Brian Davis also made a video where he explains why he quit Trainerroad.
OmG aNoThEr cLiCkBaIt yOuTuBeR, probably just trashing TR to drive his business like DJ 
3 Likes
Good for him! He states that he is not price sensitive and hence gets a coach to get an even better experience which ultimately boils down to: do you want to pay close to 200 bucks for a year of TrainerRoad or do you want to pay 5,000+ per year for a decent coach. Hiring a former pro (eg of Tim Cusickās training business) sets you back 18k per year. If you are lucky and make it off the waiting list. 
It will be interesting to see what the adaptive planning will end up delivering. If it works it will outperform many of those coaches. Perhaps Brian can then come back to TR and stop selling socks? 
2 Likes
Did you actually watch the video, bro? Brian Davis does not earn his living with youtube, training software or coaching?
1 Like
Did you watch it to the end? Like the last minute where he heavily advertises his sock gig? How he earns the rest of his living you can see in the description.
1 Like
touchƩ
1 Like
Around 4:30 in the video he says that āwe already have the technology to do this: itās called a coach. Theyāre humans, with brains.ā
Zooming out from this instance of a discussion weāre having, I think people with this mindset are going to be very, very surprised over the next 1, 3, 5, 10+ years. Iāve seen some applications of ML that are amazing. Implying that a human can understand volumes of data like these is akin to saying that a human would always be faster at math than computers in the 50s or 60s.
Is TRās ML application perfect? I am positive it is not. Will their system beat all athletes and coaches? I am positive it will not. But broadly speaking, these systems have access to things humans just canāt identify (with support.)
EDIT: I think coaches still have massive value. When he talks about how he communicates with his coach, probably feels more of a sense of comittment because otherwise he has to explain why he didnāt complete it, allllll the other incidentals (bike handling, food, etc etc). There are still huge things that a coach can add into the discussion, and arguably for some people, will probably get a better outcome. But in terms of using trends to identify actual training vectors? Eh, ML wins and will continue to win that one.
7 Likes
5000/yr will give you a very reputable coach. Not arguing with the point of your post.
2 Likes
Spot on mate! Our company was in a similar situation with our branches. When a small team worked on AI and ML they were ridiculed. Not so much after their profiling suggestions outperformed those of the humans who thought they know their customers so well. 
Absolutely, a coach can and will always have their place. Though I guess it will come down to the price point. IOW, do I want to pay 20 or 500-1,500 bucks a month and how much of a difference will it make for us non-professionals.
As far as I am concerned I would rather hire a dietician and strength/mobility coach rather than a coach who tells me to ride 70% in Z2 and then some more in Z3, Z4, Z5, and Z6.
2 Likes
Doesnāt it bother you that you have to label your workout activity using the adaptive training feature? The labelled dataset of all workout activities is more valuable than the income generated by subscriptions. I would not want to pay additionally for supplying training data for machine learning.
Yes it will, many decent coaches have a 400ish monthly rate. Though the real experts (eg former pros and pro coaches) will charge quite a bit more. It certainly is tempting to see what difference they would make. 
1 Like
As far as I am concerned TR can know that the 10x3@120% felt like a total hip replacement. All I care about is that they get me to a point that I can get such workouts done. Ideally with an FTP above 5 wpkg.
Regarding your data privacy point: TR is stating that they wonāt sell our data. If they change that fact they will have to notice us and we can decide whether we want that or not. If we donāt agree, we can challenge them to delete our data. That at least applies to the people in Europe. So no real issue for me either.
Spot on!. Itās like paying for duolingo.
2 Likes
Breaches against the GDPR would bring TR out of
business within days. Especially breaches of fraudulent nature (eg selling our data behind our backs).
Forget privacy for a second. What Mr @HansDampf is alluding is unpaid labor.
I was trying to match the intellectual level of other vocal arguments against DJ made here.
No, itās not. He stated that the data of assessed workouts are more valuable than the subscription. Though TR could only materialize on the information if they were able to sell them. Something they cannot without disclosing it.
So if they donāt intend to sell the data it will only be used for better training frameworks which directly benefits me. So I click on 2 fields and eventually get a better product for a fixed price in return. Sounds fair to me. 
8 Likes
Not only that. To the the extent the model improves and becomes a lot more valuable, but that value is captured only by TRā¦Despite the thousands of hours the collective of users have spent manually feeding the model.
Though the users benefit directly from that effort. More data, better AI and ML, better training, bigger watts and a fixed price.
2 Likes
Donāt you do this already? After each training session you evaluate how it went. At least I do so, to be sure and try and understand my body and my training better. Helps me adjust my training in time and not over or under train. Now I use the annotation box for this, when the system updates I am going to use the forms. I learn, I get better training and trainnerroad learns. Great in my opinion.
When I go to a coach and he or she asks me how the training went so they could help me better. I also aināt going to say Iām not giving you this info because I pay u and you might also learn from it.
2 Likes