How do you remember what a normal week used to look like?

"Normally I see my friends every week." "Usually I cook most nights." "I go for a run most mornings." People say things like this with total confidence, and quite often the thing they are describing stopped being true months ago. Not because they are lying — because "normal" is not a fixed point they are checking against. It is a rolling average that updates itself every day, silently, around whatever they are already doing.

Your baseline is the thing that moved

When you compare this week to "a normal week," you are not comparing it to a week you actually had. You are comparing it to a mental average, and that average is built almost entirely from recent weeks. If the last two months have quietly had fewer evenings out than the two months before them, your sense of a normal social week has already adjusted downward to match — without asking you, and without marking the day it happened. The current week can feel exactly as usual as it always has, even while the thing it is being measured against has drifted a long way from where it used to sit.

This is why a slow change is so much harder to notice than a sudden one. A sudden change breaks the pattern loudly enough that you feel the break. A slow one gets absorbed into the pattern itself, one week at a time, until the pattern is describing something new and calling it the same old name.

Why the update never announces itself

There is no single day on which "I see friends every week" became false. It happened the way most drift happens: one week ran long and a plan moved, the next week there was a good reason too, and by the third or fourth week the gap had simply become the new shape of things, absorbed without anyone deciding it on purpose. Memory does not file this as an event, because there was no event to file — no fight, no falling out, nothing to hang a date on. It is just an accumulation of ordinary weeks that, added together, look nothing like what you would call normal if you actually laid them out.

Ask someone when they last had a proper evening out with friends and most people cannot answer with a date. They can only offer a vague sense — "a while, I think," "not recently, I guess" — because the honest answer requires comparing a specific week against a specific week, and memory does not naturally keep specific weeks lying around. It keeps impressions, and an impression is exactly what "normal" is built from.

The comparison you actually need is dated, not remembered

The fix is not trying harder to notice, because noticing is precisely the thing a slow-moving average is built to prevent. What actually catches it is comparing two specific, dated stretches of time against each other — this month against a month from a year ago, this run of weeks against the run before your schedule changed — rather than comparing the present against a fuzzy sense of "usual." A dated comparison cannot quietly redraw itself around your current habits the way a mental impression can, because it is anchored to two fixed points instead of one moving one.

This only works if something was actually written down at the time, in something closer to real time than in hindsight. A week reconstructed from memory six months later inherits the same drift problem it was supposed to solve — you would be comparing one impression to another impression, both shaped by whatever you currently believe was normal. What holds still enough to be useful is a record made while the week was still just that week, not yet folded into an average.

What this looks like in practice

Someone with a few months of dated entries behind them does not have to ask "have I been seeing people less" and hope an honest answer surfaces. They can put a month from this spring next to a month from last spring and read, plainly, that one had nine evenings out and the other had two. No argument with their own memory required, because memory was never the thing being consulted — the record was. The drift that felt impossible to pin down becomes a difference you can just look at.

The same holds for the opposite worry — the sense that things have gotten worse when they might not have. A remembered "normal" is vulnerable to whatever mood you are in when you go looking for it; a bad week makes the whole recent past feel bad by association. A dated comparison does not have a mood. It just shows you what was actually there.

About TimeMap

TimeMap is built around exactly this kind of comparison. Every entry carries the date it was written, so instead of asking your memory to hold a fair average of "normal" — which it was never built to do — you can put two real stretches of time side by side and read the difference directly. Not a guess about whether things have changed. The actual weeks, next to each other.