Hacker News Reader

Comment by monjerry | Hacker News Reader

Code quality is not an issue.....until it is. Everyone will be happy if things seem to be working and moving fast, it's only until you start seeing symptoms like not being able to change things without breaking existing features or even seeing an increase in tokens burned for reading complex code bases that you will start caring about code quality (which i imagine now people will just have AI try to fix it's own mess).

View on HN

Yes. This is the result of the industry being led by only managers and non-technical people who scuff when engineers who talk about taking more time to do things properly as if they are just fulfilling their own personal creative needs. As a result the engineers who are just willing to make leadership happy are the ones who end up making technical decisions, usually short-sighted that leads to compounding complexity and lots of headaches down the line.

I've always thought that good engineering should go smoothly by definition. You shouldn't be juggling a ton of bugs and broken develompent pipelines. The problem is justifying this in a work culture that doesn't care about quality and is stuck looking at numbers that can be immediately justified.

There are plenty of decades old codebases written in old guard languages that were suffering the same fate pre-ai. What is old is new again.

We will just repeat the same pattern, have the latest and greatest frontier models rewrite the software just like how many desktop monoliths moved to React SPAs.

The real winners are frontier labs, we pay them to write black box software and we pay them to rewrite the software.

Yeah, but why are we accepting that? Do we envision a scenario where code will be so black boxed that software creators will be pinching pennies and cutting corners so they can eek out a profit margin over the cost of development? It seems insane that we’ve willingly taken this route, the power when using LLMs is immense, but the power we’re giving up is all we had.

I saw a youtube comment a few months ago that has stuck in my head, can’t remember the exact wording but it was something along the lines of “I’ve never seen an industry so excited to cuck itself out of a good wage”.

This pattern is somewhat self-inflicted. There's very few reasons for software to change, IF it is properly developed. A reasonable life span should be in the order of decades, not years, or I bet now with AI it will be months. Do we really want a constantly changing software landscape that requires unlimited Internet connectivity to function? THat seems like such a waste of electricity and bandwidth, but as you say that has been the norm.

Agreed, but it's a very big if that I haven't seen in practice much. By January, 50% of major codebases that I've worked on will no longer be running.

Something that I'm working on for my day job is essentially 32 years old though (been through a couple of big upgrades along the way) and that's how it should be, unless the underlying technology would be better off being replaced (like a COBOL or Fortran codebase).

I think unfortunately companies are going to have to learn the extremely hard way what happens when your engineers no longer understand any of your code. The drive to eliminate institutional knowledge to create a perfectly replaceable worker was always dumb at best, and just leads to car crshes

I have used Agentic AI for both really sloppy vibe coded personal projects (big balls of mud with tons of integrations- huge HTML pages with tons of JS slop doing and NodeJs that I wouldn't ever care to look at) and well-engineered, carefully crafted systems for work and I haven't seen significant regressions in either group.

Reply on news.ycombinator.com