Doesn’t that threat model defeat the purpose of the algorithm leaks?