Thatās how I interpret it as well. Some all outs required a back pedal during the last set of over unders is maybe a 9. Anything earlier in the workout or reduction in intensity maybe a 10. It probably doesnāt matter as much as people think.
3 Likes
Me too. Guess the update really wasnāt ready for prime time.
@Jonathan Iād love early access to the new TrainerRoad AI 
1 Like
I could be wrong, but think you need to go here to request beta access, instead of tagging reps.
3 Likes
Got it, thanks @mcneese.chad !
Iāll point out that Jonathan gave a brief update about the launch in the podcast (or at least he says he did, I havenāt listened yet), see below.
Iām in the beta and will say that they have been fixing multiple bugs a day since they returned from holidays. Thereās a few specific ones that I suspect they want fixed before launch. They seem like they shouldnāt take long to address, but thatās just speculation on my part.
7 Likes
If you look in the apple store entry, all they ever day is āunder the hood improvementsā
(scroll down, click on whats new)
He talked about it. Said they had hoped to have it out by now, but that the beta folks found enough that they wanted to smooth it out first before full release. Makes total sense. The complexity here with the amounts of different ways people use the app (phone/mac/pc/tablet), the zwift integration, etc., there are probably a million little annoying things to work out.
9 Likes
Thanks!
Iām not sure whether this is just relevant for the new release; I never received fullprogression level benefits for a sweet spot workout despite having completed all the intervals. I did go the toilet during the second rest interval so this no doubt the reason. I did pause the workout on Zwift beforehand though. I rated the workout as very hard. I thought completing the work intervals would be sufficient to get full PL acknowledgment.
Nope. AI will look at what you actually did - and extending the rest period for whatever reason will make the workout easier.
4 Likes
Thought so 
The closed beta has been going on for months - but at least a couple of hundred more people were added around Christmas.
As it is the software is perfectly usable in my opinion - but new small things have come to light after having so many people looking at it.
No show stoppers.
If they were Garmin it would be out 
10 Likes
Such a bug bear of mine 
The line from TR staff is to ānot try to crack the codeā
However, if you base it on wording you get something like this I posted in the past:
I dont have the inclination to test any of this though - I do my workouts using TR with ZWIFT running alongsideā¦.
3 Likes
I think that those different perceptions of how hard a workout is are exactly why documenting it isnāt particularly helpful. With an AI model then being consistent as an individual is likely more helpful than trying to get everybody to align. Similar to a human coach, over time youād expect the AI to learn the difference between the athletes who think that every workout is somewhat hard and will never use the āeasyā answer, and those who will doggedly claim they most workouts are easy or moderate and will only reach for the āhardā option when theyāve gone to the well just to finish it.
Basically, donāt overthink it, just go with how you feel and apply the answers consistently for yourself.
5 Likes
I have never understood all the hand wringing and worry over the rating. Just answer it with your gut as TR says to do, and move on with your day. I just complete with my first instinct and never ever ever think about it again. I trust that the machine is appropriately weighting the actual completed work and that the survey is likely only nudging recommendations going forward.
21 Likes
I totally agree wrt the TR ratings - but if you have athletes using a mix of Zwift and TR to do their workouts then they surely need to be consistent?
2 Likes
I guess I still live in the camp of complete the survey on your first instinct and move on. If marking a 6 vs a 7 in Zwift is going to derail your training, there are bigger issues under the hood of TR.
9 Likes
I agree that ājust be consistentā is the idea, but personally, that did not work for me. I tried to ādonāt think about it and just put what you thinkā, but that didnāt work and I felt the old AI never ālearned meā. I was never quite sure when it was going to bury me with my next workout and it was going to be my own fault for rating it wrong. I found myself having to say, ādonāt use the rating you personally think it should be. Think about whether you want the next workout to be easier, the same, or harder, and then rate itā. This left me wondering if I was leaving gains on the table. I was ready to just give up until the āChad listā came out and it made everything much easier. I can tell you that reading those descriptions, they do not align with my personal definition of those words. Itās a few years later, and I STILL have to pull it up and look because of this.
Also, I would personally NEVER rate a workout as a 1, so that number is useless and 4 and 5 are so similar that we CONSTANTLY see people posting that they took breaks or lowered the resistance and then did not rate the workout a 5, which makes zero sense to me. If you didnāt complete the workout as it was designed because it was too hard to complete, how could it possibly be less than a 5?!? No one wants to say they āfailedā, so they come up with all kinds of mental gymnastics. Having these very very clear definitions resolves all of that and makes the 5 numbers make sense and become usable for me.
tl;dr - ājust put whatever you think it isā works for a lot of people, so they canāt understand why it didnāt work for others, but that doesnāt mean that it didnāt create issues for others.
6 Likes
I think itās different personality types and any system should be designed to work for all of them. A tool Iāve seen for developing counter measures has you look at the various insights profiles and then see how it would help them. Same here, some people like rules, some like data, some just want to move on, and some go by feeling.
2 Likes