So you’re telling me whether I still have my job in 5 years will come down to whether the increased demand for software outpaces the efficiencies in producing it?

The reactions to Devin seem to fall in either: 1) devs are toast now and we’ll see large workforce reductions 2) No-code has been around for 10+ years and devs are still here.

What we did through better abstraction and tooling is over time genericize the underlying skillset where the ability to solve problems trumps knowledge of specialist trade tools. The culmination of this is the modern fullstack engineer.

The bespoke secrets of TLS, SHA256, PostgreSQL sharding is not what you pay web devs for. You pay them for the skillset and experience to solve problems end-to-end and bootstrap their way into new technical areas leveraging the dev collective intelligence

Devin’s core intelligence is a GPT-4 loop. We know the limits of this approach already. Even devs shudder at the idea of inheriting massive amounts of code from other devs, good luck to non-devs doing that with whatever Devin gives you. 60% done isn’t something you can ship.

Dev collective intelligence is already as good as googling for information and finding it, now it’s in your IDE. The productivity gains from having an experienced dev with a copilot will far outdo “agentic” systems for a while.

From the history of computing, it’s been broadly true that increased efficiency has created more demand for software development which has consistently employed more people. In the 80s, you had one “programmer” staffed by a vast team of other specialists they needed to do work.