Hadn't heard of the Blyth-Still-Casella interval or Barnard's test before, so thanks for that. By "If you want to dynamically tune a Web site to make the most

Hadn't heard of the Blyth-Still-Casella interval or Barnard's test before, so thanks for that. By "If you want to dynamically tune a Web site to make the most money as more information rolls in, that calls for a different experiment than a standard hypothesis test." you're talking about minimising regret / bandit algorithms?