Author: Peter Todd 2023-08-17 15:56:12+00:00
Published on: 2023-08-17T15:56:12+00:00
The email discusses a problem with the understanding of "normal" connections and loss events in relation to k-anonymity sets. The author argues that while their math may have been sloppy, the non-loss events in the k-anonymity set can potentially be loss events as well. They explain that a node with N clients will experience a certain number of delayed connections per year, represented by 10 r_d = N * ---- year. Additionally, the same node will also see a certain number of connections with data loss per year, represented by 1% r_l = N * ---- year. The probability of a connection attempt representing data loss is then calculated as r_l P_d = --------- r_l + r_d, resulting in a dimensionless ratio of approximately 0.1%.The author acknowledges that conditional probability may not be relevant in this context and agrees that user error can be a factor in some cases. They suggest mitigating the risk by making the process of emergency restore from seed inconvenient, such as by adding artificial delays. They argue that this type of engineering problem exists in various circumstances and should not be considered a deal breaker for the idea being discussed.Furthermore, the author addresses another concern regarding the control an attacker may have over randomly chosen blockchain data sources. They believe that as long as the probability of this happening is low, it is not a significant concern.Overall, the email delves into the concept of k-anonymity sets, the calculation of probabilities for delayed connections and data loss, and potential ways to mitigate risks and address concerns related to user error and attacker control.
Updated on: 2023-08-19T01:46:34.307356+00:00