BIP 10 could use some improvement.  I created it for offline and multi-sig tx but there was no reception to it because no one was using offline or multi-sig tx at the time except for Armory (which only currently implements offline tx).  So I made something that fit my needs, and it has served its purpose well for me. But I also think it could be expanded and improved before there is wider adoption of it.  It's a little clunky and not very rigorous.

Elements of it that I'd really like to keep:

(1) Some aspects of human-readability -- even if regular users will never look at it, it should be possible for advanced users to manually copy&paste the data around and see what's going on in the transaction and what signatures are present.  I'm thinking of super-high-security situations where manual handling of such data may even be the norm.

(2) Should be compact -- I took the concept of ASCII-armoring from PGP/GPG, because, for the reason above, it's much easier and cleaner to view/select when copied inline.  If a random user accidentally runs across it, it will partially self-identify itself

(3) Includes all previous transactions so the device can verify transaction inputs without the blockchain.

Things that could be added:

-- It needs a BIP16 script entry (this was created for vanilla multi-sig before BIP 16 was created)

-- Comment lines

-- Version number

-- Use base58/64 encoding

-- Rigorous formatting spec

-- Binary representation

-- A better name than "Tx Distribution Proposal"

I'll be releasing the Beta version of Armory soon, and after that, I'll probably be thinking about a multi-signature support interface.  That would be a good time for me to tie in a better version of BIP 10 -- one that is compatible with other clients implementing the same thing.