Sigh.
Please read this:
http://www.zephoria.org/thoughts/archives/2006/05/13/cluster...
If you're too busy to read it, a good summary applied to this situation would be that small-market browsers do indeed matter if they're being used by the sorts of critical early-adopter/well-connected people who make or break your service through their social connections. And remember that Safari is on every single iPhone out there, and so is the mobile browser used by lots and lots of those people.
I still don't feel that this even comes close to matching the massive impact that Firefox has. It is also popular among early adopters, and were Ogg to be made the standard and implemented on YouTube as the sole <video> codec, I think it would be possible to force Apple's hand on the issue.
The question is not "is X bigger than Y". You have to stop thinking in those terms if you want a successful service built on social networking effects. You simply cannot afford to lock people out, period.
"were Ogg to be made the standard and implemented on YouTube as the sole <video> codec, I think it would be possible to force Apple's hand on the issue."
No, it wouldn't, because YouTube is run by people smart enough to realize that punishing millions of users to try to stick it to one of your business partners is not a good plan. And even if they went Theora-only (which they're not going to do), they'd have to provide a fallback for IE, which would probably end up extended to Safari as well.
I'm not bothered about the actual format shipped, let the market sort that out. What I want is to be able to publish a webpage with a video and know that all standards compliant browsers will play that media.
Safari doesn't have to only use Ogg Theora, YouTube doesn't have to stop using flash.
Keeping the openness of the web means allowing content creators a path to create content unencumbered by license disputes IMO. They should go back and specify it for HTML5 standard compliance and let Safari fail compliance.
Your proposed remedy does not accomplish this. Putting language in the spec and "letting" browsers "fail compliance" leaves you in the same situation as not putting language in the spec.
Moreover, the efficiency concerns regarding Ogg will be solved in the future, so that shouldn't be a concern in the long run (which is what this standard is for).