Elon has already admitted VIN tracking is accurate with a week's delay on that call.
Yes I think we're staying pretty close to Tesla's production rate with our crowd sourced data. I continue to be impressed by the contribution of forum members and specifically those running the google sheet.
The pattern that intrigues me is the gap between VIN registration and VIN assignment. A common perception here (the one I subscribe to) is that the registration is a leading indicator, the VIN exists before a car goes into production, while the VIN assignment is a lagging indicator, meaning that the VIN is assigned when a car comes out of production. I think Elon may be referring to this lagging indicator, when he said that we're seeing things a week behind Tesla.
The gap between the two indicator is interesting, and one would expect to be somewhat constant, basically the time it takes for a car to go through the production line. Ideally if we can understand the relationship between the two, we can then predict the production rate using the leading indicator, instead of the lagging indicator.
The difficulty in that is that we saw the gap as high as 37 days during the Feb shutdown, then went down to 10 days in March, then ballooned up to 27 days during the April shutdown, and now down to 10 days again. Why would a 5 day shutdown wag the gap by so much? The two indicators don't seem to directly correlate with a constant time shift, but some "rubber banding" effect as well.
One possible explanation is that Tesla is preparing a big bucket of VINs just before the shutdown, so if the rate increases quickly after the shutdown, they can start building a lot of cars with that big bucket of VINs. But in reality the production rate doesn't step up to a new plateau immediately after the shutdown, so the bucket takes a while to empty out.
It makes VIN registration not very exact as a predictor, basically, if we see some VIN registered, we should expect to see if assigned anywhere between 10-27 days, which is probably not too useful.