About 10 years ago I stopped bicycling very much, because my knees hurt. Since then I've mostly gotten around by walking, running, and public transport. With COVID, however, I've started bicycling more, and initially my knees were hurting again a bit. A couple weeks ago I bicycled about half an hour to get somewhere, and by the end my knees were hurting a lot. On the way back, I decided to try a technique from hypermiling: pulse and glide.

I put the bike in mid gear, stood up, leaned over the handlebars, and pushed hard through about three rotations of the pedals. Then I sat down and coasted for about 30s. I repeated this all the way home, except for one hill that was too steep for any coasting, which I walked up. When I got home, my knees felt decent.

I tried this again on Thursday (40min round trip), and then again yesterday (1hr round trip). My knees feel totally fine!

I initially expected that this would be significantly slower, but, for me, I think it's actually faster! Normally I bike slowly, trying to keep things easy on my knees, so this is a bit of a low bar. Looking at my most recent trip, I biked 5.0mi in 29min, and Google Maps predicts 31min. I'm usually more like 30% slower than Maps predicts.

Two guesses about why this might feel better:

• In regular bicycling, I'm making a lot of rotations. I usually use a low gear, move my feet quickly, and don't use a lot of force. This is a lot more movement in my knees than I get in the rest of my life, while when pulsing it's probably closer to normal.

• My bicycle has a traditional upright riding position, and when I pulse I'm out of the saddle. I think it's likely that my pedaling position when pulsing is a lot more ergonomic. At the very least it's much easier to put out a lot of power.

pulsing: standing up, accelerating quickly

gliding: sitting down, resting, decelerating slowly

One thing I like about pulsing is that it works really well with traffic lights. I can choose when to pulse and went to coast, and generally get to lights when they're green. I am willing to go through red lights on a bike after looking around carefully to make sure that I have sufficient visibility and there's no one coming, but it's still much faster when green since you can stay at full speed.

I had an idea last night for how one could make a bicycle that was even better suited for this kind of intermittent pedaling. When you pulse, you're turning your body's energy into increased speed on the bike, and then in coasting you lose that speed to rolling and air resistance. Since the bicycle isn't all that aerodynamic, and air resistance goes up (linearly) with speed, it would be nicer to store that energy mechanically instead of kinetically.

What if you added a spiral ribbon torsion spring, like the mainspring in a clock or a wind-up toy, to the drivetrain? It could go inside the rear hub, functionally between the axle and the shell. When you pulse, you accelerate but you're also winding the spring. When you coast the spring unwinds, helping turn the rear wheel. You need a ratchet to keep the spring from driving the pedals in reverse while coasting.

You also need a way to keep it from driving the wheel when you're trying to slow down. I'd put a coaster brake on it, and have backpedaling disengage the spring. For bonus points, use the momentum of the wheel to wind the spring, as regenerative braking.

You could combine it with planetary gearing, with a bunch of fiddly engineering, but it would work out of the box with a standard derailleur system. I expect if I spent a while looking through historical patents I would find something like this, since all of the technology involved existed during the heyday of bicycle inventions. Which also makes me think there's something important I'm missing that keeps it from working, unless it was just that the materials weren't up to it, and perhaps are now?