I do not see any threads where this is possible to submit, but the behavior I just observed has to be considered a software bug. I’ve seen similar(ish?) reports in some of the other threads but not as cleanly described as this.
I had Geiger +1 (4x12 SS WL 5) scheduled today and felt decent, had more time, and it was sunny outside so I switched it to the workout Sandia (4x15 WL 7.1) which was under the ‘harder’ alternates tab. I know there are ongoing bugs with the alternates all reading ‘not recommended’ but this seemed like a reasonable increase in difficulty for what i wanted.
As soon as i switched the workout my predicted AIFTP went from 263 to 251 - we all know this is 100% impossible, one workout is not going to change my FTP 9 days from now, let alone LOWER after I do more work. As soon as I finished the workout, my AIFTP went back but now 262. Functionally the same, although I have no idea why it’s 1W lower, but it doesn’t matter that much.
I don’t know whats going on in the back end of TR but this behavior is absolutely something I wouldn’t expect, and it’s physiologically impossible. I know i’m not the only person seeing weird stuff like this, but I haven’t seen a thread where odd behaviors like this can be reported.
1 Like
I think the AI was predicting you would probably fail the harder workout. That’s why the prediction bounced back after you successfully completed it.
4 Likes
Firstly, the TR estimated chance of failure went from 1% to 2% so not what I would consider a significant chance of failure worthy of a FTP modification.
Even if I HAD failed the workout the same applies, one failed workout does not lead to a 10W physiological change. If I had skipped the workout entirely I am willing to bet the difference between doing the workout and not would not even be measurable. We’re not talking super critical power/build workouts here, this is at best a mid SS level workout based on the difficulty, and was more like high tempo at best.
Even if we throw the traditional definition of FTP out of the window as needed for the new model, a prediction shouldn’t be made before the model even has data. And if it has data why would it be logical to respond in such a drastic measure? A 10W change is really significant for one new data point.
I’m no expert, but I think the AI is reacting to more than just the one specific workout. It thinks your ego is writing checks your body can’t cash and then it changes its opinion of how you will do for the rest of the month. More likely to screw with the plan, more likely to fail workouts, more likely to get fatigued or injured, more likely to get morale issues and skip workouts. Something like that. Then you went and completed the workout with no trouble and it said oh never mind. Shrug. Seems like an overreaction to a single workout obviously.
4 Likes
Here’s another bug. I finished this workout today and the preview screen keeps asking me why I ended the workout early but if I click it it just shows the standard difficulty rating which I had already answered. Opening the full details screen of the ride does not show this.
I assume the system does know I finished the workout and its just a display problem. I did this workout outside which may be confusing it.
This “ended early” one is a bug 
2 Likes
It’s a prediction.
If a weather forecast predicts rain on a certain day and then the prediction later changes to clear sky it doesn’t mean that the forecaster has actually impacted what is going to happen on that day.
3 Likes
Is this REALLY a bug though? Steamboat +2 is a 60 min workout, but your ride preview screen shows you only rode for a duration of 53 minutes (11% short)? Even though I assume that’s the cooldown you cut out, it’s still not a complete workout.
2 Likes
I skipped part of the warmup and did 1 minute rest between the intervals instead of two. Yes bud it’s a bug. You can even see my power and consistency for each interval in my screenshot…
When’s the last time you saw a weather forecast change 7 days out in one hour buddy? Let alone the best example in your analogy is it’s sunny, suddenly there’s a typhoon for forecasted and immediately after predicting it that typhoon disappears. Thats the equivalent - there’s no physiological reason your body would adapt that quickly. And no meteorological reason where you’d observe such extreme phenomena appearing and disappearing in your model under standard rather than extreme conditions.
You mean how often have i checked the weather forecast once and then checked again an hour later and the forecast is different?
Er…. often 
3 Likes
Why did you ignore the rest of my message?
