Hacker News Reader

With respect to hiring, of the two points you mention - AI productivity gains for SWE and "no more free money" - I'm confident it's solely the latter, that higher interest rates cut off free money from startups, so they need to be more careful about who they hire. Additionally, there was substantial overhiring during COVID which combined with the interest rate changes lead to a lot of layoffs, and subsequently lead to the conservative hiring in a lot of companies. This leads to a lot more competition than there was previously, both to even get an interview in the first place and to get to the end with an offer.

It's possible that AI may have some kind of impact on the industry too with respect to productivity, but I'd bet that it's nowhere near a 10-20% gain and it's not a factor at all in the difficulty of finding a job.

I agree. Working as a manager in the industry my take is that AI has had a near-zero actual increase in SE productivity. That’s not to say the C-level perception matches that actual number though. It hasn’t happened at my company but I wouldn’t be surprised to learn that some layoffs from big companies were driven by that perception. I would interpret more as an excuse to get rid of low performers in general.

I think LLMs spitting out code will probably supercharge the "if engineering showed a toy demo, that means we're 90% done and can start taking pre-orders" problem.

Yup - while creating massive tech debt, and weird edge case bugs that are even less sensical than typical ones you’d get.

From what I've seen, AI is used as an answer to the question: "how are we going to be expected to maintain the same velocity with fewer people."

I second this explanation. In fact its not just startups but also a lot of big tech companies were rationally taking advantage of the low interest rates to pursue speculative bets/product features. Tech CEOs were being richly rewarded via inflating stock prices for chasing the potential for future cashflows.

When the macro changed, those bets no longer made rational sense.

The silver lining for guys like Jon Bach (as described in the intro) is that with the abundance of free agents and SF/SV office space, it's probably the a great time to form/find an ambitious team and help try to bootstrap the next Google.

> It's possible that AI may have some kind of impact on the industry too with respect to productivity, but I'd bet that it's nowhere near a 10-20% gain and it's not a factor at all in the difficulty of finding a job.

I'm not going to make any claim on the actual gain in productivity. My sample size of one for what was sometimes quite repetitive work is a very bad sample.

However, regardless of all other factors (let alone combined with them), I would say that AI has made getting a job far more difficult as AI gives people far more ability to shotgun every single job ad out there. I recently spoke to a recruiter who indicated that for a Senior Golang Dev role in Australia, he had received 400 applications. There is no-where near 400 spare senior Golang devs in this city.

I seem to remember a story that there was a lot of “defensive hiring” — you would hire a lot of technical people to keep them from starting or joining a disruptive startup. This would be a corollary of your higher interest rates pushing down employment by not allowing funding to flow to startups, but the end of such defensive hiring would serve to amplify the drop in hiring beyond just startup personnel.

Surely the section 174 changes in Trump's tax bill have to be a part of this as well. Since engineer salaries now have to be deducted as an expense amortized over 5 years rather than in the same year they were incurred, smaller companies with revenue that may not be profitable will still be taxed as if they are.

https://www.bdo.com/insights/tax/current-state-of-r-d-tax-po...

It's not an unreasonable proposition, but it doesn't explain why I'm seeing the exact same hiring trends in Australia, even before the USA election.

Those changes were made during Trump’s first term.

The use of Copilot and similar tools has been explicitly banned where I worked - we tested it out and the gains were minimal if any, and the possible cost of liability and licensing considered not worth it. There is no short-term intent to re-assess this either.

I was recently laid off with "Efficiencies gained by AI" explicitly stated as one of the reasons.

Either it's complete PR BS (most likely), or C-level execs are putting the cart before the horse - preparing for "productivity gains" that haven't actually materialized.

I have found Copilot to be useful enough to use if someone else pays for it. In some cases it was useful (writing Golang functions that perform basic SQL CRUD functions). Others times, not so much.

As I said above, I'm not claiming it's a massive increase, only that it is there. If AI moved the productivity needle 2%, and a given market has 100,000 software engineers, then that's possibly 2000 devs that are spamming the heck out of every job out there.

> Either it's complete PR BS (most likely), or C-level execs are putting the cart before the horse - preparing for "productivity gains" that haven't actually materialized.

All of the above is also a real possibility.

Honestly, I suspect the industry I'm in (System Software/Driver Development) is a poor fit for the current state of "AI Assistants" - there's too much specific that's just not got the depth of training data to "learn" from, it's a mature system so not much "boilerplate" needs to be generated, and much of the real complexity is managing the state of the hardware interface and expectations from a constantly-changing documented-mostly-in-some-hardware-engineer's-head-only specification. Or time debugging why the stated specification doesn't actually match the behavior of the hardware.

I don't think anyone in our group who was asked to trial it had much positive to say about it in that situation.

> My take for SE roles is that it's a combination of AI giving enough gains[0] to reduce head count requirements

It is not giving enough gains. It is only giving ammunition to board members and execs to cut headcount.

What is really happening is that the current cohort of tech companies has been living in the era of free money for so long that they have forgotten competent practices. They only thing they learned was more headcount==more revenue. This was partially possible by delivering large number of features and charging for each one of them e.g. AWS has a ton of useless crap services.

Of course that is horseshit logic that only worked in the era of free money. And of course market caps going up, up, and away was fantasy. At some point, markets shift and cannot be tapped any further. And that is what is causing headcount reduction - there are no saleable features to produce.

There is still a large amount of maintenance to do. But companies loathe paying for maintenance. They'd rather have customers suffer and quit than pay up front for headcount to maintain what has been built in the last decade. They would much rather scam customers with poor billing practices than provide quality and support.

Thus, we end up with lower headcount, lower quality, and a generally third world quality existence - all because execs want to protect margins over anything else aka shareholder primacy.

Everyone has been trying to get into tech for the last 4 years.

My wife works in healthcare. There are whole facebook groups dedicated to healthcare workers trying to transition into tech. A lot of these people are clinicians (Occupational Therapists, Physical Therapists, Speech Pathologists, etc).

There are a lot of people looking for WFH jobs and tech seems to be where they turn first. Honestly, I blame the glut of low effort "Day in the life of a Software Engineer" youtube videos where the "worker" spends most of the video walking their dog and going for a casual afternoon jog while fast forwarding through the actual work. There has been a lot of glamorizing of tech jobs on social media that doesn't typically really match reality.

Obviously, we all know about the trend of bootcamps promising six figure jobs after 2 months (though that seems to have tapered off a bit).

I've read a lot of resumes over the last 4 years and I feel like I've seen it all. I've had multiple senior developer openings and I've seen resumes with just about every background you can think of. Waiters, fast food workers, mechanics, pest control, construction workers. No actual real word programming experience, only a list of a few bootcamp group projects with a link to a github profile where they pushed their bootcamp homework. It was especially bad if you included React in the job description at all (I noticed a dramatic reduction in these types of resumes when the job description focused on backend work).

It reminds of that anecdote about stock bubbles: when your taxi driver is giving you stock tips then you know its time to sell. Well, when your waiter is talking to you about JavaScript because you wore an AWS shirt to dinner, you know that we're in a tech hiring bubble that is going to pop.

There are/were a lot of people looking for tech jobs that frankly have no business doing so. Some of them may have spent 5 figures on a bootcamp of dubious value.

Doesn't match reality at all. If only these people can see the daily insanity. I just got through 2 weeks, where at the end the owner of the company had to ask me why I haven't had much progress and trying to explain that everyday for the last two weeks he has shifted the requirements so much, anything I wrote had to get thrown out the window. Literally, get updated requirements, get 75% of the way through implementation. Bam, email in my inbox telling me to forget what was last said and do something completely different.

The best way to deal with such insanity is to slow down: get requirements, "design" for a few days, if requirements change "design" a few more days etc. Furious coding just burns you out, and not much progress is made in any case.

Yeah these people graduating from boot camps are super naive from my experience. Every single project is a ruby on rails twitter clone. Yawn. What happened to doing something novel or applying code to a personal interest? If you love to hike or bike, make some snazzy app with GIS involved. If you like to tinker with smart home stuff, show me something interesting with arduino or pi.

Sadly, HR doing pre-screening does not understand GIS or PI/Arduino, they do keyword match over known things, mostly known PL, popular framework, something stating that said tech were used in some activities.

If they can’t pass through pre-screen, what use is the cool projects? Besides, bootcamps must make life easy and fun, teaching anything outside packaged topics will scare away the students and the revenue.

How long until the autonomous Waymo taxis start giving stock tips? Buy GOOG?

If I were Waymo, I’d try to gain some market majority(or competitive majority, e.g. taxi vs uber), and then slowly start test pilot of showing ads while the ride is running…

Or even audio ads works, pretend it to be radio fully generated with generative AI and inject subtle ads.

Providing stock tips have zero revenue possibilities, but could be useful to have a channel talking about stock market movement today and what the experts are thinking.

Let alone startups - the higher interest rates are hurting everyone across the board. For example in my world - Telco Infra and MSPs which did a lot of Capex spend in light of 5G deployments aren't able to keep up with the interest rate hikes putting pressure on repayments for cash raised for the spectrum auction spend and the like. 5G FWA may be in the industry news, but there hasn't been a killer application that has allowed for revenue growth to match those payments. Only way to handle that has been layoffs.

I've been thinking about the AI gains a lot. If individual developers became, say, 20 percent more efficient at coding, the organization would potentially see even more gains, because all of the reduction in time spent coordinating between people. I.e. a 2-man task that needed 20 man-hours of work, 8 of which hours of work were just communicating, becomes a 1-man 10 hour task. Kind of an extreme example, but communication and coordination is extremely inefficient a lot of the time!

> communication and coordination is extremely inefficient a lot of the time

To the point where in many places it's optimizing communication and management that might be the low lying fruit - as far as programming efficiency.

After 24 months of LLM code seeding very subtle and confusing bugs into production codebases, the cracks will start to show, and we’ll probably need to hire 2x as many engineers to unwind the mess.

I don't think "AI" reduces head count, I think software that's built correctly can reduce headcount (by a lot)

> There's also the absolute flood of CS graduates in recent years contributing to the number of applicants. I know governments have been screaming for more STEM graduates, but I suspect the numbers are far higher than what is actually required.

Alot of these grads want fancy silicon valley free lunch ping pong jobs. Those are super competitive. Cities across the US have plenty of "boring" software jobs - in medical billing software, at insurance companies, at credit unions, for hospitals, in defense, etc. But those aren't in SF bay area or Seattle or NYC or wherever these young people want to live.