[gabrielhauber]

Good question, one that I did consider at the time. I figured that the normal weekly variation at $4.99 actually made the total a little higher than it would otherwise. i.e. I went with my gut to a certain degree. You may be right about the more users thing at the lower price point giving better growth in the long term. The interesting thing (something I glossed over in the post) was that during the experiment, the total during the week it was at $4.99 was a lot higher than the baseline before the experiment (when it was also at $4.99).

[antirez]

Maybe the previous weeks have an effect on the next, for example a low price point brings more new users that will bias the result of the next week and so forth. It's hard to make a controlled experiment... but at least it is possible to do multiple experiments over time shuffling the order in which the prices are tried.