Funny though, I know what I'm talking about.

Since you clarified that you're talking about control systems, I expect what you mean is:

- The power meter reports wattage output to the cyclist.
- The cyclist adjusts their efforts to maintain a consistent wattage.
- The new effort level gets reported back to the cyclist.
- Thus, the information fed back to the control mechanism (the cyclist).

Your argument seems to be that any averaging interval on a power meter will obscure vital data, which will result in improper feedback sent to the cyclist.

This is incorrect.

I suspect one reason why you believe this is because you do not understand how pedal strokes actually work.

Cyclists don't apply 100% consistent power at every instant of a pedal rotation. In fact, a cyclist is more like a two-stroke engine. You apply an increasing amount of power on each downstroke and none on the upstroke, so it's more like a pulse than a steady stream of power. You don't actually apply power on the upstroke, even if you're clipped in; all you do is relieve the downstroke from the extra effort of lifting your other leg.

As such, if you were looking at instantaneous data, it won't be a consistent number; you'll constantly see huge spikes, and as such you won't be able to properly evaluate your efforts.

Consider the second graph below ("Vector Power Left / Right"), from Metrigear. Their power meters are in the pedal itself, rather than in the BB or rear hub. The power output of the left leg is in red, and the power output of the right leg is in green.



If you were looking at instantaneous data, your output would go from near-zero to 800 twice a second. Seeing that much data would be utterly useless in terms of evaluating your efforts, and would NOT work "better" than, say, a 2- or 3-second average.

Because a cyclist is more like a two-stroke engine than a water wheel, a sample interval that is too fast will be counter-productive for providing feedback to the cyclist, because the rider won't be able to make sense of it. You need a longer averaging interval to make the data intelligible.


So, what about longer time periods?

Keep in mind the test I'm proposing: We want to find the effects of changing one variable (weight) on the total time it will take to do, say, a 10-mile climb on a 3% grade. The purpose of the power meter is to keep the power output as consistent as possible, to eliminate a critical variable as best as possible.

The shorter the length of the test period, the less likely you are to have a measurable difference. A 10% difference on a 30 second run is 3 seconds. That's going to be extremely difficult to measure without a highly sophisticated testing system.

In comparison, a 10% difference over a 60 minute run will be 6 minutes. You could easily pick up a difference like that with a wristwatch. If you really wanted to go nuts, you could compare the charts to pick up any discrepancies due to fluctuations in the rider's power output and possibly correct for them.

Thus, in this case a longer measurement provides better real-world data.


Finally: You don't seem to understand that providing feedback to the rider is a critical function of a power meter. For example, pros routinely use them to avoid going anaerobic -- and they're using 2s or 3s sample times. If it wasn't providing reasonably accurate data to the rider, it would be worse than useless.


So, in sum:
• Power meters are designed to provide accurate feedback control data to the rider.
• You need to average power data over 2-3 seconds to get intelligible results.
• You don't lose any critical accuracy with a 2-3s average sampling rate.
• For our purposes, a longer test/sample time will produce better results.

As a result, if you want a real-world evaluation of the effects of weight on rider performance, and you don't have access to a multi-million dollar training facility, your best option is to do repeated runs on a specified route using a power meter to get the power output as consistent as possible, and have the run be long enough to make the effects easier to measure.

So, NOW do you see why I proposed using a power meter?