Hacker News Reader

Comment by stefanos82 | Hacker News Reader

I feel trapped to be honest with you...on one side is young developers that their mind is a sponge comparing my old grumpy half-burned brain-cells (lol) and on the other is the AI that does incredible things that could easily replace us one way or another.

I don't know anything else apart from technology and I feel doomed -_-

View on HN

I feel similarly. I've loved computers and programming from an early age, and I was inspired by the stories of researchers like Dennis Ritchie, Ken Thompson, Marshall Kirk McKusick, Alan Kay, and others. However, after graduate school and after spending nearly a decade in industry, I discovered that the world today is very different from the world these researchers lived in during their heyday. There are no groups today that have the ethos of 1970s Xerox PARC or Bell Labs. There seems to be less room for long-term exploratory work; instead there's immense pressure to quickly deliver short-term results and produce deliverables such as papers and code. The trailer of the documentary "Message Not Understood: Profit and Loss in the Age of Computing" (https://messagenotunderstood.com) summarizes how I feel. I'm also disillusioned by the state of personal computing today and by the overt attempts by platform companies to squeeze their users as much as possible.

Even so, quite frankly I love computing and I don't want to leave it behind, even if I feel disillusioned about the tech industry. Some of my thoughts include becoming an independent researcher and also starting a non-profit software company that is devoted to improving the personal computing experience, but I have to figure out how to make a living while pursuing these ideas, and it seems that the cost of living is outpacing my ability to keep up (especially housing costs, which jumped over 30% since 2020 where I live).

> Even so, quite frankly I love computing and I don't want to leave it behind, even if I feel disillusioned about the tech industry.

The commentary here has allowed me to reach an interim decision (subject to change in the future as I ponder these things more deeply).

I am going to find another line of work. That way, rather than leaving it behind, I can make it a hobby again and do the sort of projects that improve my life.

Perhaps that way, I can honor and maintain my love of computing, and earn a living doing something else. I'm perfectly OK with taking a large pay cut if I can live a happier life in exchange for that.

I've been thinking about this, too, with teaching being my prime candidate. One of the downsides of leaving the tech industry is the reality of taking a massive pay cut at a time when the cost of living (especially housing) has been rising rapidly. As an aside as an American I feel very pessimistic about the state of the American economy and about the medium-term future of this country; it's getting harder for people to get ahead thanks to rising costs (especially housing, health care, and higher education), our politics has become extremely polarized, and the general mood is becoming more tense, as if we're one or two crises away from this nation completely falling apart. It seems that this nation's economy rewards rent seekers, and it's becoming harder for people to make a living through other means.

That's exactly what I have been thinking for a while now.

Thank you John, you gave me hope that I am not alone, that feels this way.

Where can I find the complete film? (https://messagenotunderstood.com)

The film hasn't been released yet. There is a 30-minute version that has been shown at various events, but the final version of the film won't be released until late 2023 or early 2024, as there are still some interviews the directors want to do.

Yeah I really regret going into the field now, especially the way I did it.

There’s just no where else to go that isn’t just less of the stuff I like about it and more of the stuff I hate about it. I had a plan to go back to school and switch fields, but the more time passes, the more I realize this isn’t going to happen. Everything keeps getting more and more expensive, both in terms of money, and more notably time.

> There’s just no where else to go

Yeah, that's a problem. This is literally the only marketable skill I have.

What field were you thinking of switching to?

I don’t like to talk about that too much on this site, but I’ve have been a number of ideas, mostly still in technical fields.

I thought I had my head set on something, but it was, funny enough, all predicated on having a well paid software job to fund the jump. I planned on starting this year, using community college along with statewide transfer programs to get started as quickly and as affordably as possible, but the layoff and current economic/financial situations rocked me, and I’m not sure I have the capability brains, or money to pull it off anymore.

The use of LLMs as a programming aid are another strong indication to me that I need to get out of this business.

Why? I've been employed in the industry since 2017 and I love Copilot and ChatGPT. They make everything more convenient, especially repetitive code, and suggest good new ideas sometimes. They've been a godsend for debugging, figuring out/generating command line options, setting things up like ftp servers with weird configs, etc. It's like we've gone from walking around on foot to figuring out horseback riding. I feel freshly inspired in my productivity since I've started using AI tools.

Not to mention, if you're able, you can paste in code to ChatGPT to give context and tell it to generate unit tests. I've found it does a generally good outline of code that annoys me to write, and then I can tweak the cases and add more as I need. Makes a chore I dislike much, much easier.

But writing my own unit tests and code outlines, as annoying as they are, make me a better programmer and let me produce better code. I can't tell you how many times the act of writing unit tests -- just writing them, not even running them -- has given me an important insight into the code that I would have missed otherwise.

Using tools to do these things for me means that my skills and abilities will suffer.

I make great use copilot for test cases now. I still think through what I'm trying to do and start writing it, but as I go along copilot will tend to write what I had in mind anyway and complete 20 lines of the likes of `assertEquals("overriden address 1", model.getAddress1());` before I can blink. I'll also often enough be caught offguard by some strange output and realize it's because I did something wrong, so I catch the problem much sooner.

Since 2017? You're still wet behind the ears!

It tells me I need to get out because it represents a shift to a higher level of abstraction in programming. In the ideal I hear expressed, instead of programming you'd be just describing the code you want and letting someone/thing else do it for you.

That's management, not programming. I want to program.

Understand that I'm not even saying there's anything wrong with that. It's just that it indicates that the things that I enjoy about being a programmer are the things that are being optimized away.

> It tells me I need to get out because it represents a shift to a higher level of abstraction in programming. In the ideal I hear expressed, instead of programming you'd be just describing the code you want and letting someone/thing else do it for you.

To me, LLMs are just "advanced" (read: statistical) codegen, unlike compilers and interpreters that are deterministic (should generate functionally equivalent code). But there's as much novelty there as there was when the C compiler or LISP repl was introduced.

> To me, LLMs are just "advanced" (read: statistical) codegen, unlike compilers and interpreters that are deterministic

That's fair, although I would disagree with the notion that compilers are code generators.

But using code generators is not something that is interesting to me even a little. They remove much of what interests me about programming. So your explanation -- while it makes sense and I wouldn't argue that you should feel any differently -- underlines that I need to get out of this industry.

I'm somewhat mixed. From what I've used, for relatively high level boilerplate code, the stuff I hate, it's okay with. For anything more complex, there are a lot of mistakes, and you still have to understand the technical and the domain to spot the mistakes. I think it will be all that much harder to become skilled as AI generated code becomes more common, because fewer relatively speaking, developers will be able to spot said errors and correct for them.

It's hard enough to keep up with PR reviews of a half dozen developers while doing my own work now. Hard to imagine that with 3-4x the output.

Reply on news.ycombinator.com