The asteroid currently hitting frontend web development
32 points by cgrinds
32 points by cgrinds
Maybe an unpopular opinion: But frontend dev has been over for me for many years.
I grew up learning HTML, that gave me my first job, that spiralled into doing backend DEV, and so on. Web development used to be fun and then at some point it was all SPAs and I hate React. So maybe it's just a coincidence but I completely agree with "The frontend is less risky to just hand to an agent." My current job uses React and I much preferred angular (v1 over v2 even), so it's not like I've not been touching it for 10 years, it's just that the enjoyment is long gone.
All my personal stuff will stay handmade and without anything but Vanilla JS (hey, at least I migrated from Zepto the last couple years).
All my personal stuff will stay handmade and without anything but Vanilla JS
I generally agree with everything you said, though I find myself using Alpine when I want some interactivity. I think it was made for people in the "I want some interactivity, but I don't want to build an entire app" category.
Can't help but feel the depreciation rate for web development knowledge is at least part of the picture.
Every single time I've dipped my toes in the field, the rules are different, the languages have changed, the frameworks and concepts are new, browsers are doing new insane shit that is incompatible with everything else and what used to be done before still works perfectly well but it isn't the new way of doing things so we've abandoned it, the set of things you inexplicably have to know to do for things to work has changed.
Putting in the time to become an expert in something that has the half-life of an actinium isotope does not feel like a worth-while investment at all, maybe it did for a moment when this sort of knowledge was in some sense exclusive and something you could sell, but in the presence of a machine that can know it for you, I can see why the ship would begin to take in water.
This resonates with something that's been on my mind since I read this comment on another post regarding AI use in academic environments.
(...) Another thing that a lot of people have commented on is the desertion of office hours and reduction in collaboration among undergrads.
When the clanker is so available to answer our questions then we stop asking each other, discussing amongst each other or considering other perspectives. I have seen this at my $WORK and in myself, and for now I don't have a better solution than to be more self aware, organizationally aware and culturally aware.
Front-end skills were undervalued already. For as long as this has been a job, there has been pressure to just make whatever quickly, merely to be approved by a manager who's only going to look at the website's screenshot for 10 seconds. On top of that the devs are underpaid and have to debate whether they're even real programmers. The whole field is treated as a "junior" job, which ends up being a self-fulfilling prophecy.
Your comment reminds me of a (long ago) previous job where front end developers were seen as imperfect until they evolved into backend developers. Not full stack, but pure backend.
For context, I'm a front end developer.
Yep, wasn't uncommon. I think partly it was the PHP problem. Easy enough to pick up, so there were just a lot of people who were not very good at it (and not very experienced), and so the "real programmers" coming from the backend could catch up really quickly. At least in theory, I've also seen that backfire often enough.
A good frontend developer was a very good addition to a team.
It's quite sad that anthropic and openai basically get to define which framework will be the default for (likely at least) a decade. Especially given that they can choose the "none of the above" option and go back to modern vanilla JS which is all many pages need. In some ways this could've been a good reset from the overcomplicated world of what web development became. I don't know if I'd rather like no impact, or fully going back to simple things and the "everything needs an API" era. We still have a chance for the second option from AI, but people overall would rather invent yet another meta compiler/bundler.
I share similar concerns(I don't think I would call it an asteroid tho), at least around the projects I've seen, as long as it looks good it's enough, later on it can be refactored(rarely happens, unless somebody proposes a rewrite). It's disheartening, because I'd expect good developer tooling[0] to make it easier to write more accessible-friendly code, be it through linting/tests/components or just general awareness of what kind of impact your markup has on an user.
Instead it is generates...more of everything: more code, more unreadable non-slot-driven layouts, more breakpoints, more buttons with all kinds of handlers, let's throw in some TypeScript, more random null checks if you want better code, do React and Tailwind, you want less JS? Well go shave it off or play type tetris with TS. All of this not working? Write some markdown that'll do it just fine.
I, for some reason, have to argue about why a bunch of barely readable, not tested, blob of code isn't the way to create resilient software that takes into consideration user and their accessibility needs, and for a even more absurd reason, it's an uphill battle. I do not have the marketing skills of an entire industry. It's weird, and the scale makes for some really bad incentives, in a place that's already ridden with those.
[0]: yes I know, there are people out there for whom the LLM generates better frontend code than they'd ever write, <enter some other shitty comparisons for humans vs LLMs>, and I don't care for that, I like people having jobs they like and getting better at it, which means making mistakes.
It's really heartbreaking to see a whole field disappear because some companies have commoditized it and have it on tap. It's somewhat positive that one day local models will catch up, but by then the industry will be a very different place. Maybe one where we focus more on higher-level concepts and we'll all be able to build bigger projects, but it'll be without a doubt one without this amazing journey of learning and sense of mastery over the basics.
I'm outside web dev, but I'm still amazed by it. How do you prevent things from stagnating and coming apart though, when few are left who understand it? It feels like a conversion from engineers to computer priests.
that first paragraph is an impressive example of how AI is eroding creative minds and those that used to do exploration and champion communication...