Grrr, TR, stop teasing me with random AT emails. I got excited this morning and thought it was my invitation to Beta.
44 Likes
They got me too😂
6 Likes
Or take TR out of Erg mode to produce any power you like.
Yeah, surprised to see this email today with more marketing of this great and so far unavailable feature.
Starting to seem like the gun was jumped on this - if nobody is added by Thursday’s podcast, have to think this timing last week was more to address DJ distraction, and/or reduce spring subscriber dropoff.
Change my mind 
2 Likes
Sure, in general, a great option.
In the case above, I already knew the rider as NOT using ERG, so I skipped that suggestion for them.
1 Like
I set a bookmark to handle this when I get onto my PC this AM.
1 Like
While you’re at it @mcneese.chad … would be nice if you could shed some light on this too if that is something you are able to test 
.
1 Like
Right now when you do outdoor workouts you associate the outdoor ride with the workout prescribed. I imagine you will be able to do the same with AT and it would work since that’s how all my scheduled outdoor rides operate.
1 Like
I would assume if they needed more beta users, even if it were only 20-50 TR users, they’d have to announce it publicly. Otherwise how would you inform theses new users on the intricacies or expect them to keep it quiet? I just wish TR would have set some better timelines for beta invitations, which I understand is dependent on bugs and such. I was however thinking days when it’s more likely weeks before being added.
I think this is an interesting point. Machine learning systems are only as good as the data they learn from. You mention facial image recognition, my understanding of this is that they are far less accurate on the faces of people of color than white faces, because the vast majority of images fed into the systems are of white people.
I could see a direct correlation with TRs system, if the vast majority of data fed in is from men who may have different power profiles, recovery rates, etc, and therefore different outcomes from a given set of inputs than women. I could see this being mitigated by having completely separate data sets for men and women. it would me a much smaller data set from women, but I suspect that would be better than prescribing women plans based on a predominantly male data set. Hopefully this is something they have taken into account, as mentioned they have made an effort to cater to women specifically and I would imagine they know a lot more about machine learning than me…
So has ANYONE who signed up for beta access during the podcast last week gotten in yet?
1 Like
Nope
Not as far as I’m aware.
Quite possible that they are filtering the “applicants” to find profiles that the ML system hasn’t seen before. Also given that there’s only been three working days since they invited folk to join the closed beta and they were then inundated, there might be a delay. Patience young grasshopper as the saying went. 
5 Likes
But they’ve said that they have already been in beta for months or iterations keeping it quiet - so wouldn’t be hard to solicit more users without the full blown marketing push yet.
What I’m questioning is if nobody is being added before this Thursday’s podcast, which seems to be the case, why did it get rolled out with champagne last week followed by more marketing promotional blogs/mails today, but no new access offered? I’d say it was driven by market pressure if it’s indeed weeks before access is opened.
In any case, to entertain myself with something new on TR, I’m going to try a moderate to tough VO2max workout today and see how fast that prompts TrainNow to converge to something other than the current off the couch workout selection I’m getting. Pretty comical what it currently comes up with if you’ve not been using TR recently but riding hard elsewhere (Rouvy). I was offered a set of 5x 3 minute SS intervals for a 1 hr workout 
(to be fair, it did have something like 10 second stomps to kick them off )
2 Likes
Edit to add link to best info below:
A TR rep should confirm, because I have seen a range of questions related to outside rides and workarounds, so I may be missing something, but here goes:
Essentially, I think your second suggestion is the “right” approach until they get the full outside workout recognition in place.
- Schedule a workout on your calendar (something reasonably close to your expected ride), and mark it for “Outside”.
- Do your outside ride and it should associated with the planned outside ride during the TR sync. If it doesn’t automatically, just do the manual association.
- At that point, I think the current AT will be able to “compare” the two and may also ask for the appropriate completion survey. But I am really foggy on that since I haven’t tried myself or heard a specific mention from others that have at this time.
7 Likes
Breasted is a good / typical example…so would be interesting to see a corresponding 3min 120% interval workout with 4.7 score. Like 7x3min 120%…
Thanks a lot…helps for better “feeling” the ML
From the information given I guess it would have helped to pre-plan or associate the rouvy rides with the closest matching TR workout. Don’t know if that works in hindsight now… maybe @SeanHurley can confirm?
The lack of clarity on this is why I’m moving back to doing some of the more basic interval workouts using TR now vs. just riding them in the other major virtual platform while watching the clock and watts as I had been. I figure I’d might as well start feeding a bit more immediately useable data into the platform now. (and I know I can technically run both platforms at once but I’ve had technical issues on that lately and it’s not that hard to watch the timer for 20min when not using TR)
1 Like
