What Google v. H.264 means, in summary
In the short-to-medium term
As a user: nothing changes. Almost everyone will stick with Flash as a prerequisite for delivering video content, irrespective of browser capabilities, with the limited exception of mobile devices.
As a web developer: the balance is still somewhat tipped in favour of H.264. If you can support WebM, try to, but it’s going to be a lot of hassle for a lot of people. The minimal-effort support-as-much-as-possible approach remains “H.264 with Flash fallback”.
This is decidedly murky. If we’re still in this same position ten years hence, the only sound you’ll be hearing will be people weeping and Adobe laughing all the way to the bank. There are plenty of things Flash is good for, but a single proprietary delivery mechanism for all Web video lacks longevity.
Everything boils down to two things: patents and silicon.
The patent situation is dubious across the board: on the MPEG side, people are concerned about fees (even if the potential hikes are very limited in scale), and for the alternatives (be it Theora, WebM or even Dirac), it’s pretty much uncharted waters. There will be patent trolls whichever you choose. The logic of Apple and co., is that of it being more likely to be troublesome with the “patent-free” codecs than otherwise. They simply don’t want to have to go to court to defend WebM when they get sued (and they would; they get sued for something new every day), however dubious the action.
The silicon issue is even more important to Apple, though: without hardware decoding, you can throw that 10-hour battery life on the iPad out of the window for starters. Mozilla is barely affected by this; Google is focussed on the software, rather than hardware, side of things (it dips a toe in the water, but it’s not really a primary objective of Google to shift lots of handsets, for example). Mozilla would struggle to care.
There are two ways this could go. The MPEG-LA could decide that all software decoding (and possibly encoding) of H.264 was gratis, in some fashion that would satisfy Opera, Google and Mozilla. You pay royalties on hardware implementations (which have a tangible cost in any case), but not on software. How likely is this? Who knows.
The alternative is that WebM — or whatever its successor turns out to be — is implemented in hardware as soon as possible and integrated into the chipsets used by modern handsets and other mobile devices. This still leaves the patent troll concerns, but at least then it’s a matter of a calculated risk rather than completely infeasible as it is without the hardware.
Right now, it’s not clear what will happen at all, really. All I do know is that this isn’t really a step forward. It may bring about one as an eventual consequence, and so could be considered a long-term “good”, but right now, all it’s done is complicate matters.