Instructor's notes - before lecture
Council of Europe’s definition of “democracy”: Individual autonomy: The idea that no-one should be subject to rules which have been imposed by others. People should be able to control their own lives (within reason). Equality: The idea that everyone should have the same opportunity to influence the decisions that affect people in society. Compare/contrast with Dahl’s criteria. Equal opportunity is a key part of most of Dahl’s 5 critical properties. Can we as technologists quantify the “equality” of “influence”? Social influence distributions One person, one vote: democracy (nominally) One dollar, one vote: economics, corporate ownership and governance One web page, one vote: PageRank initial condition One reference, one vote: PageRank output condition One joule, one vote: Bitcoin One machine-learning input sample, one vote: governance by AI Including learned bias Old, from 2019: Some basic premises of democracy: No one is perfect or trustworthy enough to be guardian of all the rest Knowledge and perceptions are imperfect and noisy Representation, aggregation is essential to get “good” collective decisions It’s hard to tell - and dangerous to decide to early - who is trustworthy or qualified Foundation or “presume nothing” principle Robustness: general concepts and analysis techniques Engineering: fault tree analysis (FTA), simple quantitative risk estimation Application to fault-tolerant systems: e.g., 2-disk backup Application to network routing, gossip - e.g., UseNet Single path Two paths Star network Highly-connected network (e.g., expander, small world) Application to corporate governance: e.g., directors and bad strategy Distributed systems: benign faults, Byzantine faults, rational faults Threshold assumptions: e.g., Lamport’s “Byzantine Generals Problem” Statistical: e.g., median, mean, mode as measures of central tendency Social choice: robustness to manipulation, incentive compatibility Social choice and statistics: deciding on comfortable room temperature Trust/reputation networks: h-index, PageRank Post-lecture blackboard snapshot 2019: