Observed distribution of keyword scores

I don't want to be asked all the questions for one keyword, and I don't want to be asked one question for all the keywords.

Both the keywords selected and the questions selected in the course of the chat should be moderately mixed

That depends on the distribution of keyword scores over the course of the chat.

I've never observed the distribution of keywords, and I've been fiddling with them rather randomly, so I can't predict what the distribution would look like.

Should we create a mechanism to observe

Should it be in real time?

I'm torn between being in the chat and being an observer.

Replay later?

A replay function would be beneficial in many ways in the future.

---