If we do it and then it breaks, that hurts every old thread that relied on it. I'm sure you of all people will understand why broken content would be less than ideal. Part of our success here is that we're generally trying to keep things breakage-free.
An alternative option we could consider is not attempting to scrape the data from links, but rather modifying the [youtube] bbc tag*. For example, we could add a "start" parameter there, similarly to how, say, an [img] can have a "width".
We'll find a way to make it work. I'd just rather approach it slowly than rush into something that might break things in the long run.
* - Yeah, I realise most people aren't aware of this, but e.g. [youtube]dQw4w9WgXcQ[/youtube] works.