A merkle tree could be used for verifying a tree of data (e.g. a JSON state object) that is updated in pieces. The root hash would be sent as part of each upda

A merkle tree could be used for verifying a tree of data (e.g. a JSON state object) that is updated in pieces. The root hash would be sent as part of each update, to validate that you have the right state. If it's different, a hard update isn't necessarily required - you can find missing items in log time. In addition, for large states (e.g. say all cars in a region/partition for Uber), you can use merkle trees to partition and validate that the client has one area right, without downloading all the data. There are many other approaches, but I think a merkle tree isn't a bad method for the use case of client state validation in the use cases I described. What are your thoughts? How would you architect it?