This is definitely "not broken". I really don't find this a problem, at all. And for those crying about the sender having to pay to include the larger transactions, come on, I bet you can include transactions to multi-sig addresses paying an irrelevant fee. And probably the only reason they wouldn't be included for free is due to the standard fee policy.*
I'm not saying BIP 16/17 are not interesting. Only that there's no reason to rush or stress over it.
You people are really "making a storm in a glass of water".
* There you go, an example of something I find more important than these BIPs: removing the standard fee policy. This would need anti-spam protection. I remember some development was on the way concerning this, but I don't know if it's finished...