gerald sussman once said that programming is no longer constructive, building truths from truths, instead investigative, identifying probable behavior from unreliable libraries. anyway, llms,,,

https://wingolog.org/archives/2009/03/24/international-lisp-conference-day-two

@wingo Yes. I remember the days when if I *really* wanted to know what the computer was doing, because what the logic analyser was reporting didn't make sense, I could use an oscilloscope. We knew every byte of code in the machine and we had a reasonable idea what the hardware was.

I reckon it was things like on-chip caches that broke this way of doing software - the CPU was now doing things that weren't visible on the external bus.

@TimWardCam @wingo whenever i say "I *really* need to know what the computer is doing", I am never, ever talking about behavior that could be sensibly investigated with an oscilloscope.
@PaulDavisTheFirst @wingo Not these days, no. I'm talking about prototype hardware in the 1980s.