When building becomes cheaper than specifying, everything changes

I'm the author. I've spent the last couple of years building functional products with AI tools instead of writing specs, and the results changed how I think about product development. The short version: I built a satellite monitoring module in 9 days that would have taken an engineering team 8-10 weeks. Not production quality, but functional enough for real users to test and give honest feedback. That feedback was better than months of pre-build research would have produced. The experience led me to develop two operational frameworks, SIGNAL and ATLAS, for rapid product discovery and construction. Both are running in production with real teams and real revenue. The article describes the economic argument underneath: the cost of specification (meetings, documents, reviews, estimation) is now often higher than the cost of just building a testable version. That inverts the logic of how we've organized product development for 30 years. Happy to answer questions about the practical side, what the tools actually produce, where the model breaks down, and what engineers think about non-engineers building software.

I also feel things changed a lot. We’re now getting vibe coded prototypes from management - thousands of lines of spaghetti that we must make production ready “by yesterday”. After all the difficult part has been accomplished already. Oh, and this is a list of additional features that should be very easy to add while you’re at it.

it used to be that coding was veeery tough and tedious thing, so specifications had to be really well pre-chewed, down to the bone. Think 60ies and 70'ies .

Each decade after that, less and less of the specification is being ever written, and it is/was left to the programmer to finish it as they seem fit, and more and more programmers became (business-) domain-experts. Just see job ads for last (20?+) years - half is software-stuff, half is domain-knowledge required.

so.. do you think there's some reset happening/pending? So there is more specification being written? Or is it faster prototype-and-throw-away loop, and at the end the spec is "make it like this but prdocution grade" - instead of (again) actual proper specification with all the whys/whats/hows in it ?