[sgarland]

> The real horror isn’t that AI will take our jobs—it’s that it will let people in who never wanted the job to begin with.

I fully agree. This already happened with the explosion of DevOps bullshit, where people with no understanding of Linux got jobs by memorizing abstractions. “Stop gatekeeping,” they say. “Stop blowing up prod, and read docs” I fire back.

[alexjplant]

> explosion of DevOps bullshit

The fact that somebody can "be DevOps" or work as a "DevOps Engineer" is exemplary of the fact that DevOps as conceived and DevOps as practiced are two very different things. The former would be engineers taking ownership of deployment, collaborating horizontally, and practicing tight feedback loops. DevOps as practiced is the time-honored tradition of a dev team and a cloud team playing tennis with a grenade that is a questionably-stable SaaS that people volley back and forth with rackets like "let's roll the pods" or "it worked on my machine".

> people with no understanding of Linux got jobs

This happens in every industry with every job title. I've worked with Senior+ developers that mutated React props, didn't know how to use Git, couldn't read Java stack traces, etc. I myself have been paid money to do a myriad of things that I have no business doing (like singing or playing guitar or mixing cocktails). It's the way of the world.

[sgarland]

The platonic ideal of DevOps is fine, yes. The problem I’ve consistently run into is there is a vanishingly small percentage of devs who want to do Ops, or even consider that a server with real constraints is running their code. Performance takes a backseat to DX, which AFAICT is code for “I don’t want to do anything but write code and push.”

I should note that I think this is fine, if and only if you have specialist teams who respect each others’ abilities and recommendations. A dev team shouldn’t have to worry about standing up infrastructure, but similarly, when the infra team tells them that their app is consuming far more compute than it should be, the dev team should profile and improve their code instead of asking for more compute.

[alexjplant]

> there is a vanishingly small percentage of devs who want to do Ops, or even consider that a server with real constraints is running their code. Performance takes a backseat to DX, which AFAICT is code for “I don’t want to do anything but write code and push.”

This is because many tech orgs judge performance by "business impact" and dev and infra teams have different value propositions.

A dev writing and shipping features conspicuously demonstrates value because they write things that users pay for. Product features are novel and specific to a business; a developer can write "Built an automatic cardinal grammeter synchronizer that generates $100M in ARR" on their resume. There is essentially no upper bound on the dollar value of their impact.

Infra teams, on the other hand, have no direct way of quantifying their work. Even though their job is very necessary it's difficult to move the needle as far as cashflow is concerned. People tend to consider infrastructure a commodity so it ends up being one of those things that's only noticed if it breaks. A good day on an ops team means no trouble, not more money. The cherry on top of the sundae is that more users is more money which means more prestige for developers whereas more users means more load which means more problems for the infra and ops teams to contend with.

This incentive structure is why many devs would rather spend all of their time slinging code than worrying about k8s cluster nodes or IAM roles.

As a jack-of-all-trades I don't think this a healthy dynamic. I've led infra teams and dev teams both and know what it looks like on each side of the fence - a little consideration goes a long way and it's one of many reasons why I think that it's healthy for everybody to do a bit of everything in a tech org. People are much less likely to "play defense" if they understand a team's motivations and don't anticipate a negative interaction.

[mountainriver]

This is already happening and it’s a colossal mess

[bangertho]

This already happened with Linux where people with no understanding of electrical engineering got jobs memorizing abstractions.

I look forward to a data driven system future where a few functions transform the machines electromagnetic geometry to solve a task based upon the most efficient energy model for solving a task as we continue to compress from the model all the non-essential syntax sugar of modern software.

https://arxiv.org/abs/2309.10668

[sgarland]

Fun fact, I was a Nuclear Electronics Technician, and have a decent understanding – though admittedly not to the level of an EE – of how computers work at a very low level. I also worked for a chip fab for a couple of years, and literally made silicon wafers. I also rewrote the US Navy’s microprocessor training course for Nuke ETs, updating it from the Motorola 68000 to the Intel 386.

I make no claims that I fully understand anything, but I do have a decent understanding of how a CPU works from the level of doped silicon and up. Crucially, I read every doc I could find at every one of those jobs. You can learn enough to do the job, or you can learn more. That is a choice that everyone makes.

More generally, I’ve been playing with Linux and computers in general for over 20 years, and when I finally got a job in tech about five years ago, I was stunned at how little people knew about how computers work. I don’t expect (nor do I think it’s helpful) anyone to know how a bus arbitration cycle works, but I assumed that things like IOPS and throughput would be generally understood.

[nyarlathotep_]

> More generally, I’ve been playing with Linux and computers in general for over 20 years, and when I finally got a job in tech about five years ago, I was stunned at how little people knew about how computers work. I don’t expect (nor do I think it’s helpful) anyone to know how a bus arbitration cycle works, but I assumed that things like IOPS and throughput would be generally understood.

My expertise is only in sleeping until 11am on weekends, but I too started in "tech" after being a lifelong hobbyist and have been continually shocked at how concepts like "pass by reference" are alien to a seemingly large portion of the people that I've worked with.

People often fail to know things that are basically "table stakes" in the domains they ostensibly work in, to say nothing of even being aware of something like L1 cache or how code they write could interact with it.

[sgarland]

AWS: “We’ve separated compute and storage by a large physical distance, which makes relational databases better.”

People who knew better: wild laughter.

[binary132]

It’s actually kinda terrifying the more you find out how incompetent most of the people writing most of the software most of the world runs on are

[sgarland]

Much like how the more you know about how computers work, the more mind-boggling it is that they work. The Internet? Magic. CDMA? Wizardry. EUV? Sorcery.

[SirSavary]

Could you elaborate?