that depends on your definition of "most people here". If you count every user that ever logs in in a given month, then the average could well be that.
You have to assume an 80/20 Principle distribution, which is also recursive. So 20% of users will have 80% of the minutes in the system per month, by recursion, 1% will have about 50% (rounding a bit to keep the math simple).
If we assume 60M U.S. users with any monthly activity (generous), then they would spend a total of 180M minutes if the 3 min average were true. And the top 1% of those would spend about 50% = 90M, divided by 600k users = 150 minutes on average per user per month. That would be 2.5 hours.
If you recurse it one more time, you could get 120k of the top 0.2% or so of users, spending 72M hours, or 600 Minutes (= 10 hours) per user per month.
Do you see how this works? So it could well be that Comscore is off in its estimate, but probably not by an order of magnitude.
Let's say the average was really 10 minutes. Would that make you happier about their findings?
That would mean that the 120,000 top active U.S. users use it for 30 hours or more per month in that case. Keep in mind that that sample of the 500 or so people that YOU are following on here may be among the even more active subset of those 120,000. That's actually extremely likely.
But that doesn't mean that the average for the entire system
is suddenly going to be 10 hours per month per user.
/cc +Alexander Becker +Max Huijgen