Vetter: Linux Kernel Maintainer Statistics

Am I misreading something or is there something a bit screwy with the “kernel w/o GPU” graphs? Looking at the bars for v4.16 in Figure 3 for instance, I’d eyeball that at around 1100 reviewed out of 2400 total, or about 46%. The corresponding yellow bar in Figure 4, however, only appears to be around 37%. Likewise, from Figure 3 it would appear that the percentage of reviewed maintainer self-commits was slightly lower for v4.15 (~1100/2500) than v4.16, but the corresponding bars in Figure 4 appear to show v4.15 slightly higher than v4.16.