What is the purpose of PMax? Initially I thought it was an actual measurement until it changed on me a few times. It seems like the most ridiculous and pointless metric available.

My current PMax is, once again, less than my actual 1 sec max. It's actually less than my 2 sec max! Silliness.

I'm not sure where you're getting Pmax from, but if it's from WKO or something similar it's one of the parameters used to derive your modeled power-duration curve. By comparing the parameters or the modeled curve to your actual data, you can learn things about the quality of the model and your strengths and weaknesses. For example in this case, since your modeled Pmax is less than your actual short term power, I would question whether you've done enough all out efforts from 1-60 seconds or so; or alternatively, if your 20-120 second power is lower than typical for someone with your 1-5 second power.

Okay, but the actual definition from trainingpeaks is: Pmax: the maximal power you can generate for a very short period of time (one complete revolution of the pedal stroke).

And that's a bunch of bollocks. It doesn't says it's derived from the pd curve. And when we actually have that max effort, it's pointless. We have our max sprint efforts.

We've got the power duration curve, okay, and we have mFTP, okay again, since that's a longer effort and has the most muddied and nonspecific definition of any other commonly used power term. But why actually have a display on the HERO bar (for WKO) as if it's some pertinent number (to be fair, that can be applied to FRC, TTE, stamina, too) when we have the actual real number.

Not that it really matters. Just one of those things that gets more attention (by default on WKO) than it deserves, in my opinion.

I'm not an expert on this, and I've never used WKO because GC is free, but the Golden Cheetah modeling tutorials do a pretty good job of explaining how the parameter values are estimated based on the power data if you're really interested. My understanding is that in the 3 parameter model, Pmax is used to calibrate the trade off between CP and W'. So while PMax isn't that interesting of a metric, CP and W' are, and based on my limited experience, the 3 parameter model seems to do a much better job of estimating a realistic power curve than the 2 parameter model. The thing I like about GC is that if I see a number that doesn't seem right, I can do a performance test of the missing duration and update the model with that performance data point. So in your case, if the PMax looks low, you could add some 1 & 2 second maximal efforts to the model to see what impact that has on estimated CP and W'. I agree that none of it really matters, but I've had some fun fiddling around with it.

To paraphrase RC, the purpose of a model is not so much to make predictions, but to sharpen our questions. If you're not interested in using Pmax for that, no problem; ignore it and leave it for those who find it useful. (but perhaps you might ask yourself if the developers of WKO thought it was important enough to put these parameters on the Hero bar, what am I missing if I think they're worthless?)