Cardinal Quest actually started out as an html5 game, and
I've written before about why i switched to flash.
Now, keep in mind that this article was written almost a year ago and the situation has improved some in the meanwhile, but IMO html5 isn't up to par yet, but at least there are some better tools for it now compared to before.
Good. Anyone doing it now will have a massive head start when propping up of the foundation settles.
As someone who jumped on the html5 bandwagon in April of last year (1.5 years ago) & still dabbles in it once in a while I disagree - sure, coding directly in js and to the bare canvas/audio api you'll learn some platform specifics.
But if you use one of the newfangled libraries/frameworks that are coming out in droves (and you
should) it takes all of a week (
tops) to learn the ropes if you are a competent programmer (they are about the same level of abstraction as programming in as3/flash using flixel or flashpunk, only the results you get are not yet as good), especially if you've done (non-game) web development with html/js before.
This is
not the same as learning obj-c in 2008 so you could program ios - the skill set from making flash games transfers pretty smoothly.
Well, I guess it is if you've been making mac apps with cocoa before anyway, but honestly who was doing that before ios? On the other hand making games with as3/flash is pretty much the most common situation here (next to gamemaker).
At any rate as a programmer you should regardless do your best to expand your skill set & not get too comfy with any single tech - they all come and go & you should either be able to adept or start learning some (any, even html5) new languages/platforms now so that you develop the skills to adapt quickly when your current platform of choice's inevitable demise arrives.