Evolutionary, not Revolutionary

When AI can write software an order of magnitude faster, its tempting to be revolutionary instead of evolutionary.

With the rapid development of software in the agentic era, open source has seen a huge uplift – including submissions from many people who may not have open-sourced before.

A side effect that I’m not a huge fan however, is this idea that many of efforts that used to have at least several months (or even years) of usefulness – can now be considered obsolete very quickly – often within a matter of days.

When AI can write software an order of magnitude faster, its tempting to be revolutionary instead of evolutionary.

I’m not convinced this is good for the industry. When appropriate you should “kill your babies” and it was never a good idea to get too attached to your code.

However, this ability to produce software at such an extreme pace doesn’t inherently mean AI-slop.

Yes in many cases that may be true, but there are certainly exceptions.

I’ve seen plenty of great software created in recently months, that’s almost immediately considered irrelevant by social media. This undermines the work of great engineers and in my opinion, devalues the work being done and removes the chance for iteration and hardening required in order to create robust, reliable software.