OP_RETURN can be used to store references to external resources, say for example very long meta transactions that don't fit into 40 byte, an asset contract or whatever. The storage pointed to could be a website which lists something like (for the sake of an example), a side chain or a P2P structure like a DHT. To my knowledge this was discussed several times, but never tested on a broader scale.
That's missing the point, then you're back to the question of how do we securely store data in an accessible, decentralized way, and it's onto side-chains, which come with their own problems and which no one has coded working functionality for - that Counterparty has here and now, and this is for 40 bytes of economy with only pseudo-speculative importance, when as Canton said gambling transactions are taking up 5gb to Counterparty's 5 mb.
I do agree the status of these things may change, but as I wrote, I don't see why speculations about the future should regulate implementation that a freemarket is capable of doing on it's own.
Why does the data need to be stored in a "decentralized way?" Why should it matter if someone tries to store invalid data or remove valid data? If I made a trade and the hash of that deal is saved in the blockchain, all I need to prove I have what I traded for is my own copy of the trade which can be validated with the hash in the blockchain. It's like how in Bitcoin, seven lying nodes can be overruled by one honest node. The work is proven and it doesn't matter how many peers disagree.