Samples an event from a source distribution, calculates that event's surprisal under the current model distribution, then updates the model to include the new event.
Watch how the model distribution — and its entropy — drift as more events are folded in.