It was. What you see there is the average brightness in a rectangle of image contained entirely within the kitchen doorway, divided by the average brightness in a rectangle of image contained entirely within the dining room wall (which I assumed to have constant real brightness). If I plot the two separately, it's clear that the camera is indeed adjusting, because the brightness of the DR wall goes a bit down as the brightness in the kitchen goes up.

Oh yes, score FAIL for my reading comprehension.

It would be interesting to see the two separate graphs of the unadjusted average brightnesses in your two rectangles, to see how the camera adjusts over time.

No sooner a word than a whatnot:

(This graph is a bit less processed than the last: it's got the full data set as opposed to just the bit between me hitting the switch around frame 170 and reaching peak brightness at ~7000.)

Damnit, why don't you work in our lab? That is exactly the sort of stuff we need to think of all the time.