Regarding your overall point about recalculating XSS after the activity, I can appreciate why this would be confusing until you understand what’s happening behind the scenes.
The rationale is fairly straightforward. Let’s say you start a ride with a Threshold Power of 250 W, but during that ride you demonstrate that your actual fitness is higher and Xert extracts a new TP of 260 W. When we perform the final analysis of that activity, we believe it makes more sense to use the newly validated Fitness Signature rather than a signature that the activity itself has demonstrated was underestimated (incorrect, if you will).
This becomes particularly important with Breakthrough activities. I noticed that MPA is hidden in your screenshots, but during a Breakthrough it’s quite common for your power to exceed the old MPA. By definition, you cannot actually produce more power than your Maximum Power Available—the fact that you did tells us the starting signature was no longer representative of your fitness.
Here’s a fairly extreme example from one of my recent races. Using my starting signature, I spent roughly 25 minutes “above” MPA, and the real-time calculation awarded me ~236 XSS for a ~40-minute race (and a max difficulty of over 230!!):
Clearly, I wasn’t actually riding above my maximum for 25 minutes. 
The model simply didn’t know my updated fitness yet. Here’s what the data should look like using a valid signature:
There’s an old saying in statistics: “All models are wrong, but some are useful.”
Xert is ultimately a model of your physiology, and its usefulness depends on the quality of the inputs we’re giving it. If an activity provides new evidence that the starting Fitness Signature was no longer representative of your fitness, it makes sense to update the model with that new information.
Once the activity establishes a more appropriate signature, recalculating MPA, XSS, Difficulty, etc. with that signature gives us what we believe is a more accurate and useful representation of the activity.
That being said, I do understand your broader point: ideally, the real-time and post-ride numbers wouldn’t disagree in the first place.
One thing we’re considering with the Xert 2.0 model is allowing our mobile apps to update/recalculate the Fitness Signature (and therefore MPA, XSS, etc.) during the activity itself. That would ultimately be a better solution because the real-time calculations could adapt as new fitness is demonstrated rather than waiting for post-ride analysis.
In the meantime, there’s actually a useful way to compare the two approaches. When viewing an activity, click PREVIOUS underneath the MPA chart. That applies your previous/starting Fitness Signature to the activity, allowing you to compare the chart and metrics against the final analysis. This is exactly what I did to get the screenshots above.
I’d be curious whether comparing PREVIOUS versus the final extracted signature helps explain why we believe the recalculation is necessary.
) so final XSS is a bit lower (most users are content with achieving a BT)
), so the final XSS is likely a bit higher (signature goes down a bit, but you get more XSS credit for the ride)
)
