My understanding is that the efficiency loss is strictly limited to delivering electrical power back to mechanical power via motor instead of via the chain drive. The problem with this design is that it completely eliminates the ability to deliver mechanical energy directly. The rider pedaling charges the battery which then powers the motor. Going from mechanical energy to stored electrical energy is going to introduce a much greater loss in efficiency than a mere 5%. In a more standard e-bike model, when the rider pedals, my understanding is that they are delivering mechanical energy directly, bypassing the conversion to electrical energy. I may be wrong as I am not very familiar with e-bikes, but, if this is the case, the 5% figure seems highly misleading.
See https://electrek.co/2021/08/31/radical-new-electric-bike-dri...
Real, especially cost reduced, systems could be a lot worse thougn.
The Author is arguing for replacing the chain with power wires, by generating power from pedaling and feeding it into the motor (with efficiency losses of 5%). So the power split from "human" power vs. battery power will barely change, with the benefit of less mechanical parts. It would thus also work just like a conventional E-Bike when the battery is empty (with 5% less efficiency).
> The second idea in the article which made me double-take is the idea of removing direct mechanical brakes from the bike
Many E-Scooters already use an electrical break (in addition to a mechanical one on the rear wheel), like the Ninebot E22. In my experience, I had less issues with using the front motor to break than many mechanical breaks on my previous bikes even though the communication seems kinda iffy (a bidirectional UART).
I wouldn't want to remove all mechanical breaks, but replacing the one on the rear wheel with an electric, primary one seems like a good idea.