Bob renegotiates to exploit uncertainty in future value of information

[sabina-sa]

I can foresee the following game-theoretic problem. Bob knows that with non-zero probability he has become monopolist on the resource that we can label “key to Alice’s information”. This gives to Bob the power to re-negotiate whatever contract he has signed with Alice earlier, in hope that Alice now values her information more than she used to. So, he will message every Alice (or make her message him) and say “Alice, although we had agreed that you pay me 0.01 BTC, and I had deposited a stake of 0.1 BTC, I now want you to pay me 10 BTC or lose access to your information forever”. If Alice does not agree, then Bob may later lower the price gradually, until Alice finally gives up. This game has no clear equilibrium, but under some circumstances it can be profitable for Bob to play it. Basically, Bob has the incentives to try and absorb the value that stored information has now for Alice, when it exceeds Bob’s initial deposited stake.

Objections to the above argument, with answers to these objections.

1. “Alice could have foreseen that her information will go up in value and she should have asked Bob to deposit a larger stake as insurance” - the inter-temporal value of information is quite uncertain and impossible to predict with accuracy. Surely in some cases information will appreciate for Alice beyond the insurance that she has incorporated in Bob’s stake. Similarly, in some cases the information for Alice will depreciate (but Bob is protected against that). In addition, there is asymmetry, as the bound for depreciation is 0 value, while there is no upper bound for appreciation of information (think of a password to an address with tons of bitcoins mined in 2010).
2. “The case where information goes unpredictably up in value for Alice is quite rare and should not disturb the normal market activities”. While these cases are rare, maybe 1% or less of the total volume of transactions, they are “golden”, and Bob knows it. While Bob does not know which information that he keeps is “golden”, he may know from experience that e.g. he can renegotiate every 1 key in 100 keys that he stores, for x1000 of its initial agreed value. This may constitute a viable business model for Bob, with payoffs that are higher than he gets by sticking to the initial agreements. In that case, Bob will have incentives to renegotiate with all his customers, “mining” for that rare customer whose information has unforeseeably increased in value.

[sabina-sa]

There is an obvious extension to the above problem: Bob can renegotiate selectively, targeting e.g. high-stake contracts. That would result to "adverse selection" - Alice will not want to ask a high "insurance" stake for her contract because that reveals that she holds something valuable and makes her a target for renegotiation by Bob. For example, Bob may think "since Alice has asked me to deposit a whole 1 BTC stake to guard her data, high chances that this data is priceless for her and she will be willing to pay an even bigger ransom if I renegotiate".

On the other hand, if Bob's dishonest attitude becomes widespread, Carol may see how Bob acts and profit from it. Carol will ask Bob to deposit a high stake to guard her worthless data, and when Bob tries to renegotiate she will gladly get compensated with Bob's stake. So, the market could balance itself.

[dr-orlovsky]

It seems that the problem is solved with the new proposal here: #6 (comment)

now Alice can decide to punish Bob and he (if he keeps the data) will just get less than was originally agreed, so Alice's negotiation point is strong