I've been lurking to date, but I had to register to explain to this guy what's going on.

Look, I don't know how the system works, but let's suppose it's as you say - ioi has worked out an algorithm for extrapolating NPD numbers from a small sample that he tracks. As ought to be obvious to everyone, the formulae clearly only apply as long as the numbers of Wiis/360s sold throughout the US (which is presumably what the NPD reports) continue to bear the same relationship to the number of Wiis/360s that ioi actually knows about. What happened last month? Wii sales in the US skyrocketed. Now, you could write this down to fraud immediately, or you could remember that Nintendo's been promising to increase supply for a while now, and was arguably creating an artificial shortage earlier.

Let's suppose for a moment that the NPD numbers were accurate - the Wii sold like crazy. If the Wii sold like crazy, is it reasonable to expect that the total US sales could still be extrapolated by the same formula from tracked sales? Of course not. We already have anecdotal evidence and common sense that tell us that Nintendo would have altered relative shipping allowances - Toys R Us wouldn't have seen many more Wiis, while Gamestop's would have seen many more. Unless ioi's tracked data is exactly representative of the stores in the US, this would have skewed his formula. His Wii numbers were low because he couldn't anticipate a large surge in the supply of Wiis.

That, to me, seems a far more reasonable explanation than NPD bias. I mean, given that ioi's only been trying to hit the NPD numbers, it just seems absurd to take his findings as the more accurate, since what he's doing is essentially curve-fitting.