The issue with drafting would be that it comes and goes. In a sprint finish, one moment you might be drafting, another trying to overtake the other guy (so not drafting), then fall behind again and draft, all in a matter of seconds.
Then, there's the road gradient, also a massive impact on the power/speed function, and it's highly localized. You might know that a stretch of the road is 8% on average, over say 100m segments, but locally it can be much more or much less. On sharp switchback the inside of the corner can be 20+% when the outside is only 5-10%.
Plex it's a race, power goes up and down in response to events, one moment you're sitting in a group, another you're chasing the guy in front. Presumably that's also where you'd press the boost button - when you'd legitimately be putting down more power.
In amongst all that I doubt you would see an extra power boost.
The battery and the motor would be significant weight to carry when it comes to the mountains, not to mention this deadweight must be carried all the way from start to finish and only be used for maybe 60 seconds.
Bikes have a minimum weight of 6.8 kg in UCI competitions and the lightest bikes are at least a kilogram lighter than the minimum, so they have to add weight anyway.
This would theoretically detect the opposite. The power meter (on the crank arm) detects force applied on the pedal. The discrepancy would be if they're overtaking other racers pedalling the same wattage.
However, mandatory power recording could become quite valuable over the long run. The peloton should be large enough to allow data scientists to detect the abnormalities (unless all the teams reach an agreement to cheat consistently ;)). Once a history is built over a decade or longer, "unphysiological" power output could be detected quite easily.
> https://www.garmin.com/en-US/p/658593
It doesn't need to be pedals either - it could be a standard crank arm or any other easily accessible part that can effectively house a power meter.
I'm sure there are plenty of other issues with this power meter based approach, but securing the data is probably doable to a reasonable degree.
This is not that different to amateur motor racing, where its expected you only bring a helmet that's certified by an external body (Snell/FIA etc). There are generally only a handful of pedals designs in use in the pro-peleton today, so not like we are talking about moving a mountain, and the UCI already regulate a lot of existing parts.
As I said, I suspect this is not the main problem with this approach, there are likely many other much better reasons why this isn't a good idea.
Overall, it's just simpler to just scan all the bikes.