Commentary

Why every program works for three weeks

The standard advice about program hopping is that you should stop. Pick something, run it, be patient, stop chasing novelty. It is delivered with a faint moral tone, as though switching programs were a character flaw.

It is not a character flaw. It is a rational response to a genuine information problem, and telling someone to be more disciplined does not fix an information problem.

The question you cannot answer

Here is the situation. You are eight weeks into a program. Some sessions went well, some did not. Your bench is roughly where it was, maybe slightly up. You feel like you are working hard.

Is it working?

You genuinely do not know. You cannot know, because "roughly where it was, maybe slightly up" is exactly what both a working program and a failing program look like over eight weeks in the middle of a training career. The signal is smaller than the noise.

Now someone describes a program that made them enormous. That is not evidence about your situation, and you know it is not, but it is the only thing in front of you that resembles information. So you switch.

Three weeks later you feel great and the numbers moved. Confirmation. Six weeks after that you are back to not being able to tell, and the cycle restarts.

Why the first three weeks always work

The confirmation is real and it means almost nothing, which is the cruellest part.

New program means new or reordered movements. You get better at unfamiliar movements quickly, mostly through learning to perform them efficiently rather than through building anything. That produces genuine early progress that has very little to do with whether the program is well designed.

It fades after a few weeks. And critically, it fades identically for a good program and a bad one. Every program produces this effect. So the thing that felt like proof was the one part of the experience guaranteed to happen regardless.

Which means program hopping is self-reinforcing. Each switch produces a burst that feels like evidence, then a plateau that feels like the program failing, then another switch. The loop is stable and it can run for years.

What the loop costs

Not what people usually say. The cost is not that you failed to complete a program, as though completion were a virtue.

The cost is the history. Changing an exercise means your old numbers on it are no longer comparable to your new ones. Barbell to dumbbell, one machine to another, a different rep range: all of them mean the previous data cannot be lined up against the next data honestly.

So every switch resets the only asset that would have let you answer the original question. After four switches you have four short stubs of data and no long run of anything, which guarantees that next time you ask whether it is working, you will be exactly as unable to answer as you were before.

The loop does not just fail to solve the problem. It actively maintains it.

Patience is the wrong instruction

"Just stick with it for six months" fails because it asks you to sit inside not-knowing for longer, on faith, without giving you anything to look at.

Some people can do that. Most cannot, and the ones who cannot are not weaker. They are being asked to keep doing something with no feedback, which is difficult for entirely good reasons.

The instruction that works is not "be more patient". It is "become able to tell". Those are different problems, and only the second one is solvable by doing something.

What being able to tell requires

Less than you would think, and more than most logs collect.

It requires recording effort as well as reps, because at a fixed weight the difference between eight to failure and eight with two left is the entire signal, and a log that only holds weight and reps shows a flat line across a real improvement. It requires comparing like with like, which means the same exercise against the same exercise and not carrying numbers across an equipment change. And it requires staying on a movement long enough for the comparison to accumulate, which is the part that overlaps with the patience advice, except now it is patience with a payoff rather than patience as a virtue.

Do that, and the eight-week question gets a real answer. Sometimes the answer is that it is working slowly and you should continue, which you would not have believed without seeing it. Sometimes the answer is that it genuinely is not, and then switching is the right call rather than a relapse.

Switching for a reason is fine

None of this says never change anything. Changing a program because you found a specific problem is not hopping, it is responding.

The difference is entirely whether you can say what you are fixing. "My chest is doing fourteen sets a week and I cannot recover from it" is a reason. "My hamstrings go nine days between sessions" is a reason. "I do not know, it feels stale" is the loop.

And if you cannot say what you are fixing, the productive move is not to pick a different program. It is to make the next eight weeks measurable enough that you will be able to say next time.

Common questions

Why do I keep switching workout programs?

Usually because you have no reliable way to tell whether the current one is working. Without that, novelty is the only signal available, and a new program reliably produces a few good weeks that feel like evidence.

How long should I stay on a program before judging it?

Long enough that the early novelty effect has passed and you have a run of comparable sessions to look at. That is closer to two or three months than to three weeks, and the specific number matters less than not deciding during the first fortnight.

Is switching programs always bad?

No. Switching for a reason is fine, and sometimes necessary. The costly version is switching because you cannot tell what is happening, because that resets your history and guarantees you still will not be able to tell next time.

Why do new programs always feel like they are working at first?

Unfamiliar movements improve quickly as you get efficient at them, which produces real early progress that has little to do with the program's design. That effect fades in a few weeks, and it fades whether or not the program was any good.