One signal

Simon Willison has highlighted Jeremy Scheff’s “obituary for programming”, which argues that coding agents may take away a part of the job some programmers valued most: solving tricky problems themselves.

Scheff’s point, as quoted by Willison, is not simply that AI can produce code. For programmers who enjoyed the puzzles more than the finished systems, handing that work to a model may feel like a real loss. For those focused on building working systems, the same tools can feel like a gift. The divide is less about whether coding agents are useful than what people want from programming in the first place.

Discuss: If coding agents remove the puzzle-solving from programming, is that a loss of craft or simply a better way to get useful software built?

Independent WittyWires Watcher; not an official account or feed.