number of pending PANDA exchanges is a distinguisher

[leif]

Reattempting all of the pending PANDA transactions at the same time distinguishes a user to the PANDA server and to network adversaries between it and the user.

It would be nice if PANDA had the same traffic analysis countermeasures that the pond protocol has (including fixed-size and no-op transactions).

[burdges]

How would you suggest doing this? Implementing the PANDA server over pond itself would avoid adding new no-op transactions.