Hacker News Reader

Comment by lend000 | Hacker News Reader

Not mentioned enough on this topic: not all engineers are software engineers and not all software engineers never need to touch hardware.

If you plan to work on/with web frameworks, advertisements, data science, or pure CS topics for your whole career, then there's probably no excuse to force you to come into the office 5 days a week. But good luck doing that as a software engineer at a company like SpaceX, Tesla, Intel, AMD, defense contractors, etc., let alone as any other kind of STEM professional. I think it's an important distinction to make. For me personally, some of the coolest work I can imagine doing is likely going to be done in the office/lab for the foreseeable future.

View on HN

You're right that not all engineers get to WFH but more can than you'd imagine.

I know defense contractors that are working from home. Two of them did so before the pandemic too - remoting into hardware they're not allowed to physically touch even if they were in the lab!

A neighbour is a civil engineer and the huge screens he needed to do his job were only in the office. Now that he's moved that huge screen (and I mean massive, it's like a telly), the office have told him that he only needs to go into the office to meet clients.

I have a product design dear friend who bought himself a Prusa 3D printer and now can pretty much 100% WFH. He prefers it because he no longer has to share resources.

A neighbour is a network engineer and he spent 2020 decommissioning hardware in the server room and moving it into the cloud - something he said was inevitable for his business but the pandemic hastened it. He now WFH pretty much 100% of the time.

I know someone who works for a SSD manufacturer in the states and they WFH most of the week.

All the web/data devs I know WFH pretty much full time - except one but only because his boss is draconian.

As a software engineer who just spent the past week in an airgapped lab debugging hardware issues, I can still do most of my work from home.

90% of the software I write works just fine in an emulator. We don't even have nearly enough hardware to give every engineer access.

Yep. I do control systems for 20+ tonne hydraulic robot arms (among other things). 90% of the work I don’t have to catch a plane for can be done at home.

> 90% of the software I write works just fine in an emulator

As it should be. Investing some effort in my development environment has paid back every time.

If it has paid back every time, perhaps you are not investing enough effort?

(Only slightly tongue in cheek. At the margin, some of your efforts should fail.)

Actually you are probably right. And it's not just building the environment: also unit testing, simple baseline algorithms, etc... We have a tendency to skip them.

Whatever you build, you will need to change, as this is the nature of SW projects. Being able to make changes fast will be life saving at some point

> We don't even have nearly enough hardware to give every engineer access.

This is the problem I always run into too. An emulator should be standard on any system development program that has software interacting with hardware.

I have spent my whole career working at places like AMD and NVidia, either writing systems software or doing computer architecture.

Yes, we work with various forms of pre- and post-silicon devices but you will find that either they can be accessed remotely, or you can send them by courier to whoever needs one.

So, sure, _some_ people need to be in the lab touching the hardware, especially during bringup, but the immense majority of the people, the immense majority of the time, can work remotely.

Works at lesser known places as well :) Courier the boards, remote into a leee-nucks connected to them, work on the soc manufacturer's sample which you can order from anywhere and then remote into the real board. Lots of solutions to do the job from home.

> company like SpaceX, Tesla, Intel, AMD

I think you are overestimating the requirement to touch the hardware by few order of magnitude for a software engineer there. Just yesterday I was reading a linkedin post from some senior electrical engineer in Apple M1 division and how they achieved everything remotely. For a fabless company, I don't think most of the engineers would need to touch the chip. Similarly for Tesla and SpaceX, they had great two years when almost all software engineers were working remotely.

Yep, I read lots of Verilog nowadays. I think the implication is that you need to be on site to mess with experimental hardware. But that's not true. It's all colocated and cloud-based nowadays, even for companies that are otherwise paranoid.

I worked on flight software for cargo dragon, and one time I got an email asking for my help inside one of the capsules. I had to ask where the capsule was, and everybody lol'ed.

It was a fun day; I got to wear a bunny suit and a headset and climb around inside a space ship.

I e done firmware development from home for over a decade. I need help once and a while from someone in the lab to push buttons or make connections or do reworks or whatever. But I have no trouble getting most of my work done from anywhere I have a half decent internet connection. The oscilloscope runs Windows and I can VNC to it. The (very old) logic analyzer runs HP-UX and connects to an X server to present its UI. We have network controlled relays and power supplies. We can get a lot done without physically being there.

awesome, for what company do you work?

Yeah sometimes you do need to touch things that you can't bring home.

But there are people like me, people who do touch hardware.. at home. My desk is full of hardware. Behind me there's another desk with an oscilloscope, lab psu, dmm, electric load, a pile of probes, a soldering station..

It's not like we get new hardware every day, and when we do, it's not a big deal to wait a day or two for the courier to bring a prototype to my home.

This isn't true.

There are sibling comments here that mention the same, however, to add my own point here: I'm involved in a complex hardware plus software project which involves designing and doing a board bring-up for an Arm SoC Linux-based device. We need to visit the office only once in two to three weeks. For the board bring-up when we do not have the provision to provide the hardware to our software engineer yet, we have someone in the office connect the UART lines to it and have the typical VPN+SSH set up to perform the board bring-up for it.

Generally speaking, the engineers that are required for the labs would be electrical engineers and mechanical engineers. Also, even for these physical-based engineering disciplines, a lot of design is highly computer based thanks to great CAD, simulation, and future-looking generative design software from the likes of Autodesk, Ansys, etc. That being said, I personally feel that being in person really can help in the epiphany process for hardware-based projects.

Yes, but not all non-software engineers need to _touch_ hardware.

Eg if you are a civil engineer, lots of your work is spent doing calculations and designing things. When you do touch hardware, it's usually 'on site' and not at the office.

Most of my job is working on drivers for hardware that is in lab thousands of kilometers away. I've never been in that lab, I just ssh to system with hardware to interact with it.

The irony is that while you won't do it as their employee, the same companies will happily hire offshoring companies that work just like that from the other side of the planet.

Lol I worked for Intel remotely before covid as a full time blue badge... Though I guess I did have a title you mentioned above...

Aside from what others say, there is a trend to 'cloudify' even things like bench work in biology. Companies that allow people to run remote experiments. This will probably change the way university labs teach things

It depends on how long you need to think between testing, designing, and writing.

You dont need to be anywhere specific to read a data sheet

> SpaceX, Tesla, Intel, AMD, defense contractors

One of those things is not like the other things (it's the one where you build things that are explicitly designed to kill human beings). Not sure if you were calling that the coolest work, I certainly wouldn't call it that.

I think it was more about having physical hardware that you work with which might not be easy to bring home consistently.

Those things are also insurance against being killed. Are you supposed to just hope the bad people don't pick you? No, you invest in offense so you can have a better defense.

I know man, I know. I'm not so naive as to say that if a country was enlightened enough to disband their military, then other countries would leave them be. They would be a seen as a ripe fruit, and they would be picked.

I just wish it wasn't like this. I wish we didn't have people engaged in the business of enabling industrial scale murder. I wish it wasn't profitable. I wish the citizenry would see through the propaganda and hold their leaders to account. I wish it wasn't the nation's poorest who were sent to die for the benefit of the nation's richest.

The war machine is a rotten, gruesome system, and I acknowledge it's not going away any time soon. But the least we can do is agree not to actively contribute towards it.

Modern warfare is much less likely to involve the nation’s poorest dying for the sake of the richest. There’s a recent article about how the average American soldier is now better educated than the average civilian. The specialist weapons we’re talking about shift the balance towards capital and away from labour in the killing production function.

Is that a good thing? I’m not so sure. The age of mass armies was also the age of democratisation. If war is waged by defence firms and brilliant technocrats, then maybe defence firms and brilliant technocrats will run the country.

Understand, there's a lot of distance between 'we don't need a military', 'we need a military to defend American soil', and what we have today.

I am strongly in the 'we only need to defend American soil' camp. I'm not sure how much that costs, but considering we're a nuclear power, we have two relatively friendly neighbors and thousands of miles of ocean between us and our nearest peer enemies, it wouldn't surprise me if we could get away with spending a miniscule fraction of the 700B we do today.

I think it's pretty cool. And I think that kind of work needs people who think it's cool, or else it'll only attract evil people.

There's nothing wrong with building weapons for defense. It's arguably the entire purpose of nukes at this point.

Even for offense, other countries are working hard to build out their arsenals.

I'm personally fine with building AI weapons, for example. If a contract opportunity appears, I might take it. After all, where else are you going to see the most advanced weaponry in the history of mankind?

It's arguably the entire purpose of nukes at this point.

That is absolutely not true. There are people in the governments of every nuclear power would use their weapons aggressively given the chance. That's why policies of disarmament are so necessary. As long as humans have nuclear weapons the probability of practically annihilating our species and leaving any survivors living in the stone age again remains higher than zero.

Nuclear war is the single biggest existential threat humanity faces. We can deal with pretty much anything else.

I certainly hope you're right in your implied statement about climate change.

I completely believe humanity can and will deal with climate change. It'll take a few wars, an anti-capitalist revolution in a first world country or two, and a massive shift towards equality and justice for poorer people, but it will happen.

For what it's worth, I agree with you. One of my sneaky reasons for joining the military establishment would be to help make those sorts of changes from within. Or at least that's what I'd like to tell myself while designing devastating weapons.

I really do feel that reasonable people need to join such organizations, though, otherwise only unreasonable people will be in them. I recently found (to my dismay) that Von Neumann argued vehemently in favor of bombing the Russians before their nuclear program could be built, and all I could think was "This isn't the type of scientist that should have a lot of influence."

Defense contracts aren't all weapons - in fact, I'd wager most aren't (by quantity, if not by dollar amount). Due to how the US allocates its budget, a large portion of its publicly funded research goes through DoD contracts. The startup I used to work at had a contract for making prototypes of multi-material 3D printers for creating medical training models (ie artificial cadavers).

<snark>Tesla? or SpaceX??</snark>

Reply on news.ycombinator.com