Observations

A testimony is something a resident said happened to them. An observation is a problem that is on file without that — noticed while reading the material back, inferred from an account that was about something else, or simply raised in a working session because somebody thought of it.

Observations exist because the alternative was worse. Until now the only door into this project was a dictated account, and a problem nobody happened to dictate had nowhere to go — so it either got smuggled into a policy area it did not belong in, or it was lost. Both are worse than writing it down and being honest about its standing.

The price of the door is lineage. Every observation states, in its own frontmatter and in its opening line, how it came to be here. A problem with a weak provenance that says so is usable. A problem with a weak provenance that does not say so contaminates everything downstream of it.

Rules

1. origin is mandatory, and it is a sentence, not a category. “Derived from the 10 August account, which mentions the heat twice without ever making it the subject” is an origin. “Session” is not.
2. An observation is not evidence of anything. A testimony is evidence of experience — weak evidence of prevalence, but real evidence that a thing happened to somebody. An observation is evidence only that the problem occurred to someone. It cannot be cited, quoted as experience, or counted.
3. An observation does not open a policy area. Areas still open on testimony. An observation can sit in the queue indefinitely, and most will. What it can do is tell you which testimony is worth going out and getting.
4. Promotion is one-directional and it is recorded. If testimony later arrives that covers an observation, the area opens on the testimony, and the observation is marked promoted with a link. The observation is never rewritten into a testimony — that would launder its provenance, which is the one thing this document class exists to prevent.
5. The powers gate applies unchanged. Nothing here proposes anything. Several of these note which class a lever would probably fall into; those are hypotheses written down so they can be checked, and under the evidence rules they are [UNVERIFIED] like anything else.

Why these, and not fifty

An observation costs almost nothing to write, which is exactly the danger: a list of every conceivable municipal shortcoming is not a platform, it is a mood. The ones below were kept to two kinds.

• Things the account already contains but never made the subject — mentioned in passing, as texture around a different grievance. These have the strongest lineage of anything here, because the words exist on file.
• Things the account’s own vantage point would predictably miss, identified by reading whose account this is and asking what a household in that position sees so constantly that it stops registering as a problem.

Anything that was merely a good idea about cities was left out.

Index

#	Observation	Lineage	Would open
1	The city’s own economy	In the account twice, on both sides of the argument, never opened	jerusalem-economy
2	Being told, and being heard	In the account, but split across four areas so it is nobody’s subject	consultation
3	Getting a pram past the works	Not in the account; inferred from a household it describes	construction-disruption
4	Heat, and the absence of shade	In the account as weather, twice, never as a municipal condition	public-transport or its own
5	The city in languages other than Hebrew	The account’s closing point, turned around	Nothing yet
6	What happens when you complain	The only process the account follows end to end, recorded as an aside	Nothing yet

Six observations against ten testimonies is roughly the ratio to hold. If this table ever gets longer than the testimony index, the project has started generating its own evidence, which is the failure mode whose account this is is meant to guard against.