A PALGN Podcast Special: 2007, the Year that was

One could say that VGChartz is always accurate right up to the point where the numbers get changed, at which point it's apparently accurate again. What I'm trying to say is that their most recent results are always inaccurate until they have official numbers to get reconciled against, at which point they're revised to be closer to the official numbers. They interpolate and estimate - while it isn't necessarily a bad thing, it does mean that there's significant degrees of unknown error included.

Statistically speaking, they're going to be suprisingly right some of the time through pure chance - if you're extrapolating off a trend and your errors are normally distributed, you're going to be within +-1 standard deviation of the real result roughly 68% of the time. You'll be +-.5 standard deviations roughly 40% of the time, and within +-.2 standard deviations for roughly 3 out of every 20 predictions. For most predictions, that's pretty darn close. However, errors compound, and given they often don't update their figures based on official announcements for up to two months at a time sometimes, that can lead to some fairly big errors over time.

Japan's fast with their sales resuts, but the EU / US tends to lag quite a bit, if they announce it at all. What I'm trying to say is that their estimates for the month of December may be all over the place, and as such, I wouldn't rely too heavily on them. If you look at their results, the hockey stick effect starts roughly at the beginning of December. Up to that point, even by their numbers (which have some serious assumptions behind them to begin with), their cumulative sales estimates for the Gamecube and PS3 are within a few hundred thousand of each other.

I'd suggest waiting and seeing - based on official numbers that have been announced so far, the PS3 is performing comparably to the Gamecube over the same period of time. That doesn't mean its lifecycle will be the same as the Gamecube - it's just an interesting fact.