Correct, you may be bending the rules a bit. But that depends on a few things I think.
When you select and perform the elevated workouts,
Are you completing them successfully without needing to reduce Workout Intensity, back pedaling or doing other hacks to finish them?
How are you rating the workouts in the Pass feel?
I think that TR is totally open to people using the Alternate function to adjust on the fly for times when they feel great vs feel off, so they can get a workout that meets their needs on the day and still aims to follow the planned progression. Otherwise they wouldnât have given the tool in the first place.
But if you are constantly kicking it up or down, Iâd think that something is âoffâ. Could be an incorrect FTP, unexpected answers to the ride survey or some other factors.
Yeah, I was able to complete all the workouts and actually typically increased to 101% or 102% for the last interval or so, just to give AT an extra little slap and prove that itâs not better than me.
Also, I typically answered moderate, because thatâs about how it felt.
My new FTP from the ramp test is higher than my previous PR (woooo ) and I bumped up to a breakthrough for today, so weâll see. Maybe Iâll get smacked down by the algorithm for my hubris.
I think thatâs basically intended behaviour isnât it?
If I mark workouts 3 I find I donât get adaptations to the future workouts - but if I do 2 instead, seems to bump the next ones on that progression up to the next level. It seems to essentially bring the workouts forward a week and add a harder one in at the end, if your plan has a relatively consistent structure anyway (mineâs got VO2 Tuesdays and threshold Thursdays so itâs quite easy to follow what AT is doing). I assume if I started ranking 4 (or 5) the reverse would happen.
I think I did one or two 4 rankings right at the start, but recently it has been going the other way. I think really I should update my FTP but I wanted to see what happened if I just worked through the PL progressions. The low VO2 ones are not really VO2 any more.
The exception is Endurance, since AT wasnât capturing most of my endurance work. West Vidette +1 was listed as a Stretch workout, which was a bit implausible - so I did it anyway which brought the level upâŚ
The most likely outcome here is that if you try to continue this experiment into your build phase and continue to swap every workout with a Breakthrough, youâll be in no time at all.
I use alternates exclusively now as AT would obsessively try to make me do sessions that were too easy for me. I donât try and make them particularly difficult, usually around 0.4 progression a week, give or take (once my levels are dialled in).
In addition to Chad McNeese and Ivyâs answers, the part we donât know yet is how good is the AT machine learning adaptive algorithm. That is, do we push too hard in the short term at the expense of long term gains? And the slower / easier AT ramp is actually what will lead to bigger gains over the long term?
I think this is important feedback for TR. The âadjustmentsâ that users are making may well fly in the face of the planned progression that TR intends, but it points out an area for education if TR feels this will cause issues in the long run.
So, understanding why people are getting these âoverly easyâ workouts initially and the best way to handle them going forward should be addressed. I suspect, based on the comments in the cast, that this âsoft startâ is one of the things they intend and learned would be better from a progression standpoint, according to their ML research.
But like many aspects of AT, we need to know better how this is intended to work so we can trust the system and follow it vs trying to manipulate it, especially if that is likely to blow up down the line.
Also would be great to know how TR is going to monitor how well AT is doing, and at least conceptually, what KPIs (key performance indicators) they are looking at. For example:
Change in workout completion %
Change in plan compliance %
FTP changes over time
Etc.
The other part we donât know is how well with AT adapt to us. For example: if for me, I constantly rate the workouts that AT initially gives me as âEasyâ, will it learn from that that Iâm (me as an N=1 case) able to handle / ramp-up difficulty faster than the âbaselineâ, and therefore adapt to me? If so, then âoverridingâ ATâs recommendations could theorectically negatively impact / slow down how well AT learns me as an N=1 case, and its ability to prescribe workouts for me (again N=1 case) to get the best out of me.
I do agree with this. Base phases being âtoo easyâ is not a new narrative, but itâs likely we will have to reiterate the purpose of base phases and the work youâre doing there/the purpose it serves.
and then making SSB âeasierâ youâll need to do more than reiterate. Because so far saying âML compelled us to do itâ has been received with mixed reactions.
Sure. I feel like we did expand upon this a while back to address âtoo easyâ criticism and mentioned failure rates/etc, but Iâll for sure bring it up to the team that athletes are requesting more visibility to the âwhyâ behind that. Whether or not thatâs proprietary and the team will be able to share the receipts so to speak? TBD. But Iâll ask!
Apologies if this has been asked before but how does AT populate surveys? This one back in March was marked a âStruggleâ by the system and not me (I only got into the beta last night); it doesnât look like it was a struggle either
The progression levels look a lot like progress bars in a video game. When confronted with them, I naturally wanted to make them go as high as possible, as fast as possible in hopes of maxing them all out.
I donât think itâs necessary to go through all past rides and redo the surveys, I could be wrong. I have recently been added to adaptive training Too so; im taking it on from here, kind of approach. I have an upcoming FTP test so I assume all my levels will drop as long as I get a bump in my FTP.
) and I bumped up to a breakthrough for today, so weâll see. Maybe Iâll get smacked down by the algorithm for my hubris.
in no time at all. 
