The poster child for proposals with a very large scope would be the biology proposal.
Its audience is described as:
Proposed Q&A site for biologists, including biochemists, molecular biologists, cell biologists, physiologists, and ecologists.
That's a wide range of experts, and it's missing some like microbiologists and geneticists.
Don't take my word for it, just look at Wikipedia's list of branches of biology. While all of those fields fall under the biology umbrella, an evolutionary biologist is not trained to answers to questions about if deer sleep standing up or the differences between various types of kidney surgery. You'll need a zoologist or (I think) a physiologist to answer those questions.
Each of those field could probably make a successful Stack Exchange site on their own. That's how large the Biology proposal's scope is.
Now, that's not necessarily a bad thing. Stack Overflow demonstrated that a site regrouping all experts from the same field under one room could be successful. However, you can't launch a site with so many different kinds of experts and hope it does well with the same initial population as, says, a site on the English language where an expert could answer most questions on the site.
So, what I propose is that a proposal's minimum commitment score and minimum committed users could be changed manually depending on a site's scope. There would be a default value, of course, but it should be modifiable if deemed desirable or necessary.
This way, proposals with very large scope could launch with communities of the appropriate size.