I think the fact that in this forum (or anywhere in cycling) we discuss topics like we are in the top 1% of the athletes is somewhat of a funny way to look at the issue in hand. 90% of the people in TR and Zwift are far from being Pros. In reality, the large majority of us are faster than the average human but are far from that top 10% potential. Therefore, anything that keeps accountable, has a plan and pushes my limits in a controlled matter will bring about improvements. Sooner or later, we all hit a glass ceiling and stagnate and then a change is needed.
I think the main point of conversation here should be “Value vs. Cost” (Value proposition). ML is not close to replacing coaches right now but it will get closer everyday. Right now they provide a large portion of the Value of a coach (50%? 60%? 90%?.. that is up to you) at a fraction of the cost of coach (about 10%). Therefore the value/cost proposition becomes highly favorable to most people. The new ML learning is ONLY increasing the Value proposition and this is where Coaches can only react in two ways:
1 - Increase their own value proposition (more monthly emails? one on one coaching? realtime coaching while training?)
2 - decrease their cost so that they can compete with TR and ML.
Making a youtube video about how much value you provide… its a marketing tool…probably not drastically changing the value proposition.
I really think this is based on noise on here, rather than any real stats. There’s a bias on who would come on to the forum in the first place (my anecdotal evidence of fellow TR users I know in the real world is they wouldn’t really visit here). Then there’s further bias which is across the internet that people are far more likely to post about issues, than everything just working fine.
I’m certainly not the right target audience for a human coach - I just want to stay in shape and enjoy the occasionable Cat C race on Zwift. And I’m not doubting that you find value for what you pay, and it fits your goals / priorities. Maybe you’re farther long the ‘getting the last XX% of potential’ than I am.
But I’m still curious - what ‘weekly adjustments’ are required that you would need to pay someone $200 a month for, that you couldn’t do on your own? I’m not being snarky, I’m seriously asking - what does a coach do that you can’t or could’t? What discussions do you have, what training insights do they provide that you find it worth to pay the coach for? What does a coach provide that TrainerRoad and WKO+/TrainingPeaks can’t?
Or - put another way, describe TrainerRoad’s business model
I’m sorry to read about that coach you were using. Personally, there is no way I would ever feel right about charging someone $250/month, looking at their training once a month and allowing them to ask questions twice a month. That’s not the kind of coach I would want to be. My coaching thus far has been completely free to my athletes, I update plans weekly, take their questions when they ask, and review workouts at least weekly. These people are my friends, and I’m learning from their experiences.
The main reason I am not “in business” right now is because I would be doing a disservice to athletes because I have too much other stuff going on to pay full attention to their training and give them a quality product. Your former coach’s idea of full-service coaching and value of $250/month is pretty vastly different than mine.
I’ve directed a number of my club mates who’ve inquired into my coaching them to TR because it’s a great value IMHO. But there are limitations within TR’s workout library, I don’t like the construct of many of the workouts, and ML is going to be hard-pressed to grasp the “art” behind designing and prescribing training to athletes.
But for a lot of users and athletes, TR is plenty, and AI/ML/AT probably makes it better for them. I just bristle at the idea that Tim Kusick or Kolie Moore or Brendan Housler or Jim Vance or any number of other coaches are “scared” of TrainerRoad in any way. The coaches who aren’t good at what they do, over charge and under deliver? Yeah, they probably should be scared, but that’s not because of AT… it’s because their product is poor and they likely already know that.
Hmm, I wasn’t looking for getting the last XX% of potential. Second I did better self-coaching for a year by following the principles in Carmichael’s Time-Crunched Cyclist book and cherry picking workouts from Strava CTS plans than two years with TrainerRoad’s plans. And third, doing an outside ride or two along with going to the gym twice a week for spin class (40 minutes of HIIT plus 30-45 minutes warm-up/cool-down) got me to nearly the same fitness as doing TR plans.
So for starters, I was ready to go back to self-coaching armed with only my one-time purchase $20 book and my one-time purchased FasCat off-the-shelf plans from TrainingPeaks. However I’m in my late fifties and needed help integrating resistance training with cycling 7-10 hours a week. And a year ago I tried a one-time purchased plan (FasCat 18 weeks of sweet spot), saw better results vs TR, and that was reason for hiring one of their coaches. With six months of coaching I can confidently report my workouts look nothing like TR workouts, my fitness is higher than when doing TR, and I’m able to do resistance training.
To be honest I’m spoiled now and enjoy not having to make my own coaching decisions. But if money was tight, I’d try economizing in other areas of my life as my coach is still teaching me more than I ever received from listening to any podcast or from studying TR plans. Just based on six months of coaching, if I had to drop the coach I’d still go back to that $20 book based on what has delivered more results because everything I’ve done - Stages power spin classes, TrainerRoad, FasCat one-time purchase plan, CTS book - has delivered results. It is simply a question of which has delivered more.
Objectivity. I have coached myself for years. I have been coached for two years. I learned a lot from my former coach that I now apply in my own coaching. Objectivity is huge. We make a lot - A LOT - of mistakes when we coach ourselves because we are inherently biased. It is incredibly difficult to separate ego from the data.
I completely agree that good coaches won’t be scared of AI/ML/AT; in fact if anything they’ll be eager to utilize any tool that makes them more effective.
Wouldn’t say laughable, hard to judge that sort of thing and very much a matter of whether you take life at face value or read more deeply into people.
I would say about Brendan though (even though I thought there was a bit of desperation on that video and he definitely misrepresented Nate), is that he is a talented athlete but also confident in ways many of us aren’t fit to judge. Won’t get into that too much as it can turn political, but criticism of the video above aside, I think he deserves a lot of respect on the non-cycling aspects of his life.
Also, at how much EVOQ charges (one month is more expensive than a year on TR), they should be more confident in their service. They should spend more time thinking of how to adapt and innovate instead of wasting time making these “rebuttals.”
Don’t you think another bottleneck is the ML talent?. I work in finance, in High Frequency trading. I’ve seen very talented teams trying to use NN for prediction and the results are underwhelming. I doubt TR is paying 2mm in salaries for that team. So I really don’t expect much from TR effort. But, let us hope they prove me wrong. H
Kurt, you’ve used the term “scared” several times there, and placed it in quotes as if either I wrote that or Nate Pearson said that, but neither of us did. I wrote “threatened” with good reason: probably all businesses and roles are threatened by technology in some way or other, while at the same time technology opens up vast opportunities for those able to embrace and leverage it effectively.
I’m confident that, per Marc Andreessen, software will continue to eat the world, and at an accelerating pace, so there’s no reason for me to think that coaching is some uniquely special industry or field that will not, in the by and by, also be strongly impacted and face considerable change…
Of course, Nate was in marketing mode, and at any point in time the claims of software salesman and marketers are nearly always (well, well) ahead of the reality. I actually thought he was very restrained for a software CEO, certainly compared to the Olympic-class bullsh!tters I’ve known . But, the software and capabilities will keep improving, from TR, its competitors and new entrants yet to emerge, such that great progress will be made and relentless change ushered in.
I think the underlying mechanism supporting the Rudi Dornbusch quote (insight) that I used earlier - "In economics, things take longer to happen than you think they will, and then they happen faster than you thought they could” - is that us human beings are notoriously poor at properly grasping the effects of compounding, whereby repeated shorter term incremental progress are transformed by the passage of time into massive longer term impact. I think we’ll see that play out here, and I’m fascinated to be (patiently) watching.
ㅤ
The thing is, I have the ‘benefit’ of now knowing who this guy is or his background. What you say may be completely true if I really knew him and his background/experience. I was merely stating the way he comes across in the video he posted. He first reveals the details about a private conversation (say what you will), and then takes Nate’s comment completely out of context. I watched the TR YouTube vid, and I didn’t get the impression at all Nate was saying this was going to replace all coaches - particularly in the near future.
The other issue is this guy seems to be offering a ‘premium coaching experience’. I really don’t see his world overlapping with Nate’s world anytime soon - seems pretty obvious to me sitting on the sidelines! Like I said, he comes across as lacking confidence about his own profession right or wrong.
At a basic level a coach is like any other service. You can DIY but that is true for a whole host of things. Some people mow their own lawns some have lawn services. House cleaning, financial planning, basic car (or bike) service, its all the same question.
I think this is a great example from a another field that is many years ahead in the ML/IA curve.
I get the feeling from some of the people in this thread that they think in 10/15/20 years that a large proportion of ‘good/great’ coaches won’t be using AI/ML in their coaching. I can understand the view that the current and near future AT won’t impact coaching/coaches much, but I can’t fathom thinking AI/ML won’t have a dramatic affect on this field in the next 1-2 decades.
ML is really good at pattern matching, and I think that in large part identifying what workouts work best on both a population and individual level are largely pattern matching problems.
One said, “lack of confidence in his own profession and what he brings to the table”, others have said “desperate”, and you yourself used “threatened.” Not sure why exactly we need to parse the differences in those words, frankly. The implication is that coaches are afraid that TR’s ML/AI/AT will take their business away. In 20 years, maybe.
As mentioned, I’ll stay and pay my annual fee and check it out. It’s interesting, but based on what I have seen of it (and not personally experienced yet), I don’t think it’s going to be a game-changer for me. That said, I think it will enhance the overall TR experience for a lot of their users, so kudos for that.
Personally, I didn’t really care for Nate’s comment in the podcast. I think he invited this kind of response with that statement, even if Brendan did mischaracterize what he said. Good for Nate for being confident in his product. Good for Brendan (et. al. - he’s hardly the only coach that’s questioned ML/AI since this announcement) for being confident in theirs.
This is no different than any other industry and the leverage of data. You can see it as a threat and double down on the way you have always done things, or you can leverage the new abilities to hive off the parts of your work that can be automated so you can focus your attention on providing the services that data cannot.
If you’re truly a good coach, you should welcome this type of advancement.
Hmm - interesting point. Although maybe not quite relevant; people would normally hire house cleaning or lawn service etc mainly because they -don’t- want to do those things. I’m a data geek and a nerd so financial planning and geeking out over my training plan and numbers, I kinda enjoy that so not that applicable.
Maybe financial planning is the closest analogy - you could self-teach yourself financial planning - it’s stupid easy - but the planner can ‘keep you accountable’, and maybe you’re not interested / don’t find it enjoyable so you don’t want to spend time on it.
Apples and Oranges. I’ve worked in your neck of the woods for decades, far, far too many variables in play for DNN to be useful on a consistent and reproduceable way just yet*, especially since in general strategies are only successful for as long as nobody else knows about them; any strategy that did show directional accuracy would be immediately arbitraged away.
On the other side of the spectrum - I’m CFO of a company using AI/ML with career-altering impact; it’s laying waste to entire sections of this particular industry. It’s a well-defined constraints optimization problem at this level and it’s pretty amazing. I think what TR is trying to do is somewhere in the middle of our two extremes - human biology is really friggen’ hard, but the data sets themselves are tightly defined with generally high-quality (power) data.
*My favorite HFT story: A few years ago researchers thought they had discovered particles traveling faster than the speed of light. Turns out it was a maths error, but for a while the HFTs got really really excited at the possibility that they could set up trades to occur before the trades happened.
. But, the software and capabilities will keep improving, from TR, its competitors and new entrants yet to emerge, such that great progress will be made and relentless change ushered in.
