David Grantham
October 15, 2024
11am
3147 MEB
Stability in Decentralized Networks
Abstract
Is it possible to build a metastable, ad hoc, distributed network absent of any fixed infrastructure and consisting entirely of mobile peers that join and leave the network randomly, change their IP address randomly and change their NAT/firewall status randomly?
Assuming it is possible, what are the design criteria that ensures a network will trend towards stability, achieve preferential attachment. and become THE network such as THE web or THE peer-to-peer network? Drawing from research conducted over 25 years as part of projects such as PGP, Tor, Second Life, Firefox and various blockchains, I will summarize everything I have learned—including anti-patterns—about the fundamental problems of decentralization, best practices for peer behavior and a new thesis that link identifier characteristics entirely
determine if a distributed network trends toward stability or not.