After coding parsers, my English became a little more refined (or possibly, just my interpretation of my grammatical decisions become a little more refined): I had said "one can reuse code as a platform or library", but realized that that could imply that I thought those two were different names for the same thing, and this ambiguity could be avoided by "one can reuse code as a platform or a library".
This was precisely the grammatical issue that came up in my parsing code (actually for fields that can hold a choice of types). Of course it's no surprise - Chomsky is a linguist - but such an absolutely direct connection between programming and speaking was striking to me.
An interesting point in the article is that GUI's have become synonymous with constrained choice. Constrained choice can be a good and useful thing, but a graphical representation is potentially much freer than text (consider an image of text in photoshop). I think the closest we have to this in common use is hierarchical tree views (as in vim's NERDTree, or java's JTree, or in windows directories), but they are still very constrained. We could have free-floating graphs with arcs and nodes everywhere; or handwritten algebra; or even active diagrams! (not sure what that would be!). But we don't. I think it's partly because the mouse is great for pointing; not so great for drawing. You really need a pen tablet (which many artists use), or.... a multi-touch screen. People, I think there's a deep opportunity here to change everything.
I've long been curious about Perl's growth... did it in fact grow more strongly in NT than unix? It was the best linguistic tool around with no real competition at all, unlike in unix (BTW Wall: also a linguist). Did this then form a stronghold, from which to launch back into Unix? It's an appealingly logical adoption path.