* - I picked six races as the number as it was close to being the first third of the season. If you want to count earlier in the year than that, three races makes it 21 - 8 in favour of RBR. The only way it can come out in BMW's favour is to take just the first two races (8 - 4 to BMW). No doubt you will now declare this a victory for your argument.
Consider that the cars were being developed at a terrific rate. Cars that were winning races 1 or 2 races ago were totally out of sorts in proceeding races (see how BrawnGP went backwards fast, after Button's last win).
To my mind, BMW did make a good start to the season and continued pretty much from where they left off in testing. Kubica was in 2nd, with Vettel 3rd, when they collided in Australia. As it turned out, RBR had a championship contendiing car. And like you said, by Race2, BMW were also doing OK (in terms of speed, not retirements, where Kubica had effectively retired twice in 2 races). As the season wore on they began to go backwards.
Had the other cars not developed so quickly, from one race to another, BMW would've hung on to their decent pace, for a little longer.
I stand by my original comment that BMW did make a decent job in pre-season testing and also a good start to the season. And furthermore, if you are fast in pre-season testing, you are almost certainly going to be fast in the first few races of the season. For 2009, BrawnGP are evidence of this.