This site opens with a claim: reality is not actually that hard to find. Here is the case that tests it hardest. When we judge things, we can check — measure, retest, look again. When we judge people, we mostly don't. We consult reputation instead: the compressed record of what others seem to think. Reputation is cheap to read and evidence is expensive to gather, so reputation wins by default. Most of the time that's fine. It stops being fine when reputation stops being a record and becomes a product.
My working position: reputation is supposed to be compressed evidence — the residue of what people actually saw, accumulated over time. It becomes a substitute for evidence wherever being seen as capable pays better than being capable. Under those conditions, reputation gets managed. And a managed reputation is built to survive contact with contrary evidence, not to reflect it.
Watch what happens when evidence and a managed reputation collide. Charm operates as a credibility shield: the report of what someone did contradicts the listener's own experience of them, and the personal impression usually wins. Any single incident has an innocent explanation available; the pattern only exists in aggregate, across people who rarely compare notes — and who talks to whom is itself something that can be managed. And the most effective defence is the sincere one: people who have genuinely been treated well, defending in good faith, are more convincing than any denial — without a single dishonest word being spoken in the defence. The evidence doesn't lose because people are stupid. It loses because every step of discounting it feels reasonable.
What I don't yet have is the countermeasure. A managed reputation has one real vulnerability — shared memory: people who saw the same conduct, separately, comparing notes across time. And that is precisely what modern professional life erodes. Job changes, city moves, networks that reshuffle every few years: each one hands a managed reputation a fresh start. So is the answer structural — build environments where memory accumulates — or individual, a discipline of verifying before believing? I lean structural and act individual, which may just mean I haven't worked it out. One constraint I'd hold either way: the vocabulary for all this is now popular enough that applying it to anyone merely difficult is itself a reputation attack. The claim has to stay anchored to conduct, or it becomes the thing it describes.
So that is where the document stands, at version one: a claim about what reputation is for (§2), an account of why the evidence keeps losing (§3), and a missing countermeasure (§4). Tell me where it's wrong. The version you're reading will not be the last.