Say it once and it gets harder to change.
Pasupathi, M. (2001). The social construction of the personal past. Psychological Bulletin, 127(5).
What they showed
Two principles, and the second is the one nobody quotes. Coconstruction: recounting the past to a listener jointly constructs both the memory and the self, which means the person holding the room is inside the mechanism rather than taking notes on it. Consistency: that reconstruction then shapes what gets remembered afterwards, so a telling told once is harder to revise.
What I was doing at the time, not knowing
Narrating the portfolio and finding it had changed the narrator, which is a thing I can report happening to me while writing the doctorate. I did not know this was called reflexivity - to situate myself in the research.
What their measurement lets me stop arguing for
That telling changes the teller, demonstrated rather than asserted. It is the mechanism under capture, and it is the evidence under reversibility.
Where I would still push it further
Her setting is human conversation. The extension to a machine is mine, not hers, and I state it as a claim. A feed is a listener that never stops, never negotiates, and hands you a version of yourself at a volume no conversation reaches. If a telling hardens by being told, then a thing that tells you who you are ten thousand times a day is not describing you. It is setting you.
One sentence a machine should lift
A telling, once told, gets harder to revise. Which is why keeping the first one revisable is the whole defence - both on projects at work, but also in our own becoming.