QLunch: Is there such a thing as private information?

Speaker: Charles Bennett from IBM Research

Title: Is there such a thing as private information?

Abstract:
One of the original motivations for quantum information theory was the use of quantum effects to protect the privacy of classical communications.  Yet the new theory, which has grown to elegantly encompass all of classical informatics, has undermined the very notion of classical private information, by showing that it sits on a slippery slope between quantum information and public information.  Classical privacy survives only as a useful approximation, since in principle any memory so well shielded that it can hold classical data without the environment finding out can also hold superpositions of the classical values, thereby serving as a quantum memory.  To recover a sharp notion of classical privacy it suffices to consider scenarios in which some information escapes to a place beyond the reach of one's adversaries.  
At a more practical level I argue that device independent (DI) methods of randomness generation and quantum key distribution are less secure than DIY (do-it-yourself), and measurement-device independent (MDI) methods.