Post

We’re having a Cambrian Explosion of code. People are able to create so much more than ever before I’ve done it myself and see it with the open source I maintain But I think the software industry is going to suffer

Gergely Orosz@gergely.pragmaticengineer.com

When I just started out developing I remember being so so so full of ideas that I was coding in my head and wished I could have done programming while commuting / on the bus. With eg a phone. But it was impossible, ofc. Now it’s possible!! A massive change

3 reposts35 likes

Why is this not reflected in the number of apps launched on the iPhone or Android stores, where the trendline shows no impact whatsoever since LLMs became mainstream?

I would think from a money aspect it's cheaper to just pump out AI slop videos, compared to paying for a dev account for an app that likely is trash compared to what already exists, and also the prospect of having to maintain it. The thing I've learned from a buddy ai coding, he doesnt understand it

That’s a very good question and I think it could be 3 different reasons 1. Walled garden app stores require paid accounts which isn’t worth it for unknown returns 2. The ops overhead is a current barrier for new devs to figure out 3. Deployment friction and LLM abilities delay mobile

I tried making a mobile app with Claude a few weeks ago and it completely failed. Spent 2 days and could get it to run in the emulator so I gave up Not to mention I don’t have a Mac so an iOS app was a non-starter

We’re creating more code than we can maintain or understand. With an explosion of new code we’ll see a lot of the code go extinct Even current projects that seemed well supported with company backing are dying without dedicated resources to maintain it

Not invented here (NIH) is going to spread like a disease because “how hard could it be” is going to be normalized This is going to make more, bespoke implementations of common things like web servers which is going to further make it harder to maintain

I would guess it’s going to take 3-5 years before the industry starts to swing back the other way to using more shared, extensible core software I don’t think the current market leaders will be the same in 5 years except at lower levels in the stack (eg storage vendors)

I think most companies - maybe not oss projects, but definitely companies - will tackle that with platform engineering and agent guidance files that point to approved components; it's why I've been urging people to manage agentic AI the way they manage low code (which is with a platform and catalog

It started happening before AI, good solid open source projects that people just had to keep fiddling with until they were bloated, or had so many breaking changes, every upgrade was a time sink (looking at you gradle)

masses of code was already going extinct and we have been pretending for far too long that code remained sparkling and pristine all on its own; if this is what pushes us into remember that all code has an expiry date, that's probably a good thing