Adding the numbers, since they settle part of this. For anyone assembling their own picture: Tmax is one to three days, terminal half-life about 165 to 170 hours, steady state at four to five weeks, and subcutaneous bioavailability near 89%. Those four numbers explain most of the questions people ask about timing.

A narrower follow-up, since the general answer is now clear:

How much of the between-person variation is pharmacokinetic and how much is just adherence measured badly?

TirzTom said:

For anyone assembling their own picture: Tmax is one to three days, terminal half-life about 165 to 170 hours, steady state at four to five weeks, and…

Coming at TirzTom’s question from a different direction. Albumin binding above 99% is the whole reason weekly dosing works, and it is also why the trough matters more than the peak. People who dose late are not losing a peak, they are letting the trough fall, and the appetite effect tracks the trough.

Worth separating that from the trial evidence, which this thread keeps folding into the same question. They behave differently and the advice does not transfer.

That is the short version; the long version is somebody else's post.

Closing the loop on my own question.

Update since I posted: I held at 1.7mg for a further twelve weeks and lost another 4kg slowly, which settles it for me. I was escalating because the number was available, not because I had stopped responding.

hans_munich said:

Albumin binding above 99% is the whole reason weekly dosing works, and it is also why the trough matters more than the peak.

Agreeing with hans_munich, and the qualification matters more than the agreement. The dose-response is real but shallow at the top. Across STEP 1 and STEP 4 the gap between 1.7mg and 2.4mg is a couple of percentage points of body weight on average, and the average is carrying a wide spread — plenty of people at 1.7mg sit above the 2.4mg mean. If a dose is working and tolerable, "working" is the relevant variable, not "maximal".