[TechniSol]

I'm 20 full days in since charging, 37 hours of reading on lowest lighting level, and at 40%...

As for expectations of linearity that are notoriously not so, it doesn't prevent me from hoping that someone might actually get it right. If they monitored current usage rather than voltage, as most devices do, the battery gauge could be quite accurate -once the actual capacity of the battery was determined. If that's not practical because of sleep cycles between page turns then perhaps it could be pretty accurately faked by monitoring what the processor is actually doing for how long and multiplying by predetermined average current draw based on what processes are active. Even that would probably be better than the commonest method of simply hanging an A/D off the raw battery voltage and interpolating between the highest monitored voltage and the lowest they are comfortable with to ensure charging is mandated before internal circuitry prevents further discharge of the battery.