There is legitimate reason to reject this on the basis of nondeterminism. While we are surely obliged to "hold our noses" and support "SELECT *", as: A) The SQL standard obliges us, and B) People already use it a lot,
Neither of those factors hold true for the EXCLUDING notion. So all things are decidedly not equal. By all means I find it an interesting feature, but that shouldn't be mistaken for necessarily being a desirable feature. I don't think I wish it. We're telling our developers not to use "select *", and I don't think having "select * except " would change that policy, beyond requiring us to waste time explaining : "No, we're not changing policy. The fact that PGDG added this to 9.2 does *not* imply our policy was wrong."