I like the concept of allowing people to store data on some peers, and letting the market decide the rate at which the storage costs. This is in contrast to storing on chain or in the cloud, both of which are inefficient or undesired in the later case.
Redundancy becomes a major issue so I'd like to hear about this new idea of using delegates perhaps as redundant data servers incase peers shutdown and data "shards" need to be reconstructed.
Perhaps the best first step would be to use all delegates to offer their VPS storage capacities (which they probaly arent using unless its the chain folder) to offer free storage services for something like maidsafe or storj... then people can store data via delegate "shards" and it will be limited to the storage capacity of the delegates. If market desires more data then more delegates can be voted in with bigger storage capacity.. creating a secondary market for delegates other than what we have now. Now we have free storage which delegates are paid for, funded by the blockchain and an added incentive for new delegates to stand up with bigger storage facilities, making a profit aslong as their storage costs < delegate pay. We need to leverage the server machines as much as possible and create value added services such as this. Perhaps a revenue generation model would be to buy more space than is available or add extra redundancy for X amount of bts per day or something.
These are some general ideas of what most projects are doing with a touch of DPOS.
The market certainly would have a roll as you mentioned, but all the current models that have attempted it in their current methods have been received with limited/little adoption.
There are other issues to consider in regards to privacy and encryption that I do not believe would be advantageous to the bitshares blockchain for what it is designed to do (ie, be a financial platform). Graphene however can be adapted in its own side chain (which will share drop on bts of course) to enable proper standards that make it more than just a side show curiosity in storage.
There is a fair amount of overhead, major bandwidth costs, and performance issues and security to consider. Our implementation will be with the objective of mass adoption that can serve both consumer and enterprise. So with regards to redundancy you cannot create a one size fit all system. It needs to be able to scale with both standards and costs. Joe six pack with his home movies library is fine with mirrored redundancy, while Medical Widgets Inc. requires HIPPA compliance for their storage with DR... list of scenarios goes on... as do regulatory requirements in privacy. The costs of redundancy should be dynamic.
We already had plans to create bridging tools to make some of what you suggested to interface with bitshares easier/seemless.
There is ultimately the opportunity for the bitshares blockchain to live in this new decentralized network itself eventually if delegates choose to make the move there. There will be many many very good reasons too I assure you.