cost/benefit analysis of new URI scheme vs overloading "host name"

I looked in the draft, and what I'm missing is the
cost/benefit analysis of this proposal against the
alternatives.


http+srv://hostname.tld/blah/blah

vs.

http://hostname.tld.SRV/blah/blah

vs.

http://[SRV:hostname.tld]/blah/blah

or 

http://SRV*hostname.tld/blah/blah

or some such. It seems like if you're going to introduce a new way of discovering a route to a host, overloading the host field would be better than overloading the scheme/protocol.