I am currently conducting a survey on censorship resistance and have identified a promising approach to mitigating frontrunning in the context of multiple concurrent proposals. In exploring this direction, I am operating under the assumption that validators are honest, though not necessarily rational, a premise I cannot conclusively prove.
Given this, I am contemplating whether all protocols must be incentive-compatible to be considered for production or serious consideration. Without ensuring incentive compatibility, there is a risk that strategies could emerge which may destabilize the protocol or undermine the integrity of the system being evaluated. Would you agree that this is a necessary requirement?
DSIC is very restrictive property. It is nice to have, but certainly not a necessary requirement in my opinion. Do you have an example of a DSIC protocol in the context of blockchains?
In the context of blockchains, when conducting an auction, it is expected that bidders will adhere to a dominant strategy incentive-compatible (DSIC) mechanism. In Tim Roughgarden’s book, Twenty Lectures on Algorithmic Game Theory, where he explains DSIC in the context of English auctions and its applied to various protocols. This works or are you considering other non-auction-related examples?
Sure, auctions are probably the most famous example, which can also be applied to blockchains in various ways. I was more curious about other types of DSIC mechanisms/protocols, which are more blockchain specific. One example is this paper by Babaioff et al.: On bitcoin and red balloons | Proceedings of the 13th ACM Conference on Electronic Commerce, in which they look into mechanisms in which forwarding user transactions is dominant strategy incentive compatible. They show that it is impossible to have DSIC and Sybil-proofness. The latter is a property you want to have in the blockchains, where Sybils are cheap(er) to create. Related to this, recently, we showed that in a single item auction, one can’t have DSIC payments and Sybil-proofness: [2407.14485] On sybil-proof mechanisms (arxiv.org).
What are other examples in blockchains where you want to have DSIC property? I can imagine you want (optimistic) rollup validators to check state all the time, and design rewards so that it is dominant strategy to do so. There are probably more examples.