Hacker News Reader

Comment by bradlys | Hacker News Reader

Are you advocating for everyone to create their own SaaS here or what?

End of the day, most engineers need to join employers. We can’t have 10M+ different SaaS out there and each engineer develops their own personal brand of it. That’s not how software scales. Most ICs are never going to be in a position where the CEO is going to listen to them about what product we should develop and be given the time and resources to investigate that stuff.

Your assertion doesn’t even acknowledge infra engineers who clearly don’t work on product facing work. It’s a bit ridiculous.

View on HN

I see the era of permanent tech employment as coming to an end, and tech will have to transition to being closer to a trade. You don't expect to pay builders/plumbers/architects/electricians and keep them around forever after your house is built right? You will at best only need them occasionally on a one-off basis for any fixes or alternations.

A properly designed tech solution (aka, not one designed to pad a resume) for upkeep should only require a small fraction of the manpower needed to originally build it. I expect companies keeping in-house "skeleton crews" to run things and handle maintenance and small changes, with larger ones being done by a (potentially returning) cast of contractors who come in on a project basis.

So the solution for permanent employment would be to either join one of those skeleton crews or join an agency/consultancy. Otherwise, the only way out is freelance and curating your own client portfolio. Maybe you can partner with others and form a coop to have more manpower and handle bigger projects - essentially building your own agency.

> Your assertion doesn’t even acknowledge infra engineers

My assertion is that engineers who want to succeed in the current market will need to know enough infra/sysadmin to support their solution, or farm it out to someone who will. If they don't, someone else will outcompete them.

Infra is also a huge breeding ground for resume padding and architecture astronomy (and there were no incentives to discourage it - in fact cloud providers are still laughing all the way to the bank). It doesn't inherently mean all that complexity was necessary. So I expect "infra" specialization to become a nicher skill, one or permanent skeleton/maintenance crews or hyperscalers who actually need the complexity.

> who clearly don’t work on product facing work

I'm intentionally being vague when I talk about "solutions" and solving business problems. I do not mean it should be an app, a product or even have a UI. A cron job running a shell script to copy files from one place to another may very well be the right solution to some business requirements. Infra engineers would be a perfect fit for problems that don't involve significant UI work (I myself prefer those as I hate frontend with a passion).

I think the era of high specialization in tech might be in decline; tools are good enough and hardware is good enough and the field is mature enough that most business problems can be solved just fine by a generalist (bringing in specialists on an ad-hoc basis when needed) - or otherwise they'll get outcompeted by someone who can.

Yeah, this doesn’t really land.

Your argument is basically “we’re all going to be outsourced by Indians.”

Your EU-centric view of how contracting works (probably more due to labor laws than anything else) isn’t an accurate assessment of how Silicon Valley is going to work.

> we’re all going to be outsourced by Indians

Outsourcing has its place and a lot of gruntwork/ongoing maintenance is fine for it. Post-ZIRP it turns out the true value of software is a lot lower than previously thought, and for a lot of solutions, Indians is all they can afford (because the alternative might be to pay someone to do it by hand).

Where outsourcing fails is primarily when business & domain knowledge is required - outsourcers will obviously not be aware of internal company processes or even the target market for the required solution (just like we would be bad fits for trying to spec out a product targeting the Indian banking market for example).

So I see the industry and market readjusting to the true value of software, and as a result people will pick a mix of high-value implementation work (where local domain knowledge is required) and boring/maintenance gruntwork, or get outcompeted by those who can (and this is a global market, so you are competing with Indians(.

I'm not advocating for one way or the other btw, I have no skin in this game (technically trying to push people into freelancing and consultancy works against me). But I'm just trying to be realistic - the tech comp during the ZIRP era was never going to last, and even if ZIRP continued, more and more people entering tech means comp is still bound to decrease. You can either remain in denial hope those jobs come back, or plan for the worst.

The true value of software, post-ZIRP, is still obvious for the largest and most powerful companies in the world. They print cash.

Yes, ZIRP increased tech comp. It did not magically inflate the value of software, which is always directly assessed against what people actually find useful.

The value of most software for most companies has a limit, pushed down both by supply of software engineering labor and that manual labor is actually quite cheap and your software must inherently cost less than what it would to just pay a human to do it.

There are always outliers obviously - whether it's the adtech giants being in the right place at the right time, or niche industries (avionics software costs a lot more to develop, and yet even then the actual SWE salaries there don't reflect it), but by and large the value of software is nowhere near what a decade of market distortion led us to believe.

Remember most of us (including me) are plumbers, not rocket scientists. It's just that a decade of market distortion allowed us to play rocket scientist (with only a tiny minority of those rockets actually ever needing to fly). Now we're back to being what we really are.

> The value of most software for most companies has a limit, pushed down both by supply of software engineering labor and that manual labor is actually quite cheap and your software must inherently cost less than what it would to just pay a human to do it.

Software, in the limit, is FREE. It gets written once for a finite cost and then can be copied ad infinitum. Even stratospheric labor costs are acceptable because of this.

> (avionics software costs a lot more to develop, and yet even then the actual SWE salaries there don't reflect it)

But the cost to develop a piece of software (or anything, really) has nothing whatsoever to do with its value!

Avionics software is expensive not because it's inherently more complicated (high-quality OSS hobbyist implementations like Ardupilot exist for free), but because it has to undergo system integration tests because of FAA regulations.

And, despite lower pay for avionics engineers, the value to the end users of quality avionics is literally the difference between life and death.

> but by and large the value of software is nowhere near what a decade of market distortion led us to believe.

Where do you get this idea? Tech, as a whole, is a wildly profitable sector of the economy. It is a primary driver of US GDP growth and productivity [1].

> Remember most of us (including me) are plumbers, not rocket scientists. It's just that a decade of market distortion allowed us to play rocket scientist (with only a tiny minority of those rockets actually ever needing to fly). Now we're back to being what we really are.

I get that ZIRP is ideologically unpopular, I really do. You can only see so many "Uber but for dogs" startups get multimillion dollar funding rounds before you get cynical.

That being said, interest rates have been falling for centuries [2]. Low interest rates are a fundamental reflection of a society that's gotten better at allocating investments efficiently.

It turns out that software, because of the whole "zero marginal cost" thing, is a spectacularly efficient investment.

1: https://cpram.com/fra/en/individual/publications/megatrends/...

2: https://www.nber.org/digest/202212/real-interest-rate-declin...

Or you know, you can just change jobs. The average tenure for software developers at any one job is less than 3 years.

I do work in consulting. But it’s by far not the only way.

> I'm intentionally being vague when I talk about "solutions" and solving business problems. I do not mean it should be an app, a product or even have a UI. A cron job running a shell script to copy files from one place to another may very well be the right solution to some business requirements.

Completely true, but instead this would be an event-driven serverless microservice that does some complex ETL thing to move files between three S3 buckets before landing in a "data lake" where analysts run three queries a month.

(I've seen many of these exact scenarios)

Upthread regarding astronauts is absolutely correct---there's been years and years of incentives to create messes like this, and yes, cloud providers absolutely facilitate this, and in many cases are the direct cause.

Reply on news.ycombinator.com