The death of software development
mike.tech
The death of software development
1–10 of 29 posts
Re: The death of software development
#2Re: The death of software development
#3>If an idiot like me can clone a product that costs $30k per month — in two hours — what even is software development? (...) The new software developer is no longer a craftsman. It’s the average operator, empowered
If entire industries employee counts are decimated and/or commodified, this means the "new software developer" wont find people to pay for what they create (whether software or a software driven service). For the majority, it also means further degradation of the kind of jobs one will be able to find.
Re: The death of software development
#4Re: The death of software development
#5Beware the AI+climatechange Armageddon!
Re: The death of software development
#6^^ Its been a rough couple of years with these kinds of upshots being constantly posted.
Yes, things have changed. Is the entire software development world about to collapse because of LLMs? Sorry, no. I'm impressed by the capabilities we now have at our disposal, but the LLMs still do a bunch of dumb stuff. Some of the worst bugs and edge cases I've dealt with this year were almost unnoticeable things an agent added. With that said, there is still a ton of juice to squeeze from this paradigm. I really just wish everyone would stop pretending the sky is falling.
Re: The death of software development
#7A Bloomberg Terminal doesn't cost $30K/mo. It's more like $30K/yr and less if you buy in bulk.
Re: The death of software development
#8What's dying is the programmer-first job. That guy whose main use is that he knows how computers work, and secondly that he is a human who can understand how some business works, and can do the translation.
The other type of programmer is the business programmer. I started on this end before an incredibly long rabbit hole swallowed up my life. This is the person who thinks he's a finance guy, or an academic, or an accountant, or any number of things, who realizes that he can get a computer to help him.
This type of person is grounded in the specific business they come from, and has business-level abstractions for what he wants the computer to do.
AI is still imperfect, so it is still in your interest to know how the computer works, especially as you dive into things where your model of the machine actually matters. But it allows the person with the business view to generate code that would previously be their second job. He can QA the code on a business level. This used to just be called Excel, which would generate horrors for anyone who could actually program, but it is still the glue behind a huge number of business systems, and it still works because ugly often works.
I liken this to previous revolutions in IT. At one point schools had begun churning out literate people, and they started spilling out into the business world as clerks. You could learn how to read and write, and that would get you a job sending correspondence to India, that sort of thing. And that would be your way into the organization, and maybe you'd eventually learn the business itself.
People who typed stuff had a similar fate. There used to be rooms of people who would type letters and send them. Now the executive just types the letters and sends them off by email.
If you're a translator first, AI is not great for you. If you managed to turn your translation skills into executive skills, then you are happy to pull the ladder up.
Re: The death of software development
#9This is the part that terrifies me. Generating code has never been the bottleneck; understanding it has. If you aren't reviewing the code, you are effectively introducing a black box into your stack that you are responsible for but do not understand.
While on point, I like the patterns Effect is introducing, but I have already been using them pre- and post-AI for a long time now (especially in TypeScript). It's not an innovation. We shouldn't be surprised when LLMs are better at working on robust code bases with modular design.
Re: The death of software development
#10Well, I suspect increasing use of code generation mechanisms will lead to improved design and debugging skills. Practice makes perfect, eh?