In this update we are introducing a new system for gathering data on the type of changes that are harder to evaluate or predict without very large number of games, such as gold bounty formula changes.

We expect the number of times that we actually do this during the year to be very low, however when we do it it'll be during weekdays only at the lowest point of user activity, between 4 and 6PM Seattle time. In addition, we are adding a convar "dota_disable_experimental_gameplay" that can be used by players to disable it. Any gameplay changes will only be active in matchmaking games where no players have the convar set. The changes will never be active in lobby and league games.

We recognize that in some cases players will back solve what changes were done, however we won't announce changes ourselves in part because we don't want players to overcompensate with knowledge of them and because generally we prefer to not reveal potential upcoming changes.

Although the type of changes we are thinking about, like gold distribution, take much more than a game or two to show their full impact, we will still be able to get meaningful data and analysis on the direction and scale of changes.

More...