Official statistical systems were largely designed to provide the government and the public with reliable information upon which to make many, possibly competing, decisions. Because statistical sophistication is heterogeneous, statistical agencies have struggled to communicate uncertainty effectively and to educate users on appropriate methods. Public opinion research had many of the same challenges, but one rarely sees polling numbers without margins of error, and the public has learned to expect such information from reliable pollsters. This talk will consider methods that might be considered by statistical agencies to assist users in processing the uncertainty that differential privacy and other disclosure limitation systems introduce. It will comment on privacy-protected statistics from both statistical and computer science viewpoints.