Progress isn't linear

Opening scene

We're looking at one person's twelve-week log.

Week one: twenty seconds. Week two: twenty-three. Week three: twenty-seven.

Week four: twenty-seven.

Week five: twenty-four.

And that person quit in week five.

Because week five read like a failure. Three weeks up, then a stall, then a drop. The story broke.

Now let me show you the rest of the log — what would have happened if that person hadn't quit.

Week six: twenty-seven. Week seven: thirty-two. Week eight: thirty-five. Week nine: thirty-one. Week ten: thirty-six.

So the drop in week five wasn't a glitch. It was the curve's normal shape.

And knowing this shape is a tool for persistence. Because the most common point of quitting is an expected but unrecognized plateau or drop.

Today we're building that curve's real shape.

What this lesson covers

First you'll see the curve's three features: the plateau, the drop, and capacities advancing out of sync. Then we'll talk honestly about why the plateau happens; we'll separate the known side from the unknown side. After that we'll establish the window rule needed to read a trend. Then we'll count three things that look like regression but aren't; you've seen all three in this course already. After that we'll make clear when a drop is a real signal. At the end I'll give you a three-step decision rule for the question "should I change something." We'll work through sixteen examples. By the end of the lesson you'll be able to read your own log knowing the curve's shape.

Starting point

Recall variability from Y1.01: the same person's same measurement coming out different on different days.

Recall decoupled progression from Y1.44: two capacities advancing at rates independent of each other and being recorded separately.

Now we're adding two concepts.

Mechanism

The three features of the curve

The first feature is the plateau. Weeks in which the measurement doesn't change. It's normal for a twelve-week log to have two or three plateau periods.

The second feature is the drop. Weeks in which the measurement regresses. There are several causes: within-day variability, accumulated fatigue, sleep, and total life load.

The third feature is being out of sync. Decoupled progression from Y1.44. If you're tracking three capacities together, it's almost never the case that all three rise in the same week.

So a log with three capacities doesn't show three smoothly rising lines over twelve weeks. It shows three wavy lines, and the waves are independent of each other.

That last sentence matters, and it's this lesson's point of honesty. You can't know the reason for the plateau. What you can know is that the plateau is normal.

Why the plateau happens

Let's start with the known side.

The first factor is habituation. As the same stimulus repeats, the response to that stimulus decreases. The twelve variables from Y1.65 exist for exactly this; ways of changing the stimulus.

The second factor is accumulated fatigue. The capacity you've gained is still there, but fatigue is masking it. So the measurement drops, the capacity doesn't.

The third factor is measurement variability. From Y1.01; even on the same day, two measurements come out different.

Now the unknown side: in a given person, which of these three is dominant? A log kept at home can't determine that.

And you don't need to know. Because two of the three factors resolve on their own over time; fatigue with recovery, variability with repetition. Only the first calls for an intervention, and you only make that intervention if the plateau is genuinely continuing.

This is the reasoning behind the window rule.

The measurement window

In Y1.36 I said this: measurement is done every six to eight weeks, because with weekly measurement, variability is larger than real progress.

Now we're turning that rule toward reading.

Reading a trend needs at least three measurement points. Two points draw a line — any two points draw a line; three points show a trend.

If you measure every six weeks, three points means twelve to eighteen weeks.

So what about the log you keep every session? That's more frequent, and it can be read too — but not point by point.

Here's the method: compare a three-week average with the three-week average before it. Don't compare week to week.

Concretely, back to the opening example. First three weeks: twenty, twenty-three, twenty-seven; average twenty-three. Second three weeks: twenty-seven, twenty-four, twenty-seven; average twenty-six.

So the second three weeks is higher than the first. Despite the drop in week five, the trend is upward.

That person quit because they compared week to week. With the three-week window, they would have seen a rising trend.

Three things that look like regression but aren't

You've seen all three in this course, and now they're together.

The first thing: the criterion changing. In Y1.42, someone who starts measuring plank duration up to the leak point sees their number cut in half. The capacity didn't drop; the criterion tightened. The same thing happened in Y1.49 when torso uprightness was added.

The second thing: dose reduction. From Y1.66. Someone who raises the surface height by one step writes a smaller value in their log. Because the load-bearing action is preserved, this is a dose decision; the capacity didn't drop.

The third thing: that day's state. From Y1.04. A number from a sleepless day isn't a trend, it's a day.

These three are distinguishable when they're written into the log. When they're not written down, all of them look like regression.

This is why I said in Y1.66: you record the variable value, not a quality judgment. Now you're seeing the second half of the reason.

When a drop is a real signal

Now the distinction to watch for.

A drop in one capacity: most likely noise or a plateau. Apply the window rule and wait.

A drop in every capacity at the same time: that's something different.

If you're tracking three capacities together and all three dropped in the same week, there's a shared cause. And that cause is usually not in the capacities themselves — it's in recovery.

Supporting indicators: the next-day line is bad for three sessions in a row, a change in sleep, general reluctance.

That picture is the next lesson's subject. The only thing you need to do here is tell it apart.

Summary: wait out a drop in one capacity; a drop in all of them means stop and look at recovery.

The "should I change something" decision rule

Three steps, in order.

Step one: how many measurement points do you have? Fewer than three — wait. There's no data to decide with.

Step two: is the drop in one capacity, or in all of them? If it's in all of them, this isn't a program-change question, it's a recovery question. You're headed to Y1.73.

Step three: if it's in one capacity and it's been a plateau across three points, change one variable. One from Y1.65 — and only one.

Which variable? One that changes the nature of the stimulus. Duration and repetition have usually already been tried; pace, lever arm, or input removal is usually more effective.

And a warning: the first reaction to seeing a plateau is usually to do more. Raising volume is the most common and least effective response to a plateau; because a plateau usually comes not from a lack of volume but from habituation to the stimulus.

Reading the examples

Master layer

Now let's refine this a little.

First: the three-week window method is a blunt tool, and better ones exist. More statistically reliable methods are available, but they aren't practical for a yoga log. A three-week average is easy to compute and clears out most of the noise. The principle from Y1.36: a comparable rough number beats an incomparable precise one.

Second: I said we can't know the reason for the plateau, but there's one clue. If the plateau is in one capacity and the next-day line is clean, habituation is more likely; if the plateau is in several capacities and the next-day line is bad, fatigue is more likely. So a single capacity's plateau and a general slump read differently. This distinction is the reasoning behind the decision rule's second step.

Third: this lesson's most practical benefit isn't a number changing, it's an expectation changing. Someone expecting a smoothly rising curve thinks something is wrong at the first plateau, and usually does one of two things: raises volume or quits. Both are bad responses to a plateau. Adjusting the expectation to a wavy curve does more good than a program change.

Fourth, and the subtlest point: this lesson solves a framing problem, not a knowledge gap, and we did the same thing in Y1.66. There, the obstacle to reducing wasn't mechanical; here, the obstacle to quitting isn't knowledge. People don't quit at the plateau; they quit when the story breaks. And knowing the curve's real shape fixes that story. A course's long-term effect has more to do with the expectations it builds than the techniques it teaches.

Common mistakes and misconceptions

Practice

Keep in mind

Progress isn't linear; the plateau, the drop, and capacities advancing out of sync are normal features of the curve.

A trend is read with at least three measurement points; in weekly logs, you compare three-week averages.

Three things look like regression but aren't: a tightened criterion, a planned dose reduction, and that day's state.

A drop in one capacity is expected; a drop in every capacity at the same time is a recovery question, and what you look at is rest, not the program.

When you see a plateau, the first reaction isn't to raise volume; if needed, you move a single variable that changes the nature of the stimulus.

Bridge to the next lesson

I mentioned the case where a drop happens in every capacity at the same time, and I left it for the next lesson.

Now we're getting there.

That picture has a name, and it has readable signs. And for the first time in this course, what you'll write into your log won't be about practice.

Sleep, appetite, reluctance, how long muscle soreness lasts, resting heart rate.

The next lesson builds those signs, and more importantly: separates which of them a movement lesson can read, and which it can't.