To play on a level field with native apps? No. Wrong discussion! What about for the same sort of tinkerer/hobbyist who wrote Hypercard stacks? Seems to me, with a little hosting and a few tools, Javascript would be dandy!
(And as for commercial apps on Hypercard -- hey, you're free to try. But I'm not talking about targeting those folks.)
My understanding is that there are entire classes of programs where "let's put it on a website" is not a viable method. You can write many types of apps that way, certainly, but you pretty much give up anything so much as pretending to be performance. Additionally, it prevents you from doing anything particularly low-level - if you were desperate enough, you could always set up a web interface to gcc or something, but that won't help you if you want to hack on a device driver or something. Systems programmers are people too, you know. ;)
But if you recall what Hypercard stacks were like, then putting it on a website that's specially tailored to make it easy, combined with a local app to make things more seamless, might actually be a money making proposition.
Put the hacking open/closed politics aside. What do you think about the idea as a business?
But the difference is that HTML5 games are games, and demos from 15 years ago were demos. Writing an entire app like you write a demo is quite difficult; hand-optimized assembly takes a lot longer to write than Java. The improvement over the last 15 years is in the programming language tools and runtimes, not the graphics that the programs produce.
Yeah, it's exciting that Quake2 runs in Chrome/Safari, it really is. But in 95/96 I was participating in demo-scenes, with graphics that weren't accelerated by the GPU, in 386 real-mode, and it was a lot easier and a lot more exciting (demos that got distributed by a local PC magazine, having as target more people than the people that actually tried that HTML5 Quake2 demo).
So welcome to 1997.