“Activist” users will always give you grief over anything. You decide how you want to react to them.
That’s a negative, Ghostrider. The pattern is full.
Do I understand correctly that all ML based work is quite far out? It might be good to lower the hype on the forum a little then. People have ridiculously high and unrealistic expectations in AT here (e.g. “giving the AI a name” thread :D).
Might be good to distinguish the “adaptive” features (good and correct name by the way) from all the “ML” features (good marketing idea by the way).
From what I understand, what AT users will get is:
- ranked workouts
- workout progressions based on compliance (questionnaire)
So I suppose there was some ML applied to rank the workouts, but that’s it so far, right?
Anything else being rolled out in the next weeks?
ML is Ranking workouts and also using 100s of features to update your plan and adapt future workouts in that plan. Right now that means it will update the workouts on a current TR plan but it won’t build a plan for you from scratch because they need more data to feed the ML before that becomes a reality. Hence the closed beta and feature roll out on the way.
So to your point, yeah some people are jumping the gun on what is actually available in the moment and what is on the horizon. Either way the potential of this is quite massive and all very exciting as it could be a literal revolutionary development for training.
See, I do not think “update your plan” will be a reality any time soon, other than choosing one of the ranked workouts based on your progression. How do you imagine this “update your plan” or “build a plan for you” to work? ML is not self-operating AI wizardry. ML can give you correlations for complex relationships. Every adaptive feature still has to be coded manually based on these correlations you find.
Let’s say a concrete feature would be “plan adapts to have more workouts of type X if I have a weakness in a skill that is crucial for my A race and workout of type X improves on that”. That must be really hard to implement and years out from becoming reality.
I think this conversation would be more appropriate for the below thread so we can keep this thread focused on CEO Communication. Here are some of Nate’s responses in that thread that can be used as some starting points for conversation. Frankly, he’d have a better response than anything I could share due to his immediate and intimate knowledge of AT as it stands right now.
Ditto. What’s wrong with good old YouTube?
I know what thaaaat means
I already added trad base for now while i wait for this
I think what @Nate_Pearson and his team are doing is great! Those little insights keep a user like myself excited about the platform and the brand, all the while knowing that it is not just for show - they are actually actively trying to make their platform better and stick with their mission statement of making us faster.
As far as knowing exact dates, they only real way to do that (and as a developer myself I know) is just set one so far out that it ends up not meaning anything anyways. Either that or waiting until you are so close to completion that you can actually give a specific date - both of these rob the user and the company of valuable excitement, anticipation and TBH marketing value.
You may not be able to please all of the people all of the time, but it looks like you can please 91% of them all of the time…
I think the main risk you have is if someone can get better data about a ride. How you parse the ride to get the information you pass into the machine learning. Which metrics are useful and which ones aren’t. Machine learning can’t really make up new metrics of how to measure a ride so it can’t create a more useful metric on its own, that has to be figured out first.
My thinking is if you can better model human physiology that will train the machine learning better.
Even saying what is in Beta could get people’s hopes up. Sometimes betas identify that the product as currently conceived doesn’t work as intended / meet “MVP” (minimum viable product), and you need to go back to the drawing board.
In the software industry, this is one of the reasons for doing controlled testing (betas, etc.) with actual customers. To identify if a product actually works they way you think it should when used by people who don’t know how the product was conceived to work, identify scenarios that the developers didn’t think of, and therefore the product isn’t ready for prime time.
In my engineering days, I actually had to demonstrate to a customer that every conceivable pathways that could happen / every part of the product spec was tested, and that every test was robust and passed. This was only possible because I was dealing with an embedded system, so I didn’t have to deal with humans doing something I never conceived of.
Nate, Thanks for the update. Keep up the great work.
Share whatever you want. Your doing a great job now with the information. When your excited about something share it when your ready. If it takes longer shit happens. If your not comfortable sharing or not ready then don’t.
Your the captain of this ship. It’s your call. You have not run us into any icebergs yet. Although a lot of us are unsure if this is actually Nate asking or the skynet AI.
Relax, your doing great!
Keep doing what you do, share what you think is appropriate. Frankly, you have no further obligation to me other than to provide access to the current product per my subscription. I am a bit taken aback that folks may be “pushing back” on your dissemination of information per your own free will. You can share what you know but running a business is not linear, things change, it’s just the way it is.
Keep it up, and thanks to all of you for taking the time to share what you feel is in your best interest.
The advantage we have over everyone else is plan vs actual. No one else has the same data set as us on that.
I think your current approach is great! From a practical standpoint, it gives us an opportunity to provide feedback on where your priorities in a way that we couldn’t if you were more close to the vest. More importantly, your approach seems to reflect your company’s culture. I think living your culture, not just saying it, is part of what makes great companies great!
Anyone in software knows you can’t ship until you are comfortable with the stability and bug count. What I like knowing is what you are working on and at least the priorities to help me make decisions based on my needs. I love transparency and honesty.
The team at TR have an amazing product and an even better team. I am a card carrying, rooftop shouting fanboy for everything that TR offers, and I’ll continue to do so with constructive criticism if its warranted. Keep up the good work!!!
You want them to make their bug tracking public info? Next they’ll have to be open source too? He already said what the major bug was that was holding things up