Not everybody is using power meter for training.
You might use it just as a simple "power estimator", to help you control your efforts on longer rides.
+-5% or even +-10% is good enough for that kind of use. It is still much better than using just HR alone.
And why pay 500+$ for +-1% device, if you don't need that kind of accuracy?
I might buy it for my GFs bike. ¯\_(ツ)_/¯
I am not going into the philosophical debate how effort has to be paced or how one should train, but 10% drift in power on the same training could be a difference between blowing yourself up and making it through at exact effort. It is mandatory for power meter to be:
- reproducible in its technical error
- reliable in its ride-to-ride comparison
- able to compensate for temperature/pressure drift
It MATTERS WAY LESS how high or low it reads, how much it lags on sprints or how much it weighs or costs. If it makes it possible to compare today's training to the same training from 2/6/12/128 weeks ago, than it is good enough. If not, then it is more convenient to install some RNG plugin on your Garmin screen and call it a day, there is no need for a power meter and you are better off with HRM and RPE only (yes it is possible to train effectively that way too).