It's worth noting that the change is probably indicative of a large number of community accounts being deactivated by the switch, rather than the result of 10,000 subscriptions disappearing over the last five months, and is further evidence that the number displayed in the group membership reflects only a subset of the actual number of D&D Insider subscribers.
While it's not impossible that 10,000 subscribers ended their subscriptions in the last half year, it runs counter to the projected increases we would expect based on data over the last five years. Based on rates of decline in growth over the last few years, we shouldn't expect to see net losses of subscriptions until 2015. It's possible that the announcement of D&D Next caused some to end their subscriptions, but I see that as unlikely given that the practical utility of the subscription won't decrease at all until the game is actually released (and it also occurred well before the last measurement period).