a. I'm using the blockchain to make an objective indication as to whether an offer still exists, whether it's been satisfied, by whom, and when. My method automatically cancels the order when filled, and picks a winner if two people accept it at the same time, automatically and objectively.
This is true, and I think it has its advantages.
b. My offer is much easier to accept. By signifying that sending bitcoins accepts my order, one does not need to bother with getting in contact with me and hoping I am able to respond in a timeframe that works for them. They don't need to do anything other than determining that I'm trustworthy enough to perform as agreed, and then send some bitcoins.
This is true, it does resolve that part of administrative problems, but it creates a whole new set: someone needs to keep track of all the offers from all the parties, eventually this would have to somehow be built into a database, which rather than scouring the web for signed documents would more reasonably be pushed into by the very party offering. Which would make it a sort of OTC-OB.
c. I believe that writing an actual agreement using language with legal meaning and signing it myself has greater enforceability than putting an unsigned line in the OTC order book.
Iirc you can only put lines in the orderbook if you're id'd with gribble, which means everything in there is actually signed.