But here you are saying unique addresses....earlier Aminorex was saying 'active' addresses.
We're talking about different things. Peter R's chart referenced by AnonyMint was based on unique non-dust addresses. He also ran the correlation to the active addresses, which I prefer because I consider it the best proxy for a commerce node. I believe the gbianchi's chart, which described the I^2.26 fit was the first published fit of price to the latter factor. I think that was done some time in November.
Now, I just went onto a couple of exchanges and created half a dozen new addresses. But I'm still a single user.
The number of people using bitcoin doesn't matter as much as the number of commerce nodes. If everyone creates 10 addresses a day, and uses them for a transaction, then the number of active nodes will be the number of people times 70 (7 day rolling window). In fact everyone creates addresses at some average rate.
The point is that the number of active users is proportional to the number of active addresses, to a first order approximation. The number and sizes of their transactions will also play a role, but will be second order factors. The same proportional reasoning applies to the unique addresses, but multiplied by the average age of the addresses. The exponential fit is scaled by the multipler. It doesn't matter what the multiplier is. Thats how math works.