though I expect we're going to find out the painful way that allowing an AI to make unsupervised changes to things is dangerous. The problem is AI is really bad about hallucinating when it doesn't know the answer
And those working with low level languages like C were paid well because they were hard to do well. When people can’t get the hang of pointers that was an automatic weeding out of developers who did not have the intellectual chops work with those languages . But it’s very good that more productive languages like Python are now dominant. AI already seems to be another boost for people with a more general skill set.
Or the revolution man. The over produced elite are the real danger and the right wing knows it . Know your second amendment rights and learn how to aim. Freedom is only won at gunpoint baby 😳
I think talented people are by definition rare and on the higher bump of the bimodal curve. They are literally irreplaceable . They will dictate their own terms if companies know what is good for them!
Some will be able to dictate their own terms. But these will be those who can program AI at an advanced level and also work with math models at a high level. I can program in Python or write bash scripts at a basic level. When confronted with JAVA programs dealing with Markov chains, I glaze over. But those who are comfortable with these (and similar) types of probability programs will do well.
It has been pointed out that the US gained its technological leadership when total immigration was around 30,000 per year. Gross. Net migration was negative.
If you are not getting the 0.01% and only the 0.01% , you are just importing people to get in the way of that 0.01%.
My story: I was hired off the street. Literally: Off. The. Street. Just walking along the pavement. I had most of a bachelor's in mathematics, mainly abstract algebra. I was given a six-week training course. And I was paid a full salary while doing it. I cannot conceive of that happening today.
There is no shortage of programmers. Not unless they're hiring people off the street, and paying them while they train.
In 2000, I saw job ads for Java developers saying "must have 6+ years' experience with Java". Even James Gosling didn't qualify. HR is a blight upon civilisation. Want to know where productivity has gone? Look in HR.
>HR is a blight upon civilisation. Want to know where productivity has gone? Look in HR.
Well said. The main thing that makes companies afraid to hire are the legal consequences of firing anyone. If employment was truly at-will with zero liability risk then "getting your foot in the door" would be trivial. De-risk the hiring process for companies: no employment protections should exist *at all* for a yearlong probationary period.
Maybe part of it is that more American companies like Microsoft are becoming H1B body shops, and the experienced native programmers they are letting go cannot get jobs at FAANG companies who want the youngest and 'brightest'.
The hiring process has been broken a long time, it’s miraculous that the industry has done as well as it has in the west. IBM proved by example before I was born that LOC is a terrible metric for programmer productivity, and yet that’s still a used standard.
In my second year at uni, 3 professors in different courses began by telling us more than half of the previous cohort had been *caught* cheating. I wonder what the uncaught ratio was.
I did not complete my CS degree, for a few reasons, but I was only attending at all because I had some small scholarships. I have never been paid to program, but I’ve continued writing some small projects on the side. I studied hardware design (my university was at the time the only one in north american to REQUIRE it for all CS students), and we all had to design an 8bit CPU in vhdl. Great experience, and I wish I could’ve done something with what I learned, and kept learning.
The industry rejected a lot of people who would’ve and could’ve done the work, and the education system overall is broken, and now I do day labour in construction. I tried to get trade work too after university, but they’re screwing us around too.
Oh well. The elites have shorted the society, and now the outlier intelligent people have little reason, not even personal greed, to prop up the system anymore, because there’s no carrot.
Canada is in a worse situation than the USA for immigration troubles, but if recent events are any indication we’ll be in *all* the exact same problems combined, together, and I hope I can scrape enough cash to get out by then.
> Basically: lower the cap and make it an auction instead of a lottery. We wouldn’t have all this talk about depressed wages or true economic need if companies had to compete with one another to hire the talent they say they need. Reduce the cap by at least half to start, and auction each visa off to the highest bidder.
True for tech - even people like me who work for FAANG accept that the folks moving through consultancies are a different tier compared to big tech. However, you might also be inflicting damage on fields that require the talent but cannot afford to pay tech companies' compensation. I know an electrical engineer in chip design, a chemical engineer designing silicon wafers etc, someone in biomedical engineering etc - talented people in their fields but miles away from what FAANG would pay if they were a software developer. Maybe - defining the fields and mandating they're paid top of their field compared to the relevant years of experience * might * be one way to address this but your proposal would ensure majority of the H1-Bs go to big tech software developers.
No. Market mechanisms, while imperfect, are less imperfect and much simpler than anything else. If FAANG is paying more for workers then that means FAANG is getting more value out of them. It's therefore socially efficient to let them hoard the H1-Bs. If the chip designer is really more valuable then let his employer signal that by paying for it. If they can't then he's not more valuable.
This idea that “global leadership” is a thing really needs to die. It’s a “crops rotting in the fields” argument. There are no crops rotting in the fields. It’s OK if jobs destined for an H1B in the US go to somebody in Germany or Canada or India. There is no such thing as global leader in technology. It’s as fake as employee of the month or best good boy in the world. It’s not something that a serious person should worry about. Yes, the jobs will probably go somewhere else. That’s OK. More for another country is not less for the US. That’s not how economics works. The pie is always expanding and your share of that pie is not determined by how large your country is or how many ringers you bring onto your team. The shortages are not in talent but in coordination. Societies that coordinate grow and societies that fail to solve the coordination problem don’t grow. Coordination is the limitation, not talent. If your society has (for the moment) solved the coordination problem, then don’t screw it up by making that problem harder for future generations. (This is posted from my blog, that's why it sounds like I'm shouting).
Non-linear benefits to who? I'm not in Sergey Brin's will. And are there no negative externalities? As I argue, the supposed benefits of immigration are small beans. Most Americans don't want growth, that's why they block new factories, vote for more regulation, and want to tax wealth. Being pro-growth is more important than immigration when it comes to contributing to growth (immigration is actually negative but that's what's in dispute).
To talented people and wider society alike. Having tens of thousands of software engineers in the bay area makes it easier to start a new company, cross-pollinate ideas, etc. Plus tax revenue stays where the people are.
Do you know how many natives have moved away from the Bay? Practically all of them. Golf clap. Nice work. You have a city where everyone arrived from another country and at max, 25 years ago.
The other major complication here is that no matter what the H1-B rules are, companies can also just hire foreign nationals to work in their own countries, either through contractors or directly through establishing overseas branches. I work with a company that has done both, with (sincerely) a wide range of success--ranging from duds who wouldn't even bother to give ChatGPT the right requirements, to some absolute rockstars. This obviously isn't directly comparable, since for example you can't usually hire foreign nationals for government contracts, but it's very much a factor at play in the private sector side.
Overall, though, every American national I know who's actually good at coding has ended up with well-paid and stable work in the long term--there's such a demand for competence that as long as you can demonstrate that, and have minimal people skills, someone eventually will find you and pay you.
When I was coming up, the panic at the time was about offshoring, everyone was going to lose their jobs to it.
This both happened and didn't, companies would like to do it more but it just doesn't work very well at scale. There is a reason that Google Hyderabad will forever be a tiny fraction of Google mountain view in number of employees.
Yeah, in my experience gains from paying offshore salaries get heavily offset by timezone differences, cultural/linguistic frictions, offshore HR/finance overhead, infrastructure complications....
Our US headcount is only very slightly down from peak, but its share of engineering is only 50% now, as all growth and most backfills have been overseas. Stasis/slow bleed is obviously preferable to unemployment but it’s a big downgrade from 2010s hypergrowth.
I think we’re broadly in line with “elite tech” trends on this.
One fun effect here is that they’re flexing their labor market power to demand more and more office attendance, but getting projects done now requires a lot of IST/PST Zoom meetings during prime commuting hours.
I have a story.
On a Saturday ten years ago, I went into the office to do something. I was surprised to see Sunil (not his real name), whose wife had delivered their first child the previous night.
"Oh hello, Sunil," said I, "what are you doing here? You should be home with your kid."
"I can't leave!" he said, absolutely frantic. "I have to finish this project!"
Of course, Sunil is an H-1B.
Now, maybe Sunil is just a workaholic with issues. But it was clear to me from our discussion that he legit believed that if he didn't finish THAT project on time, the company would fire him, and he and his family would be sent home.
This guy was good at his job, too. He wasn't on a performance plan or anything like that. It struck me that the H-1B program creates a kind of indentured servitude, and employers exploit that.
That’s a significant human problem with this system. He’s almost certainly better off than in his native country money wise but he really is being exploited!
Can you explain the difference between exploitation and motivation? If you framed this differently then I don't think that it would feel exploitative at all. Imagine that he was in India hoping to get a promotion which would take him to America, but there's only one spot available and he has to out-compete 20 other engineers. Would his decision to work late instead of going home to his newborn still feel exploitative? Or would he seem like a go-getter who's working hard to help his family? What if he was a student studying all night before the final? Is that exploitative?
I think the original poster says it pretty well:
Now, maybe Sunil is just a workaholic with issues. But it was clear to me from our discussion that he legit believed that if he didn't finish THAT project on time, the company would fire him, and he and his family would be sent home.
This guy was good at his job, too. He wasn't on a performance plan or anything like that. It struck me that the H-1B program creates a kind of indentured servitude, and employers exploit that.
Yes it's possible but he's a workaholic/go-getter but if this was a worker who was a citizen the employer would respect the person's personal life and not pressure him to work at a time in his life that literally happens just once!!! That's the exploitative part, you only have kids a few times in your life and that should be kind of sacred!
So no, you can't explain it.
I absolutely support a high skilled visa auction. But with both a number cap and a salary floor.
If you say you want the best <whatever> on earth, you can afford to pay at least $300K for them.
And maybe it will make some companies fire their HR drones and reform their hiring process to actually find the best Americans for whatever job they are doing instead of the best at playing HR-keyword bingo.
Also, a better hiring process would incentivize universities and other training systems to actually teach people the thing they are purportedly being taught to do.
The problem is universities feel they need to award students with a degree even if those students aren't skilled enough for one. This is good for universities (increases revenue) and crap students, but bad for good ones.
Regarding whether H1B is about attracting skilled workers or exists to push down wages, I would suggest the following:
- companies that employ H1B workers have to pay extra taxes (e.g. 20% of those workers' wages) for importing and employing them; this means that employers are disincentivised from employing H1B workers to cut costs
- H1B workers are free to move to different jobs; this means that an employer can't treat them like shit as they'll simply leave.
We live in an age of remote work. There’s no need for Indians to actually be in our country.
“Reduce the size of the program and make companies pay full price for the privilege of its use, and we’ll go from there.”
Worth a try. What the commentary here does is to focus on a particular aspect/field of the H1-b VISA program, so I hate to recommend too many sweeping changes based upon such.
For example—to keep it focused upon CSc—my department had any number of *faculty* hired with such VISA’s from overseas. I can guarantee you, they were the best we could find in their subfields. Salary was never the issue and was the same for all new hired faculty anyway. It was always a matter of strengthening the department and our reputation in the field. In short, merit! This positive aspect of H1-b VISA’s of is what you’ve acknowledged in your commentary today.
On the other hand, horror stories such as with Disney abound. Disney back in the 80’s (90’s?) fired their entire IT staff and replaced them with H1-b VISA holders. Brazenly, they even had the audacity to incentivize their former (US) IT employee’s to remain on for 6 weeks to “train” their foreign replacements! That opened my eyes wrt H1-b VISA’s and their use in cost cutting to the detriment of American citizens. Such abuse must be eliminated.
But how?
The problem is that these organizations see these
workers as expense items and not part of the “core mission “ of their business. Disney needs to produce high quality films and theme park experiences. Every thing else doesn’t matter. Is that short sighted?? Maybe but that’s how many business view IT.
A factor that several of your posts have touched on but not really explored is what colleges are *supposed to be for* and whether they're the best format for teaching certain subjects and skillsets. An example I like to use is that I am frequently told that it is absolutely necessary to have a four-year degree (but really, a Master's!) to become a librarian, because there is really a lot of information you need to know to be a librarian. And yet it takes the Navy 18 months (22 if I'm counting Basic Training) to teach a high school kid how to operate a nuclear reactor. I feel pretty good assuming that there is more information in operating a nuclear reactor than in being a librarian.
I think college is good for teaching certain skillsets and knowledge but when made widespread tends to revert to theory-based concepts and are less good at practical knowledge and application, at least outside of certain disciplines. Also, related to "College English Majors Can't Read": college as a filter now is only the most coarse grain imaginable and anyway colleges have zero incentive to actually act as filters and not pumps.
I have no solution, it's turtles all the way down, I fear!
Personally, I learned assembly language in my CS studies, and it was our primary language in school. This is because I am an Old Person and that was the curriculum then. After a few years of working in assembler, pointers will never be a mystery and learning C was straightforward.
Iam not particularly smart, either, but have succeeded in making a living in software for 50 years.
In summary - your success in coding depends on how you were taught.
Two important things I think you missed: first, immigration doesn't only suppress wages via supply and demand. The semi-indentured nature of H1B harms their negotiating position, as someone else said, but there's more: legal status in the US is worth tens of thousands of dollars per year or more to some immigrants. H1B visaholders have been getting work-anywhere permits for their spouse via H4 EAD for 10 years. Their children become/became citizens automatically depending on what the courts say. And either they or their children can sponsor future family chain migration, something that can be worth a lot culturally depending on where they came from. All this is worth money, and the company hiring them knows it. This is probably less of a factor in the higher tiers.
Second, if the pool of H1B immigrants disproportionately comes from cultures that have less taboos on lying and self-promotion compared to US SWEs, this will massively muck up the hiring pipeline. The hapless HR lady is going to see page after page of perfect keyword matching, bragging, and questionable master's degrees from people who are willing to say anything to sell themselves, then throw the honest ones in the trash. In some cases, like having to leave the country with your family if you don't get a job in 90 days, I can't say I blame the visaholders.
>All this is worth money, and the company hiring them knows it.
So what's wrong with that? The visa worker is getting significant non-monetary benefits from being here. Why shouldn't those benefits be part of the negotiation? It seems economically inefficient for them not to be.
You are hiring H1B people from peasant cultures. They would do anything for a buck or a shot at the “American dream”. Americans actually want to be treated like human beings and most of them want to have a work life balance. Americans cannot compete with this mindset in general. We have higher expectations and want to be able to leave an employer if we’re treated like trash. H1Bs know this is their only shot so they are willing to eat that shxx sandwich all day long.
AI will replace or lower the skill level required for many programming and networking jobs, especially at the entry to intermediate level. The result will be hiring at reduced salaries and staffing cuts. This is especially true of any web-related programming. The types of people who are needed in IT are those with advanced math and engineering skills focused on AI, who will eventually put themselves out of work through further enhancements to AI. My advice is to enter a blue collar construction-related field, which (for now) can't be automated.
I doubt it personally.
My nightmare AI scenario is that it will be used to replace juniors. Whether it can perform in that role is an orthogonal question in the short…medium run. It’s even almost irrelevant whether it’s actually used to replace juniors or whether that is just the vibe/meme.
You’ll never have more engineers that started in 2024 than you have now, and gaps in onboarding create non-contiguous cultures of practice. Cf. how hard of a time the Germans will have restarting nuclear if they ever come to their senses since nobody in their right mind over there has studied for that field in the last at least 10 years, and probably 30, the way the mood shift was going. It’s not easy to kickstart a disrupted domestic talent pipeline.
H-1B is basically the same effect as AIs here, or Greens. Each of the substance or the suspicion of being replaced/obsoleted alone is enough to make the career seem like a dodgy, and low-status, prospect.
And yeah, most candidates are terrible and so is almost all HR. A huge haystack, few needles, and a gaggle of farmhands who seem to think their job is loading a wagon with hay. It’s so bad that for all the impracticality, sometimes there’s a real benefit of “fringe” languages that have a smart ingroup like Rust or HS/F# or OCaml, or the CL diehards, etc. It is less the tech itself (no matter how good in principle – BYO library ecosystem will rarely be the trade-off of choice) than that the hiring haystack shrinks much faster than the amount of needles. They’re actually often the very same needles.
(My dream AI scenario is that AI keeps getting more productive but never becomes reliable. Basically commoditised complement for programmers, a co-worker with encyclopedic knowledge, superhuman speed, and dodgy epistemic habits that needs competent supervision.
Those AI scenarios are obviously compatible, at least in the “apres moi…” mode.)
Er no. We'll see fewer entry level jobs but not zero. But it will not reduce the demand for higher level ones so much as make the higher level ones more efficient.
AI cannot program as well as it's proponants claim - see https://xcancel.com/esrtweet/status/1927559794514186664 and for that matter the academic paper behind https://arstechnica.com/ai/2025/08/researchers-find-llms-are-bad-at-logical-inference-good-at-fluent-nonsense/
I didn’t say zero entry level jobs. There will still be some. But they will no longer require guru-like levels of knowledge of JAVA, C++, Perl, C#, and so on. The new jobs will be semi-skilled computer technicians who are good users of AI.
Programming is expensive and error prone, especially when updating software. AI will make this process far less dramatic than it can be. It will be able to automatically update drivers or roll back software versions. Something which now can produce significant amounts of down time.
I don’t think the rise of AI is a good thing. But there is nothing standing in the way of its use from a business standpoint. It’s best to prepare for the inevitable. The IT workplace has remined somewhat the same since the 1990s. It’s about to change for good.
> [AI] will be able to automatically update drivers or roll back software versions.
And make a fucking pigs ear of it from time to time, I'm sure.
Yeah I think we're in some agreement. Though I expect we're going to find out the painful way that allowing an AI to make unsupervised changes to things is dangerous. The problem is AI is really bad about hallucinating when it doesn't know the answer
> I expect we're going to find out the painful way that allowing an AI to make unsupervised changes to things is dangerous.
I expect at least one big company will go down because of this
And those working with low level languages like C were paid well because they were hard to do well. When people can’t get the hang of pointers that was an automatic weeding out of developers who did not have the intellectual chops work with those languages . But it’s very good that more productive languages like Python are now dominant. AI already seems to be another boost for people with a more general skill set.
Soon many jobs will be automated, then most of them. At which point its either UBI or most humans starve to death.
Or the revolution man. The over produced elite are the real danger and the right wing knows it . Know your second amendment rights and learn how to aim. Freedom is only won at gunpoint baby 😳
I think talented people are by definition rare and on the higher bump of the bimodal curve. They are literally irreplaceable . They will dictate their own terms if companies know what is good for them!
Some will be able to dictate their own terms. But these will be those who can program AI at an advanced level and also work with math models at a high level. I can program in Python or write bash scripts at a basic level. When confronted with JAVA programs dealing with Markov chains, I glaze over. But those who are comfortable with these (and similar) types of probability programs will do well.
It has been pointed out that the US gained its technological leadership when total immigration was around 30,000 per year. Gross. Net migration was negative.
If you are not getting the 0.01% and only the 0.01% , you are just importing people to get in the way of that 0.01%.
My story: I was hired off the street. Literally: Off. The. Street. Just walking along the pavement. I had most of a bachelor's in mathematics, mainly abstract algebra. I was given a six-week training course. And I was paid a full salary while doing it. I cannot conceive of that happening today.
There is no shortage of programmers. Not unless they're hiring people off the street, and paying them while they train.
In 2000, I saw job ads for Java developers saying "must have 6+ years' experience with Java". Even James Gosling didn't qualify. HR is a blight upon civilisation. Want to know where productivity has gone? Look in HR.
>HR is a blight upon civilisation. Want to know where productivity has gone? Look in HR.
Well said. The main thing that makes companies afraid to hire are the legal consequences of firing anyone. If employment was truly at-will with zero liability risk then "getting your foot in the door" would be trivial. De-risk the hiring process for companies: no employment protections should exist *at all* for a yearlong probationary period.
Maybe part of it is that more American companies like Microsoft are becoming H1B body shops, and the experienced native programmers they are letting go cannot get jobs at FAANG companies who want the youngest and 'brightest'.
The hiring process has been broken a long time, it’s miraculous that the industry has done as well as it has in the west. IBM proved by example before I was born that LOC is a terrible metric for programmer productivity, and yet that’s still a used standard.
In my second year at uni, 3 professors in different courses began by telling us more than half of the previous cohort had been *caught* cheating. I wonder what the uncaught ratio was.
I did not complete my CS degree, for a few reasons, but I was only attending at all because I had some small scholarships. I have never been paid to program, but I’ve continued writing some small projects on the side. I studied hardware design (my university was at the time the only one in north american to REQUIRE it for all CS students), and we all had to design an 8bit CPU in vhdl. Great experience, and I wish I could’ve done something with what I learned, and kept learning.
The industry rejected a lot of people who would’ve and could’ve done the work, and the education system overall is broken, and now I do day labour in construction. I tried to get trade work too after university, but they’re screwing us around too.
Oh well. The elites have shorted the society, and now the outlier intelligent people have little reason, not even personal greed, to prop up the system anymore, because there’s no carrot.
Canada is in a worse situation than the USA for immigration troubles, but if recent events are any indication we’ll be in *all* the exact same problems combined, together, and I hope I can scrape enough cash to get out by then.
Let them all keep coming! I’ll leave!
> Basically: lower the cap and make it an auction instead of a lottery. We wouldn’t have all this talk about depressed wages or true economic need if companies had to compete with one another to hire the talent they say they need. Reduce the cap by at least half to start, and auction each visa off to the highest bidder.
True for tech - even people like me who work for FAANG accept that the folks moving through consultancies are a different tier compared to big tech. However, you might also be inflicting damage on fields that require the talent but cannot afford to pay tech companies' compensation. I know an electrical engineer in chip design, a chemical engineer designing silicon wafers etc, someone in biomedical engineering etc - talented people in their fields but miles away from what FAANG would pay if they were a software developer. Maybe - defining the fields and mandating they're paid top of their field compared to the relevant years of experience * might * be one way to address this but your proposal would ensure majority of the H1-Bs go to big tech software developers.
No. Market mechanisms, while imperfect, are less imperfect and much simpler than anything else. If FAANG is paying more for workers then that means FAANG is getting more value out of them. It's therefore socially efficient to let them hoard the H1-Bs. If the chip designer is really more valuable then let his employer signal that by paying for it. If they can't then he's not more valuable.
The 80s called, it wants its economic theory back.
This idea that “global leadership” is a thing really needs to die. It’s a “crops rotting in the fields” argument. There are no crops rotting in the fields. It’s OK if jobs destined for an H1B in the US go to somebody in Germany or Canada or India. There is no such thing as global leader in technology. It’s as fake as employee of the month or best good boy in the world. It’s not something that a serious person should worry about. Yes, the jobs will probably go somewhere else. That’s OK. More for another country is not less for the US. That’s not how economics works. The pie is always expanding and your share of that pie is not determined by how large your country is or how many ringers you bring onto your team. The shortages are not in talent but in coordination. Societies that coordinate grow and societies that fail to solve the coordination problem don’t grow. Coordination is the limitation, not talent. If your society has (for the moment) solved the coordination problem, then don’t screw it up by making that problem harder for future generations. (This is posted from my blog, that's why it sounds like I'm shouting).
There are nonlinear benefits and positive externalities to talent accumulation. It's better for the US if those are captured by US markets.
Non-linear benefits to who? I'm not in Sergey Brin's will. And are there no negative externalities? As I argue, the supposed benefits of immigration are small beans. Most Americans don't want growth, that's why they block new factories, vote for more regulation, and want to tax wealth. Being pro-growth is more important than immigration when it comes to contributing to growth (immigration is actually negative but that's what's in dispute).
>Non-linear benefits to who?
To talented people and wider society alike. Having tens of thousands of software engineers in the bay area makes it easier to start a new company, cross-pollinate ideas, etc. Plus tax revenue stays where the people are.
Do you know how many natives have moved away from the Bay? Practically all of them. Golf clap. Nice work. You have a city where everyone arrived from another country and at max, 25 years ago.
The other major complication here is that no matter what the H1-B rules are, companies can also just hire foreign nationals to work in their own countries, either through contractors or directly through establishing overseas branches. I work with a company that has done both, with (sincerely) a wide range of success--ranging from duds who wouldn't even bother to give ChatGPT the right requirements, to some absolute rockstars. This obviously isn't directly comparable, since for example you can't usually hire foreign nationals for government contracts, but it's very much a factor at play in the private sector side.
Overall, though, every American national I know who's actually good at coding has ended up with well-paid and stable work in the long term--there's such a demand for competence that as long as you can demonstrate that, and have minimal people skills, someone eventually will find you and pay you.
When I was coming up, the panic at the time was about offshoring, everyone was going to lose their jobs to it.
This both happened and didn't, companies would like to do it more but it just doesn't work very well at scale. There is a reason that Google Hyderabad will forever be a tiny fraction of Google mountain view in number of employees.
Yeah, in my experience gains from paying offshore salaries get heavily offset by timezone differences, cultural/linguistic frictions, offshore HR/finance overhead, infrastructure complications....
Our US headcount is only very slightly down from peak, but its share of engineering is only 50% now, as all growth and most backfills have been overseas. Stasis/slow bleed is obviously preferable to unemployment but it’s a big downgrade from 2010s hypergrowth.
I think we’re broadly in line with “elite tech” trends on this.
One fun effect here is that they’re flexing their labor market power to demand more and more office attendance, but getting projects done now requires a lot of IST/PST Zoom meetings during prime commuting hours.