I think this raises an important point. Specifically, should the classification of the ride difficulty be focused on the difficulty of the final one or two intervals or the entire ride? Using Nateās difficulty scale, if a rider considered the first 4 intervals āhardā, the fifth interval āvery hardā and the final interval āall outā, should the ride difficulty be categorized as a 5 (all out) or a 4 (very hard) that reflects and average difficulty of all the intervals?
Further to this point, if a user needed an additional 30-60 seconds to recover prior to the final interval, and paused the workout to account for this, should the workout be categorized as a āfailureā or should this pause be ignored when rating the difficulty of the ride upon completion?
In sum, I think itās really important that TR clarify/answer questions like these so users are completing the post ride difficulty survey in a manner that will be most beneficial to adapting their subsequent workouts.
I would love the ability to ātargetā these workouts and have a progression to get there. Iām training for Leadville and Gibralter +2 looks like good training for Columbine for big slow guys like me.
Indeed. And to expand on another point you mentioned, Nate said they tested the ML for assessing the TSS of a workout outside using only HR and cross referenced it to the actuals with the PM data included. He said that the ML TSS assessment is super close with HR only that you could use it as a proxy. Having a PM is obviously better, but still, thatās a huge boon for people who donāt have the budget for PM but want the benefit from the outside ride data once it gets added into the mix.
Going by the podcast Iām pretty sure they arenāt just using tss. Remember the main problem with machine learning is you need to feed it good data. Tss doesnāt really fully describe a ride so isnāt good data on itās own. Quantifying a ride is hard
Pauses are accounted for in the ML, and a post workout survey allows you to tell it why the pause occurred. As has been mentioned several times in the thread, the ML is smart enough to know when those pauses are because the work was hard or because you got off to stretch or refill a bottle or something. And if it doesnāt know, it asks you why.
As for the 5-point scale, I think youāre over complicating it. A hard workout should feel hard. An easy one should feel easy. If you completed it but it felt All Out at the end, you mark All Out and it will adjust based on numerous features in addition to your subjective response to the post workout survey. And the more you do this, the more it will learn and the smarter it will become at defining your patterns.
If I want to add endurance to the end of a workout I just extend the cooldown and use the intensity button to increase the target power to the desired level. This way you can progressively up the intensity as the workout target power ramps down into cooldown. Intensity usually ends up around 250% for an endurance (cooldown) block.
Huh, plan builder had been awesome for me. I occasionally fail workouts, and have to juggle a few things, so this improvement is awesome, but plan builder was not a pile of ding by any stretch of the imagination.
So I didnāt have a plan built out yet and I did a plan builder run and dated the start date retroactively to when I started SSB1 in November and Iām getting recommended to pick up on short power build for a couple of weeks, although Iām not sure Iāll dive into something tough to start the weekend lol I havenāt seen any adaptations triggered yet, based on my current vo2 progression of 4 and going into a 7.2 straightaway doesnāt seem like something the AI would like lol
For some reason I thought intensity increase stopped at 200%. That may have been some time ago(older software) as Iām not often increasing to that level. I think that is what Iāll do going forward.
Chapeau @Nate_Pearson and the whole TR team. This is absolutely awesome. Clearly a huge amount of work has gone into developing these new features, and I cannot wait to see the results on my training.
What is really exciting is that it should help to answer many of the ongoing questions related to the best structure for a training plan (periodized, pyramidal, traditional base vs sweet spot) and how much volume is needed for gains and what is a minimum effective dose. Most studies on these topics are necessarily based on a few self-selected participants in artificial scenarios. TRās ML and huge data resource should be able to discern real-life results based on a wide-range of athletes. This will be a huge game-changer, not just for TR, but training in general.
One specific Q, and one area that I do not think has been mentioned so far, is how does your power curve feed into plan adaptations? And in particular, new power PRs. This is most relevant to data from outside rides, races etc. For example, during the course of my last two training blocks I have set a whole new set of PRs, but the majority of those have been when racing hard online. Of course, they are a result of the inputs (the training I have completed), but they too are a data point as to the power I can currently hold for a given duration - so will that also be taken into account in the progression?
It worked based off of ramp tested ftp. Hereās a bowling analogy. The old system had bumper rails, but no strike guaranty. Now the system is going to start steering your ball down the lane. Time to expect more a lot more strikes. The perfect 300 is probably years away and will require integration of additional sensors, but we are on our way.
It did create a rest day next Friday but thatās the only rest day created. Not sure about picking lower volume and having it build up, that may be something folks would want to do using the TrainNow feature
Given that he stated thereād be one more wave of invites to VIPs, and that bugs are already cropping up, I think itās probably fair to guess that the rank-and-file rabble will be seeing invites in 1-1.5 months
I think that is what Iāll do going forward.
