Hi CuriousCarl,
Bitcoin transactions are actually chained together in a more general way than simply "address to address". You might find
this overview informative. While ordinary spend-to-address transactions are created such that only the owner of an address[1] can redeem the coins. However, it's possible to create transactions whose outputs can be redeemed given other conditions: some trivial ones are "anybody can spend" and "nobody can spend". What you're seeing here is an example of the latter. Such transactions are used because it is possible to embed other, unrelated data, into such transactions. This is antisocial behaviour because it consumes Bitcoin users' bandwidth and storage (which presumably they are donating for the sake of Bitcoin, not whatever unrelated thing is being stored), but it's technically possible so people do it.
Addresses aren't directly part of the Bitcoin protocol; they're just one way of using
Bitcoin's scriptable signature system to manage asset transfers. blockchain.info uses a broken model of what transactions and addresses are, and therefore gets confused quite easily. Generally if you encounter a weird transaction, checking what bc.i thinks of it is exactly the wrong thing to do.
[1] By this I mean somebody in possession of a corresponding private key; this often doesn't really correspond to "ownership" on the user level, for example if your coins are stored at an exchange. Then the exchange might own the address even though you own the coins. It is generally true that addresses are a pretty low-level thing and don't correspond well to user expectations, so in a more user-friendly system users would not ever see addresses at all. There is progress in this direction but it's slow.
Andrew
