Building Blocks · Training · Chapter 11 of 11 · 3 min read

Where the model stops

Ten chapters have described a model. A model is a simplification that is useful because it leaves things out. It is only fair to end with what this one leaves out, and what you can do about it.

What the model can't see

  • Very short and very long efforts. The curve in chapter 1 fits best between about two and twenty minutes. BlockRide measures your peak power instead of calculating it for that reason. For rides of several hours, the curve says you could hold threshold forever. You can't.
  • How you recover. The battery refills by one formula for everyone. Skiba's research shows real riders differ. If your battery seems to come back faster or slower than the screen says, you are probably right.
  • Tired legs. Power late in a long ride falls in a way neither the curve nor the battery describes. Researchers call it durability, and they are still working out how to measure it.
  • Your response. Load is the dose. Fitness is an average of doses. Neither is a measurement of how your body answered. Two riders on the same plan end up in different places.
  • Heat. Indoors, without wind, the same watts feel harder. In Mieras's study riders held clearly more power outdoors than in the lab at the same perceived effort, and the difference was cooling. A good fan is worth more than most upgrades.
  • Life. Health signals catch some of it. They don't know about the deadline at work, the cold you are starting, or the child who woke you at four.

How BlockRide guards against its own errors

  • It measures where it can. What you can do comes from your rides, through stack-ups, not from a forecast.
  • It corrects in small steps. A week can grow only so much. A feel answer moves the next sessions by a small, capped step. Health data needs several signals to agree.
  • It recounts on the server. The battery is replayed from the saved ride, so a glitch in the game can't raise your profile.
  • It shows its reasons. Every changed day says what changed and why.
  • It keeps AI out of the numbers. AI writes the coach's sentences after a ride. Formulas and fixed rules decide the training.

What you can do

  • Go all out sometimes, at different lengths. The profile is only as good as your best efforts.
  • Answer "how did it feel?" It is the one thing the power meter can't tell us.
  • Keep your weight up to date. Climbing speed and watts per kilogram depend on it.
  • Import what you do elsewhere. A plan that can't see your Sunday ride will plan Monday wrongly.
  • Overrule it. You can make a week easier or harder, leave a session out, change the goal or end the plan. The model advises. You decide.

About the numbers we didn't print

These chapters gave the formulas that science has published and left out BlockRide's own settings: how fast a week may grow, how far a light week drops, where one tank ends and the next begins, how many signals it takes to ease a day. That is partly because they are our work. Mostly it is because they change. They are tuned against real rides and adjusted as riders tell us what was too much and what was too little. The shape described here will stay. The dials will keep moving.

Start at chapter 1, or open the Training tab on your account and look at your own three numbers.