For any correction factor, in all industries, it is only accurate within the parameters that it is outlined in the testing methods. So that’s within a certain +/- a certain percent total correction, but also includes deviations +/- % deviation on air temperature, oil temp, gas type and temp, coolant temp (and often it’s the jacket temp), humidity %, barometric pressure, stabilized power reading (so no sweep), knowing reciprocating losses, etc.
Just applying it without controlling anything the correction calls for introduces correction error, and then go beyond what the correction is accurate for, and you introduce increasing error the further you get away from that range.
Applying it with sensors that are half on the fritz, weather stations who knows where, calibration who knows what, etc, is like the like that meme of the infomercial guy slapping flex tape onto a leaking tank, or a landlord painting over moldy drywall, where people are trying to use the correction factor outside its intended use to fix errors.
I don’t think Eric is trying to mislead anyone, but that dyno needs sorting, and it’s something everyone should be looking at with their own graphs or anyone who shares them. It’s one thing if your dyno is repeatable and reads high or low, those numbers only mean something to you, but if people are selling horsepower results or are bench racing, then it should fair game to know.
Everyone likes to think air density, but humidty is going to effect combustion, so it just all depends.