...Couldn't it be possible for this waffling to happen just based on slight variations of cell/string/pack voltage? As every measurement has some inaccuracy and even slight system loads could affect the measured value, I could see this happening. Also, you note that you were plugged in at the time. I believe the car will start to charge at times that the climate control is enabled, so perhaps your car picked up a very brief charge.
Yes, it could and likely is -- although no one on these boards knows as an absolute what the algorithms are that Tesla uses. My point is, if the underlying measurements in the battery, temp, etc are changing, Tesla should even out some of the variation before presenting the information to reduce concern. IMHO, simple (almost "dummy proof") data presentation is very important in mass marketed products -- which even our MS is starting to become as it approaches 100,000 on the road. For a number of us with 90kWh batteries, we're seeing perhaps faster initial degradation of Rated Range (see other 90D Thread on all that), so some of us watch these numbers very closely and odd numbers give us even more concern, especially if they change before our eyes for no apparent reason.
While I agree MS will pull charge sometimes when it's parked and plugged-in, I don't believe that was the situation here... There was a full 90% charge (which started at 12:00AM and completed sometime before I checked it at 4:30AM). From my experience, if I leave my MS sitting for 2-3 days and don't drive it, but leave it plugged into my HPWC, it does not begin it's own charge cycle to replace the vampire loss until it reaches a larger threshold of several miles (8+) under the set charging limit. It also does not come on, go off, in rapid succession within seconds or even minutes if I'm not doing something to the vehicle with climate control. It was also between 60-70 degrees F during all this time so there is no reason MS would have likely tried to keep the battery warm. To me, it's hard to say how or why the number varied up and down by 2 miles in less than 15 seconds -- it may be real, but if it is, I'll go back to the point that Tesla needs to normalize the result in what's displayed to the user -- Remaining "Rated Range" is after all, much like what ICE owners are used to when reading a gas gauge to determine how far they can still travel -- the gas gauge keeps going down but does not present the data to some infinitesimal decimal point in how much is remaining and even when going up/down hills in more modern vehicles, the gauge doesn't generally vary enough to make a driver worry -- it's all been normalized using physical devices and in the way the data is presented to make it more usable. IMHO, Tesla needs to do the same to instill driver confidence.