Is there any ideas for a file storage contract?

prosterproster CETMember Posts: 2

I'm thinking of a scheme where B gets paid to store a file* F for A until at least a time T.

For this to work, I'm thinking that it is needed for incentives for B to..
1) actually store the file until time T.
2) transfer the file to A on request.

I have a sketch for this, but my question is whether it is possible to remove the trusted third person from the scheme?

For 1), I'm thinking that a secret random nonce N is generated by a third party, C, and V=HASH(N+F) is calculated by C**. At time T, C reveals N to the world, and B can prove that he kept the deal to C by calculating V.

I haven't though through 2) yet, but I'm thinking it can be possible for A to initiate some kind of escrowed buy of F from B.

I hope this forum is the right place do discuss this or similar schemes.

Best regards,

*) Or rather, a forward erasure coded encrypted N-of-M chunk of a file.
**) This is an unfortunate significant overhead, as F must now be transferred from A to C.

Best Answer

  • nejucomonejucomo Posts: 40
    Accepted Answer
    Here's a thought on incentives:

    Sally the storage provider will *store* a gigabyte for a smallish fee, and will *retrieve* a stored gigabyte for a sizable fee. Perhaps the retrieval fee raises over time on some curve to anticipate future costs/profit.

    Sally's incentivized not to cheat, because the retrieval fee is where the profit is. The storage fee is more of a spam prevention.

    This (simplistic) incentive structure assumes Sally is being used as a backup service, and since the main payment only happens on retrieval rather than periodically, the revenue model is a bit more like (inverted) insurance rather than a cloud service provider.

    Maybe this incentive structure leads to simpler systems, without a need for tricky proofs-of-storage protocols, although I don't see why paying Sally a small fee to prove correct storage periodically wouldn't hurt, and it would probably help backup customers sleep better.

    FYI, I'm an occasional contributor and fan of the Tahoe-LAFS storage system. There have been discussions about both proof-of-storage and integrating payment systems into Tahoe-LAFS over the years, which you might find on mailing list archives.


  • JasperJasper Eindhoven, the NetherlandsMember Posts: 514 ✭✭✭
    Two days and no-one mentions there is a dropbox example in the whitepage. There is code for it, for instance here(but i am not sure if that is entirely up to date) It works by asking random paths in the merkele tree.(leading to a particular part of the data) And(apparently) by just paying to get the data back bit by bit at the end.

    There is also maidsafe and other distributed hash table approaches. I am not sure what state this in, but i think it is an important part of web 3.0 as Gavin Wood forsees. I think such a future probably indeed does require that serving and holding data for other pays, preferably in a way that doesnt require a lot of co-ordination between providers and consumers, the better providers win out automatically. It would also be important that providers dont get too big positions of powers.
  • BitLeafCloverBitLeafClover Member Posts: 7
    edited April 2014
    Came here to answer with "MaidSafe." Good call @Jasper
  • JasperJasper Eindhoven, the NetherlandsMember Posts: 514 ✭✭✭
    edited April 2014
    I am not sure about maidsafe yet, but i cant say i am well read enough.. The youtube video and whitepaper essentially contains the promises, they're all like "look we figured it alll out for you, trust us, and it". And i dont understand the mechanism(s) like i understand the blockchain yet. (love to hear if anyone else does.. or the better places to start)

    Even if we understood it, it isnt guaranteed their solution is the solution we should choose.
  • JasperJasper Eindhoven, the NetherlandsMember Posts: 514 ✭✭✭
    @nejucomo cool to hear Tahoe-LAFS is thinking about this stuff.

    Tend to think that the dropbox example essentially does proof-of storage, but you have to figure out payment for servers on retrieval. And it has to work against both Sigil attacks of servers failing to deliver, and requesters attempting to not pay, or falsely claiming they did not receive data.

    Dont know how things work, well, MAIDSafe claims they have it..
  • prosterproster CETMember Posts: 2
    Thanks for all the good answers. I think I need to study the proposals here to understand what they can and cannot provide. I'm also a fan of Tahoe-LAFS, so I'll try to dig into their mailing lists.
Sign In or Register to comment.