Testimonies

First-person accounts of living in Jerusalem that anchor the platform’s policy work. Policy here starts from lived experience, then gets stress-tested against data, against comparator cities, and against the question of who actually holds the lever — not the other way around.

Ten captured, all from one dictated account on 10 August 2026 given across five sittings, and between them they open the project’s first eight policy areas. Who gave the account, and what it cannot see, is set out in whose account this is.

Rules

1. Only cleaned testimonies are committed. Raw dictations, recordings, and transcripts stay out of the public repo. testimonies-raw/ at the repository root is a gitignored staging area, deliberately kept outside the published content tree — drop raw material there for processing.
2. Every testimony is dated and named YYYY-MM-DD-<policy-area>.md. Where an area gains a second testimony on the same date, a short descriptive suffix is added — 2026-08-10-construction-disruption-pedestrian.md.
3. Cleaning preserves voice and facts. Editing removes disfluencies and tightens prose; it never adds claims, softens them, or changes their substance.
4. Redaction before commit. Full names of private individuals, exact addresses, and identifying details of third parties are removed or generalized. Municipal testimony de-anonymises unusually easily — a street, a building, a school and a date together identify a household — so the default is to generalise to the neighbourhood.
5. Frontmatter is mandatory: title, date, policy-area, status, source.

Use the /capture-testimony command from the municipal-policy-assistant plugin to process a raw account into this format.

Testimonies are not the only door

A problem that nobody happened to dictate used to have nowhere to go, and would either be smuggled into a policy area it did not belong in or lost. Since 10 August 2026 it goes to observations instead: problems recorded without a first-person account behind them, each stating in the open how it came to be on file.

The boundary is strict in both directions. An observation is not evidence of experience and cannot be cited as any. A testimony is never derived — if the words are not a resident’s, the document is an observation no matter how well-founded it is. And an area still opens only on testimony, which is what stops the project generating its own evidence base.

What a testimony is for, and what it is not

A testimony is evidence of experience, not of prevalence. One resident’s street proves something can happen; only data proves it is common. At city scale that distinction is sharper than it is nationally, because a single neighbourhood’s problem can be completely real and completely unrepresentative — and the research stage is expected to find out which.

Testimonies also carry a second payload that is specific to municipal work. Every account contains assumptions about who is responsible for the thing going wrong, and those assumptions are frequently wrong in an interesting way. /capture-testimony extracts them as a separate list of attributions to test, which goes straight to the jurisdiction map. The gap between who residents blame and who actually holds the lever is one of this project’s more reliable findings, and it is not treated as an error in the testimony — the testimony records what a resident believed, accurately.

Index

Date	Policy area	Testimony
2026-08-10	renting	Ten years renting in Jerusalem
2026-08-10	housing-supply	A poor city building luxury towers
2026-08-10	construction-disruption	Everything dug up at once
2026-08-10	construction-disruption	A diversion map like a plate of spaghetti
2026-08-10	small-businesses	Shuttered shops, and nothing offered
2026-08-10	construction-noise	One hundred and ten decibels
2026-08-10	environmental-quality	The horns, and the air
2026-08-10	capital-priorities	A bridge I have never used
2026-08-10	public-transport	An hour to Talpiot, four hours home
2026-08-10	public-transport	Roving gangs of inspectors

All ten come from one household on one day. That is a real limitation and not a formality: eight areas resting on a single vantage point is exactly the case where testimony gets mistaken for prevalence, and the data annexes in stage 2 are what will have to carry them.

The concentration got worse rather than better on 10 August: six of the ten came from a second sitting offered as a scattergun list, and the rest from three more. Several are two or three sentences long. Each area’s tracker states how thin its own testimony is; none of them pretends otherwise.