Excellent use of back of the envelope math. I wonder if maybe the loss is non-linear, and a bit more is coming off because the tires are brand new. Perhaps there’s some coating on new tires before they ‘break in’?
Perhaps, but that would have to be a pretty steep curve, and one would expect that kind of falloff to have shown up in the data here (unless they literally drove the car once for only 1km, which...would be its own problem for credibility).