# Remembered As

> A peak-and-end memory chart for one everyday experience - a trip, a day out, an evening - that
> already happened or is still ahead. Split it into up to ten moments, give each a feeling from -5 to
> +5 and a number of minutes, and a free in-browser engine works out how it was lived (feelings
> weighted by minutes) against how memory tends to keep it (its most intense moment and its last one
> averaged), with a verdict: sweeter in memory, harsher in memory or true to life. A draggable feeling
> strip, an ending what-if and a snow globe that holds only the peak and the end. A metered keepsake
> reading names what memory leans on, what the ending does, what memory mostly leaves out, one small
> change to the ending and a caption. Derived from @owl-listener/peak-end-rule.

https://remembered-as.skillsafe.ai/

## What it does

- The numbers, in a fixed order: lived average (every moment's feeling weighted by its minutes); peak (the most intense moment, high or low; a tie goes to the longer moment, then the later one); end (the last moment); kept score (peak and end averaged, or that one feeling when the peak is the end); gap (kept minus lived).
- Verdict from the gap: +1.0 or more is sweeter in memory, -1.0 or less harsher in memory, anything between true to life.
- The ending what-if: try a different feeling for the last moment and see the kept score and the verdict move before changing anything.
- Free downloads: the keepsake drawing (SVG, PNG), the moments (CSV), and everything with the reading (Markdown, JSON).
- Import a list of moments from a text file (one a line: what happened, feeling, minutes), and share the keepsake picture with its summary from the device share sheet where the browser supports it.
- A moment counts as at most 720 minutes (12 hours); a longer one is read as 720 and the page says so.
- The keepsake reading is not run on an experience that mentions hurting yourself or someone else; the page points to a helpline instead.
- Saved keepsakes (opt-in) keep the moments and the latest reading, and show how the verdict moved between runs.

## The paid task

One task, `keepsake`, on gpt-terra, metered per run. The input is four string fields: `task`, `experience` (name, whether it already happened or is still ahead, and a note), `moments` (one a line with its feeling and minutes) and `facts` (the page's numbers). The reply is one JSON object: verdict (sweeter-in-memory | harsher-in-memory | true-to-life, always the page's), headline, peak_note, end_note, left_out, next_time and caption. The page re-checks the reply: the verdict and the headline's numbers must be the page's, each note must name the right moment, the change for next time must not set a day or time, and the caption stays short.

The idea is a rule of thumb from psychology research about how people look back on experiences, not a law. A thinking aid for everyday experiences, not professional advice.

- API tutorial: https://remembered-as.skillsafe.ai/api.html
- Source skill: https://skillsafe.ai/skill/@owl-listener/peak-end-rule (owl-listener/designer-skills, MIT)