Here's a well-documented quirk in how memory sums up time, and once you've clocked it you'll catch yourself doing it constantly. You don't *average* an experience. Look back on the last couple of weeks and your brain doesn't tot up all fourteen days and divide — it files the whole stretch by two things only: the most *intense* moment, and how it *ended*.
So picture a fortnight that was, honestly, ninety per cent unremarkable — ordinary days, nothing to report — but it had one genuinely rough afternoon in the middle, and it happened to finish on a flat, low day. Ask a week later how it went and the answer that comes out is 'rough, hard going'. The ninety per cent that was fine — the bit that was actually *most of it* — gets silently deleted.
The spike and the ending do all the talking. And it's not the usual suspects: you didn't forget the ordinary days, and it isn't today's mood colouring it. It's a specific *compression* — a long, mostly-flat experience squashed down to a couple of vivid landmarks, the flat middle thrown out because it didn't feel like much and the brain doesn't store 'didn't feel like much' well.
And it's not the usual suspects: you didn't forget the ordinary days, and it isn't today's mood colouring it.
So your sincere summary is built almost entirely from the two least representative moments of the whole period. The only thing that fixes it isn't a better memory — you can't out-remember an automatic compression. It's a record that never compressed in the first place: thirty short, equal, one-a-day lines, each weighted exactly the same, *are* the flat average peak-and-ending memory can't give you.
The rough afternoon is one line of thirty, not the headline. That's why 'I think it's been a bad month' and the same month read back from the notes can flatly disagree — and the notes are the fairer witness, because they gave the ordinary days their fair share of the vote. And whatever you make of what you see is a conversation for your prescriber, not the app.