3 ms·

> Keep telling yourself that. What we do is soooo hard and special and we're soooo smart. I flip-flop every 6 months between thinking that everybody can do thi

by acchow 4y ago

> Keep telling yourself that. What we do is soooo hard and special and we're soooo smart. I flip-flop every 6 months between thinking that everybody can do this and being absolutely floored that I myself can keep up with this. One day we're writing HTML and CSS, and a few years later we're learning about the shadow DOM and why we may want to use React Native Web to run React Native on the web. One day we're writing a LAMP CRUD app, and a few years later we're deploying a service mesh for resilience, visibility, and configurable networking - concepts that didn't even exist when we were writing CRUD apps. I'm not a doctor/surgeon, but I don't think that the kind of work they do changes year to year at this pace? I guess you could say that none of these concepts are really that difficult conceptually, but...what is...? Algebraic Topology? Chaos Theory?

jghn 4y ago
It's a bit of both. Most of the work in our space is a lot lamer than we like to make it out to be. Almost anyone, with a modicum of training can do this. The hard problems, and by "the hard problems" I'm not talking about crazy algorithms but rather what does one do when the poop hits the fan, that's harder. Most people could still do it, but it takes a lot more training and experience.
vineyardmike 4y ago
> flip-flop every 6 months between thinking that everybody can do this and being absolutely floored that I myself can keep up with this. I agree that it takes a certain way of thinking to be able to reason about strict logic and symbolic relationships. Just ask someone what (X or Y) equals when X and Y are both true. I knew people in my CS classes right up until the end that couldn't tell you correctly (because people in regular English think of "or" as "exclusive or" and its hard for some people to change their understanding). I disagree with the concept about the rapidly changing tech jargon. I think in SF tech startup scene it may be that crazy, but once I start a job, I use their stack, and it stays that way for years. Different companies different stacks though. Big-Tech is even easier since they have their own custom stacks that are pretty stable. > I'm not a doctor/surgeon, but I don't think that the kind of work they do changes year to year at this pace? On the contrary. It changes all the time. They always have to be learning about new medication, new surgical techniques, new diseases. Ask a doctor in 2019 to diagnose the Sars-Cov-2019 Novel Coronavirus. Then ask them about ivermectin. > I guess you could say that none of these concepts are really that difficult conceptually, but...what is...? Algebraic Topology? Chaos Theory? I would say that real theory is actually pretty hard in CS. Math is famously unapproachable to people who don't "get it" and I think CS theory shares similar logical rigor - honestly similar to law and how it is a "different" logic than most people are ready for. I know rocket scientists, and their classes were actually-intense math - but conceptually/logically easy IFF you could do the math. The basic biology of how a viral infection works is pretty straightforward (and taught to 13yo's), but again, thats not the hard part of the job of a doctor. But most CRUD apps are not. And most people write CRUD apps. At my BigTech jobs, the company always had custom-in-house build tools that were great at doing deterministic reproducible builds and they were always based on strong theory and rigor and had white papers. But how many people actually have to do that at those companies?
- Apocryphon 4y ago
  A lot of the difficulty to modern web development and so forth don’t even feel like engineering, so much as navigating bureaucracy. The scale of complexity in modern codebases has turned us into software bureaucrats.
maneesh 4y ago
One is development. The other is system ops. Some people like to handle both, and others don't. I spent the last year migrating our backend off of heroku onto bare metal Proxmox servers with unlimited VMs. It blew my mind that 0, literally 0, of my developers took up my offer to give them free VMs (including KDE/Gnome with NoMachine/NX!!). It turns out that my devs just want to be devs, and not use their free time to play on sys ops and k8s and ceph clusters. When I saw this happen, I started hiring staff for sysops separately than expecting my devs to do so, and see much more clarity in output. In summary, if you're writing the LAMP CRUD app, maybe someone else oughta be the one to deploy the service mesh.

by acchow 4y ago

jghn 4y ago

It's a bit of both. Most of the work in our space is a lot lamer than we like to make it out to be. Almost anyone, with a modicum of training can do this. The hard problems, and by "the hard problems" I'm not talking about crazy algorithms but rather what does one do when the poop hits the fan, that's harder. Most people could still do it, but it takes a lot more training and experience.

vineyardmike 4y ago

> flip-flop every 6 months between thinking that everybody can do this and being absolutely floored that I myself can keep up with this. I agree that it takes a certain way of thinking to be able to reason about strict logic and symbolic relationships. Just ask someone what (X or Y) equals when X and Y are both true. I knew people in my CS classes right up until the end that couldn't tell you correctly (because people in regular English think of "or" as "exclusive or" and its hard for some people to change their understanding). I disagree with the concept about the rapidly changing tech jargon. I think in SF tech startup scene it may be that crazy, but once I start a job, I use their stack, and it stays that way for years. Different companies different stacks though. Big-Tech is even easier since they have their own custom stacks that are pretty stable. > I'm not a doctor/surgeon, but I don't think that the kind of work they do changes year to year at this pace? On the contrary. It changes all the time. They always have to be learning about new medication, new surgical techniques, new diseases. Ask a doctor in 2019 to diagnose the Sars-Cov-2019 Novel Coronavirus. Then ask them about ivermectin. > I guess you could say that none of these concepts are really that difficult conceptually, but...what is...? Algebraic Topology? Chaos Theory? I would say that real theory is actually pretty hard in CS. Math is famously unapproachable to people who don't "get it" and I think CS theory shares similar logical rigor - honestly similar to law and how it is a "different" logic than most people are ready for. I know rocket scientists, and their classes were actually-intense math - but conceptually/logically easy IFF you could do the math. The basic biology of how a viral infection works is pretty straightforward (and taught to 13yo's), but again, thats not the hard part of the job of a doctor. But most CRUD apps are not. And most people write CRUD apps. At my BigTech jobs, the company always had custom-in-house build tools that were great at doing deterministic reproducible builds and they were always based on strong theory and rigor and had white papers. But how many people actually have to do that at those companies?

Apocryphon 4y ago
A lot of the difficulty to modern web development and so forth don’t even feel like engineering, so much as navigating bureaucracy. The scale of complexity in modern codebases has turned us into software bureaucrats.

maneesh 4y ago

One is development. The other is system ops. Some people like to handle both, and others don't. I spent the last year migrating our backend off of heroku onto bare metal Proxmox servers with unlimited VMs. It blew my mind that 0, literally 0, of my developers took up my offer to give them free VMs (including KDE/Gnome with NoMachine/NX!!). It turns out that my devs just want to be devs, and not use their free time to play on sys ops and k8s and ceph clusters. When I saw this happen, I started hiring staff for sysops separately than expecting my devs to do so, and see much more clarity in output. In summary, if you're writing the LAMP CRUD app, maybe someone else oughta be the one to deploy the service mesh.