Poster: Scalable Bias-Resistant Distributed Randomness

Bias-resistant public randomness is a critical component required in many (distributed) protocols. Tor hidden services depend on a fresh random value generated each day for protection against popularity estimations and DoS attacks. Anytrustbased systems such as Herbivore, Dissent, and Vuvuzela use bias-resistant public randomness for scalability by sharding participants into smaller groups. TorPath critically depends on public randomness for setting up consensus groups. Public randomness can be used to select parameters for cryptographic protocols or standards transparently, like in the generation of elliptic curves where adversaries should not be able to stir the process to select curves with weak security parameters Voting systems, elections, polls, lotteries, and some Byzantine agreement algorithms also critically depend on public randomness and illustrate its importance. The process of generating public random values, however, is not trivial because sources of good randomness (in terms of entropy) are difficult to obtain for regular users. NIST deployed a publicly-accessible randomness beacon that provides hardware-generated random output and addresses the challenge of producing good randomness in terms of entropy. The beacon’s design assumes that everyone trusts NIST which is not a reasonable assumption, though, especially given issues such as the Dual EC DRBG debacle.