Your point? Using inheritance the way he's suggesting is a rookie mistake.
Either that, or I'm completely misunderstanding him. He can't be talking about hating composition, though, can he? He uses it in his own Erlang examples!
I don't think Joe ever claimed to be a great programmer. Erlang looks as odd as it does because it was hacked up in Prolog, which few compiler engineers would choose as a starting point.
I think Erlang is interesting precisely because it was created by someone with a little distance.
I take it you have never written anything significant in Erlang?
Try writing something in erlang and I bet you'll see the real power of pattern matching, supervision trees, and the whole message passing infrastructure. If the people who wrote Erlang weren't good programmers they certainly got lucky.
Jokes away - the "really great programmers" for me are people who can do extremely complex things very simple.
For me, Armstrong is somewhere near Peter Norvig. Both made me change my point of view, some complex things became simple (and some simple things became not so simple).
"Erlang was hacked up in Prolog" - it's the same as "Java was hacked in C". First implementation of Erlang was written in Prolog, first implementation of Java was written in C.
No; because C is a fine implementation language. It's low-level enough to be efficient without a lot of moving parts, there's very little magic.
For me, Armstrong got interested in the idea of building reliable systems, went off and did a PhD on the topic, and then created a prototype implementation of the concepts developed therein in Prolog - a declarative language essentially built around combinatorial search with pruning. In other words, he's a fine high-level thinker, and a very competent wielder of Prolog; but I believe I've read elsewhere that it was Mike Williams who rewrote the VM interpreter in C, and all the low-level imperative stuff like GC.
The original author is Joe Armstrong (http://www.sics.se/~joe/), the creator of Erlang.