Digital EPO is flawed, you can see the peaks on the speed trace when analysing the segment to know if someone has doctored their ride.
That used to be true, the old version would just remove a trackpoint every so often and then shuffle up the timestamps accordingly, so an original track moving at a constant pace would be interpreted as "20kph, 20kph, 20kph, 40kph, 20kph, 20kph, 20kph, 20kph, 20kph, 40kph, 20kph, ..." and easy to spot.
The revised implementation does interpolation between the points to make the speed increase look consistent, that way it can still correlate to the peaks/troughs in the power/cadence/HR rates in the data, but it requires frequent data points (ideally one per second) otherwise the interpolated locations may be off the road (and easy to spot that way).
At one point in time (before they fixed Digital EPO) I was going to write a version that did the above (smoother modifications) but included a way of watermarking the resulting data in such a way that it was detectable by companies like Strava at some point in the future, i.e. change the least significant digit of the lat/lon values to be a checksum against the other digits. I'd wait for a while so lots of people were using it and then hand over the details to Strava/GC/etc so they could flag such rides
en masse.
A device that logged cryptographically signed trackpoints would be nice.