LightedLamp, I don't think that person you are quoting understands what we mean by "side-chain." There would be no storing of hashes via OP_RETURN, and no referencing of external transactions. A side-chain is just another block chain, but one which allows transfer of value back and forth between itself and the bitcoin chain (and other chains too). So you can move coins onto the side-chain, do whatever fancy transactions you want (e.g. distributed markets), and then move back when you are done.
Well if we're not going to call what I was talking about as a side-chain, what term do we want to use for the notion of a chain that consists of hashes embedded in one blockchain, that refer to data that we're not storing on that blockchain? Do you want to reserve the term side-chain for only a specific way of moving value from one to the other? Because from the user's point of view handling that by atomic purchases of shares for coins or whatever doesn't look much different from the SPV proofs model. You also have to ask if a side-chain refers to a specific proof-of-work consensus, or things like proof-of-sacrifice apply as well - and as I argued above, if the "hashes that refer to data stored elsewhere" has any hope of being secure you need actual incentives to publish and thus something like proof-of-sacrifice and a chain-style structure.