Hint: valid bitcoin addresses begin with "1"
Are you sure?
How do you explain this address then?
https://blockchain.info/address/3A5XQWZPzG4GrpVeYFaDBZ1s4sXkYB69Wf
But don't let a fact get in the way of a strongly held belief
I'm not understanding how the 3 addresses aren't valid...their relation to op_eval, multisig, hashes of scripts, etc....but by this link it appears to me that a 3 address can both store Bitcoins and transfer them to a 1 address: https://blockchain.info/address/3DLCRW4v2zcMoWfk8HH95JvMtQgCKhgKYt
Now, please, I don't want to be responsible for killing anyone...if you feel you might rupture an intracranial aneurysm, please stop reading and go see an interventional neuroradiologist or neurosurgeon.
OK so differentiating between "valid" and "usable" is a bit pedantic IMHO but let's go ahead and change the language we're using anyway. Whether or not they're valid, they're not USABLE. Let's see what happens if we try to send coins to 3DLCRW4v2zcMoWfk8HH95JvMtQgCKhgKYt with some various clients/services:
Blockchain.info wallet:
MtGox withdrawal:
BitcoinSpinner Android Client:
Bitcoin Wallet for Android by Andreas Schildbach:
I haven't had a chance to test with the Satoshi client yet since this is a work PC and I don't (can't) have Bitcoin installed on it, will update when I get home. At the very least I can say at this moment that none of the web-based or Android-based clients support this Bitcoin address version yet, which makes it largely unusable. I suspect that the 0.6.3 Satoshi client doesn't support them without special configuration either. Even if it does, none of the exchanges, eWallets or other services will send to such an address so it would only be usable in the one direction.
I suspect that all the 3* addresses in the blockchain folks are linking to as "proof" are live tests of OP_EVAL using experimental forks.