That doesn't matter.  What matters is that metcalfe makes the simplifying assumption that each node has an equal likelihood of transacting with any added node, and those transactions are of uniform value.  Clearly this is not true.  In fact the value in the first half of the adoption curve should be increasing, and in the latter part decreasing.  I would expect value to exceed metcalfe in future, and later to fall below.

As marketcap rises the upper bound on transaction value rises.  As adoption increases the likelihood that a transaction can be made in btc increases. As interest and participation by wealth holders rises the available value to transmit increases.  All this is yet to come for the most part.

As adoption saturates marginal players are added to the network.  Their value-add will tend to zero.  That is several years in the future.

Using recently active addresses tends to compensate for variations in proclivity to transact.  It fails to account for factors affecting the average value of individual transactions.