You're already a Senior/Principal/Lead developer
Sure. You remember, toiling late at night over Uncle Bob books and hacking away at experimental code. Been there, done that, and have the badges to prove it. But now you're at the top of your game, so that stuff is well and truly behind you. Why should you bother learning a whole new paradigm now? You're already set, and it's almost 4:30.
Processes and toolchains just get in the way
We get it. You've heard it all before. The wacky CI/CD fever, linting fascists, and test-all-the-things lunatics. None of that nonsense was necessary then and it certainly isn't necessary now. You've been emailing firmware updates to the support team forever and it's almost never been a problem. The AI fan-boys can just go spin.
Your floppy disks have a larger context window
Let's face it. So-called 'frontier' AI models are topping out at 1M context windows. That's pathetic. You have a drawer full of 1.44M floppy disks to prove it. Besides, the biggest source file in your codebase is only 800k. So, what does it matter?
Failure is where all the fun is
"A man works hard for his filth just to have vagrants come and steal it." — George Liquor, 1991
I mean really. It's taken you decades to figure out how to debug this stuff. That's how you add value. There's no AI built that can understand the kind of breakage you have to tolerate in your codebase:
✦ "ERROR [0xFEEDFACE: 0xDEADBEEF] Interdimensional pointer dereference at
src/core/quantum_scheduler.rs:88:23. Expected FnOnce<(&'static Arc<TemporalSingularity>, &mut
[EventHorizon]) -> Result<FluxField, EntropyViolation>>, but received an unholy invocation of
std::mem::transmute::<&'a str, &'b mut CosmicRadiation>.
Product managers and the sales team just couldn't handle it
Have you met those folks? They can barely convince people to buy your products as it is. You can only imagine what circle of hell would open if new features and functionality started flowing like water. Besides that, we all know that customers are best kept wishing for more.