Every time I've let somebody else's tools write HTML for me, it's caused me grief. Either because the tool author had an agenda that was at odds with my needs (Microsoft), because of bugs in the tool (everyone), or because the problem set shifted since the tool was written as browsers and techniques changed (everyone). Likewise, I'm wary of entertwining HTML output with the rest of the program the way Arc does because it makes it too much work to change things down the line.
It would be interesting to write a templating system that separates HTML output without making Arc programs take much longer to write. I suspect that it could be done, but it's not entirely clear to me what shape it would take. If done right it would give users the option to leverage Arc's existing HTML output functions where desired. I'll have to play with it.
A great, usable prototype :)
I think you're confusing types of software with organizations that sponsor them.
Arc certainly wasn't designed for use in the sort of environment where programs are 10% functionality and 90% the embodiment of the random bureaucratic processes of the organization paying for them. More for the opposite end of the spectrum, in fact. That's where you need power.
That said, if you care to make your challenge a precise one, it's likely I (or someone else) would try implementing it in Arc, if only out of curiosity.