Since OP_Return was reduced to 40 bytes right before the release of .9, Counterparty is currently using Multi-sig (which was supposed to be deprecated) which is undesirable for reasons aforementioned by Peter Todd. On the other hand I do not think I've seen anyone present a line of technical reasoning as to why OP_Return 40 is alright while OP_Return 80 is undesirable.
My understanding is that Counterparty is functioning, right now, using Bitcoin as a transport layer. In order to do so, it must be using existing, accepted features of Bitcoin. It might not be the most ideal or efficient use of Bitcoin, and it may not be the full implementation of Counterparty, but it is functioning.
Wait for Zerocoin to boot BTC out.
You prefer an altcoin that requires centralised trust, over a decentralised Bitcoin?

Decentralized Bitcoin, which requires getting the permission of ~15 powerful individuals in order to have transactions confirmed; transactions which fully comply with the existing Bitcoin specifications and implementations?
Nobody is forcing you to run the reference client.
Fork it. Please.
I am providing material and other support to several groups (in particular, the BitShares folks) who are in the business of making Bitcoin irrelevant. Getting rid of mining completely, as it inevitably results in centralization of power, no matter the algorithm chosen.... not to mention a huge waste of resources. Of course, Bitcoin has a pretty strong first-mover advantage, and Counterparty is an interesting use of it. (Just in case it is not perfectly clear, I am NOT involved with Counterparty in any way, other than holding some XCP that I bought recently. And I hold BTC as well, and mined it until recently (just sold my last miner today), although I'm seriously reconsidering my BTC holdings due to revelations in this thread.)
And no, the Counterparty writeup is not an "existing Bitcoin specification".
No, I meant that Counterparty is already running, using existing features of Bitcoin. It is a
fait accompli.
For that claim, the developers should collaborate with the existing Bitcoin development community to formally define a BIP.
So that it could be sat on and debated by committee for months? Inventors don't wait for approval, they get out there and do it. Did Satoshi run it by the Federal Reserve before releasing Bitcoin? Did Linus Torvalds consult with Richard Stallman before he released Linux 0.01? Counterparty already has code written and functioning. They have achieved the first functional distributed exchange. An accomplishment not quite on par with the invention of Bitcoin, but definitely one of the most important developments for crypto-currencies since Bitcoin.
I'm sure that the Freimarket guys have some good ideas, but a "let's put out a white paper describing every function of our software and then take several years to develop it" philosophy just ain't gonna work. Their stuff is going to be so late to the party, it won't even matter how good it is because several competing systems will already have been running, debugged, and iterated multiple times.