I think that Greg has done an excellent write-up of the positions.
Position 1 was originally taken by Microsoft Caller ID.
Position 2 was originally taken by SPF.
I see position 3 as the best of both worlds, and heartily support it.
I see position 4 as the worst of both worlds.
I believe that "syntax" extensibility will be crucial in the future.
I believe that "feature" extensibility has a positive but small value,
but probably isn't worth the cost.
Just to be clear, the current compromise draft takes position 3 (parsers
support both SPF and XML, domain owners publish one or the other). It
also enables "syntax" extensibility. While it doesn't go out of its way
to support "feature" extensibility -- there's no ALT mechanism, for
example -- it doesn't completely rule it out, either.
-- jimbo