THE UNDERLAYER · TREND, NOT DAY
On the slow drift nobody notices, and on why a single reading is worth almost nothing.
A bad afternoon announces itself. It has a cause, a shape and an end, and everybody involved can point at it.
The other thing does not announce anything. It is the general level you have been operating at for eleven weeks, which has moved perhaps fifteen per cent, and which feels like the way things are rather than like a reading.
The second one determines much more and is much harder to see, for the same reason: you have been inside it the whole time, so there is nothing to compare it against.
One looks only at spikes. The other treats every fluctuation as a trend.
Attending only to what is loud. The dramatic day gets a full investigation and eleven quiet weeks of drift get nothing, because nothing about them was ever notable enough to look at.
The opposite: a single bad reading treated as evidence of a trend, which generates a response to something that is noise. One point is not a direction.
One reading at this level tells you almost nothing. Thirty tell you a great deal, and they tell you something a single reading cannot in principle contain, which is direction.
That is not a deep observation. It is just what a noisy signal requires, and it is the reason this station is weekly rather than daily.
The Daodejing keeps returning to the claim that difficult things should be dealt with while they are still easy, and large things while they are still small.
Read as counsel it sounds like prudence. Read as a claim about detection it is sharper: the window in which something is cheap to address is the window in which it is nearly invisible, and those two facts are the same fact.
The text does not supply a method for seeing the small version. That is what the two-axis mark below is for, and the method is modern.
Core affect is the relevant construct: the claim that moment-to-moment feeling can be usefully described on two dimensions, roughly valence and arousal, underlying whatever more specific emotion is named on top.4 The two-dimensional structure is one of the better-replicated findings in the area, though whether it is the right structure is argued about at length.
The practical finding is about sampling. Experience-sampling research consistently shows that retrospective reports of how a period went diverge substantially from the average of readings taken during it — end points and peaks are weighted far more heavily than duration.5
What that supports is narrow and useful: your impression of the last month is not a summary of the last month, and a daily mark is a better estimate than a memory. It does not establish that the two-axis mark measures anything deep. It measures what you wrote down.
The heroic material is absent here and the absence is informative. It has an extensive vocabulary for crisis and none for the unremarkable stretch, because the unremarkable stretch is not where anything is decided in that world.
The Stoic material is closer but aimed elsewhere — at the judgment inside a state rather than at the level the states are running at.
The station is genuinely modern. It requires the idea of taking a reading repeatedly and treating the series as the object, and that is a habit of mind that came from measurement rather than from reflection.
The objection is that some things get worse when watched.
The case against
There is reasonable evidence that self-monitoring of mood can increase attention to negative states in people already prone to that, and the tracking literature in mental health is honest that measurement is an intervention rather than an observation.
It also generates an artefact: a person who marks their state daily will notice patterns in noise, respond to them, and then read the response as a finding.
The mitigations are dull and they work: one mark, not a journal; weekly review, not daily; and a rule that no single reading justifies a change. If the marking is making the states more present rather than more visible, stop.
Two axes, one mark: how pleasant, and how activated. Ten seconds. Not a paragraph — the paragraph is the failure mode from the previous essay.
Review weekly rather than daily. The daily number is noise and looking at it invites responding to noise.
Hold the rule: no single reading justifies a change. Three weeks of drift in one direction does.
And check one external anchor against it — sleep, hours worked, how much you have been outside. A trend with no candidate cause is much less useful than a trend with one.
This is the longest block in the collection and the length is the point.
The loud day has a cause and an end. The drift has neither, which is why it runs for months without being noticed by the person inside it.
Thirty marks give you something a memory cannot: a direction, uncontaminated by what happened most recently.
And the cost is real. Tracking is an intervention, it makes states more present as well as more visible, and if the first effect is winning then stopping is the right call rather than a failure.
One mark a day on two axes — how pleasant, how activated. Ten seconds, not a paragraph. Review weekly, never daily.
No single reading justifies a change; three weeks of drift in one direction does. And if marking is making the states more present rather than more visible, stop.
Research is cited by author and publication, with effect sizes and replication status where they matter. Where a claim rests on a single study, the essay says so.