Multicore comments on Humans are very reliable agents

[Multicore]

Robust Agents seems sort of similar but not quite right.