the new math: old battle of the lovees was: PDP1 3600

We agree on that.

It's the structure of natural languages. One will always find exceptions.

During the OS wars for Unix in the 80s I had another personal thread. John Backus was trying to convince people to program in FP, but he never had the full support of IBM. He was bitter. He indulged me emailing him questions of Fortran history and trying to get him to give a talk at work on FP (he would not talk about Fortran history), and later FL which was partially implemented in 4.2BSD as experimental software.

We will never how if functional languages could have really helped software more.

People went overboard like Rachel Carson's Silent Spring. They over generalized.

Yes.

posts everywhere, that's one problem of distributed systems.

Basically programmers map everything to 1-D like chronological languages. You have to be careful to avoid being like certain LISP fanatics (some friends). In a 2D case (did I write 2-D? andling non alphanumerics can be problems), in imperative languages like Fortran and C you get column-row major order in mapping integer multiplies and adds for indexing. So on good measurement environments, nested array indexing has a measureable performance cost array (i,j) can cost more in time and operations if an outer loop (or recursion) loops i or j first. F or C, they are both problems. This should never have to be. Duncan Lawrie did a PhD thesis on how to architect a memory where order doesn't matter, but compatibility rules the day. Generalize to higher dimensions (more integer *s and +s).

In turn when windowing systems like X appeared, the guys at MIT, DEC, and IBM thought graphics systems were 2-D. They had text and business graphics on the brain, not full 3-D graphics on the brain. They didn't care really about the features of their languages and systems.

You have to beware of the AI arrogance (and mix in the arrogance of the other fields as well).

Others brought up OSes, PLs, and networks. In other threads. Back in the 80s, a really bright guy from the early ARPA era from SRI named privates Watson, now at LLNL also saw that the problems of OS were merging with PLs and nets. Someone pointed out that PL's don't involve queueing. Well, newer compilers in fact do. Scheduling is what VLIWs, and parallelism and optimization are all about.

So every one is reinventing various part of each others fields again. It can be dismal, but we have to remix various ratios of different ideas again and again to get the right (for the moment) combination.

Any ways I have to pack.

--