The Keynote presentation by Hal Abelson and Gerald Sussman at the fourteenth RacketCon is now available at https://youtu.be/_2qXIDO-cWw
Any plans to support emacs?
Seriously: It was interesting to hear them discuss (1) the pendulum swinging between "science" and "industry", and (2) using natural language for programming.
IMHO: The latter is an example of the former -- "industry" swinging for the fences.
Natural language might not be optimal for humans to "write code". But it's already how managers get software projects done -- natural language conversations with project managers and engineers. Their goal now is "just" to eliminate most of the other humans.
(There is a certain breed of manager who is indignant about unseen remote workers... only if they're human employees. I'm not a psychologist, so I can't diagnose sociopathy, but I know how to duck-type. )
Don't miss the special surprise at the end (around 1:05:05) . (thanks to 'iguanathesecond' for the reminder)
Anecdote about natural language and duck-typing: for years, I was under the impression, having never read-up about it but being aware of the term and its use in context, that duck-typing meant being able to duck under something.
That is, if some object X
can duck under a bar I set for it (has property Y
, e.g.), it must be the thing I am looking for.
So, yeah...
That's great!
Let's go to the LLM prompt and ask it to summarize the thread so far; it says:
"If objects X, Y, Z walk into a bar, and Z quacks in the bar, then Z must be a duck whose cousin is a sociopath seagull manager."