Consumer Behavior and Algorithm Design
summary
Claims that consumer-facing computing reshapes algorithms and gives concrete wins: a web adjacency scheme compresses links to about three bits per edge (MSR pushes toward two bits per edge) by lexicographic URL ordering and referencing one of seven recent adjacency lists, producing adjacency queries with short reconstruction chains (≤4–5 hops) and one-pass encoding after a sort. Observed power-law in-degrees fits a copy-with-probability model; MapReduce-era parallelism needs data-aware fixes — e.g., triangle counting becomes M^{3/2} and load-balanced by assigning each triangle to its lowest-degree vertex.
AI-written summary · about