Actually, I've been doing some testing with both my Model 3 (approaching 10k miles) and an S on newer firmware with an older 90 pack (not my car). There is almost certainly some subtle number fudging going on with both.
On the Model S, when calculating energy usage from various BMS messages, the SoC (and thus rated miles) display appears to tick down a hair less per unit of energy than expected. Then, below 30%, this seems to ramp up to expected, and below 20% accelerates to worse/faster than expected... getting worse as SoC drops.
Comparing resting cell voltages throughout the cycle would also seem to confirm displayed SoC being higher than expected. I also calculate and compare reported power usage with other sources, such as debug messages from the inverters, which aren't lining up as expected nor as they did in much older firmware. I'm not sure when this disparity began.
Everything is pointing to there being some fudging going on with the Model S + 90 pack. I'm not going to pounce on Tesla on this just yet, as I only have one datapoint on this so far... and I'm going to give them a chance to explain.... but it's not looking good right now.
On the Model 3 I'm still working out exact details to decode on CAN, but I'm seeing something similar but possibly even more damning there. On my trips, efficiency would always seem to follow a pattern: bad for the first dozen miles or so, epic for the next 100-150 miles, and then worsen from there.... despite driving style. I chalked the initial inefficiency up to initial climate control and such... but haven't been able to explain the latter drop off, especially with consistent driving conditions.
Again, I haven't fully deciphered everything I'm looking at on the Model 3 CAN, so I could be misinterpreting some things... but if what I've decoded is correct......... then the 3 is lying about power usage for the first ~50% of SoC usage or so, making it look more efficient than it is during this time, then compensating for the lie during the remaining SoC. Starts slowly once at < 50%, and accelerates as you get closer to 0%. It's done in a way that makes it look like efficiency is worsening, when in fact power usage per mile hasn't actually changed.
Again again.... I *could* be wrong here... but I've gathered enough data so far to be pretty suspicious and worried about the actual efficiency of the Model 3. So far I've been able to drive it about 260 real miles on a charge with reasonable driving, which isn't terrible... but is only 82% of rated range. My lifetime Wh/mi displayed is also around 255, which is 13% higher than rated.
Long story short......... I see some evidence that suggests Tesla is trying to cover up extensive degradation of the 90 packs by shifting some apparent usage towards lower SoC, making the displayed SoC (and thus rated miles) show higher than actual SoC on the upper portions of SoC. This would seem to fit what
@supratachophobia has been describing here. Again, I only have one data point currently, but I will be gathering more as possible.
And then with the Model 3........ I really just hope I'm wrong there. Because if what I'm seeing and interpreting is correct and affects all cars......... then Tesla is *really* going to have some explaining to do. Like, dieselgate type explaining. Crossing my fingers that I just screwed up somewhere.