"I believe the hard part of building software to be the specification, design, and testing of this conceptual construct, not the labor of representing it and testing the fidelity of the representation."

Fred Brools described it well in 1986 in his article "No Silver Bullet". Even by then, coding wasn't observed to be the main constraint with software delivery

Highly quality dumb TV which comes with a universal remote (so can control e.g. TV streamming dongles), take my money.

Want same for connected audio - dumb speakers that just connect to each other via wifi have BT etc

Great writing, and a good argument for turning the internet off 😅

Tests are the most important bit. With good tests, the code becomes more fungible, less important.

* (behaviour, not implementation)

"I believe the hard part of building software to be the specification, design, and testing of this conceptual construct, not the labor of representing it and testing the fidelity of the representation."

Manchester Met (MMU) have a good reputation for their degree apprenticeship

Hope its useful!

1993 – Steve McConnell shows late defects dominate cost and schedule

2000 – Joel Spolsky argues teams lose time by committing to decisions too early in code

2009 – “Typing is not the bottleneck” becomes a meme because it was already obvious

1949 – Maurice Wilkes realises most of his time is spent finding mistakes, not writing programs

1975 – Fred Brooks shows progress is limited by understanding, coordination, and rework

1979 – Tom DeMarco argues analysis matters because discovering misunderstandings late is ruinous

(cont..)

FWIW, having now scanned the paper, it's not saying GenAI is comparable to e.g. steam, it's saying it will need a power revolution like steam or electricity to sustain it

I've literally just written a piece on why GenAI is not like steam or electricity

blog.robbowley.net/2025/12/16/f...

No it isn't. Funnily enough Ive just written a piece on why it isn't the case:

blog.robbowley.net/2025/12/16/f...