Ok. The relevant statement is this one:
With our Horizon product, the pod will communicate via Bluetooth to Dexcom CGM
So internally it gets glucose readings directly from the G6. Those are, presumably, the unsmoothed readings, since the smoothing seems to take 10-15 minutes (2-3 readings) and that is pretty much useless for a closed loop.
So, I still don’t understand; Insulet are not measuring BG, they are taking readings from the G6, probably with a lot of support in interpreting them, but still from the algorithm in the G6 transmitter. They say, emphasis added:
The identified anomaly could result in the system using an incorrect glucose value which has the potential to impact insulin delivery.
So, the G6 delivers shxt and Insulet use it. They came to the same conclusion as @Eric in this thread;
He compared G6 with, just G5, and had problems with the G6 reading high. Quoting Insulet again, “an incorrect glucose value which has the potential to impact insulin delivery.” @RCA221 was more explicit (yes, this is a cherry picked comment):
G6. 7.1% of the time, or 9 readings out of 126, your bolus, which was intended to take you to 100, would have instead dropped you BG into the 50s and 60s. (4 50s, 5 60s).
Despite the cherry picking I can see that Insulet might concur and be concerned for exactly the same reasons.
If the G6 transmitter delivers consistently high readings that can, potentially, be an issue for loop algorithms. It depends on what BG the algorithm is targetting and how big the errors are; having a higher BG target mitigates larger systematic errors at the cost of killing us sooner.