tasks.md

The honest state of the project, kept in public. Verification debt is listed here rather than quietly carried, because a platform that argues from sourced evidence does not get to hide its unsourced parts.

Where it stands

The method is built and the content has just started. The pipeline, the plugin, the powers doctrine and the context scaffold exist. On 10 August 2026 the first testimony landed — a single dictated account, given across five sittings, which opened eight areas at once. All eight sit at stage 1, and under our own rules nothing in them may be proposed until the jurisdiction gate has been run.

Eight areas and zero jurisdiction maps is the wrong ratio, and it is the main thing to fix. The account was offered as a scattergun list and it was taken at face value: an area opens when there is testimony behind it, so eight opened. The cost is that the queue in front of /map-powers is now more than twice as long as the work done to date, and some of those areas rest on two or three sentences.

The same day also produced the project's only positive material — what is good about the city — which matters more than its length suggests. It is the only thing on file naming what a proposal must not break, and the only statement of a diversified economic alternative to the bets the platform criticises.

Layer	State
Pipeline and evidence rules	✅ written
municipal-policy-assistant plugin — 7 commands, 4 skills	✅ written; /capture-testimony now run once, /capture-observation added 2026-08-10, the other five still untested
Powers doctrine	✅ written
Context — the election	🟡 2024 result recorded, from a secondary source
Context — City Hall	🟡 seat arithmetic only; coalition and record not researched
Context — powers primer	🟡 hypotheses and research plan only; nothing verified
Context — the argument the city is having	🟡 framing claim recorded, entirely unverified
Context — what is good about the city	🟡 the asset base recorded, entirely unverified
Policy areas	🟡 eight open, all at stage 1
Observations	🟡 six recorded, none promoted; a new document class as of 2026-08-10
The emerging spine	🟡 five mechanisms named, all pre-gate hypotheses
Program for the city	⬜ empty, by design

Two of those layers did not exist that morning. Observations is a second door into the project, for problems that have no first-person account behind them — the previous rule was that a problem nobody dictated had nowhere to go, and it was losing material. The price of the door is a mandatory origin line stating how each one got here; an observation is not evidence and cannot open an area. The spine reads the eight areas as one body of material rather than eight, which is the only honest way to move toward a manifesto while the jurisdiction gate is still unrun.

Policy areas

Area	Testimony	Jurisdiction	Research	Paper	Manifesto
Renting	✅	⬜	⬜	⬜	⬜
Housing supply	✅	⬜	⬜	⬜	⬜
Construction disruption	✅	⬜	⬜	⬜	⬜
Construction noise	✅	⬜	⬜	⬜	⬜
Small businesses	✅	⬜	⬜	⬜	⬜
Environmental quality	✅	⬜	⬜	⬜	⬜
Capital priorities	✅	⬜	⬜	⬜	⬜
Public transport	✅	⬜	⬜	⬜	⬜

✅ complete · 🟡 in progress · ⬜ not started

All eight rest on one household's account given on one day. That concentration is itself a finding to work against, not a base to build on — see whose account this is. Testimony depth also varies sharply between them: renting runs to several hundred words of first-hand detail, environmental quality is two sentences. Each tracker says which it is.

Verification debt

Claims currently resting on secondary sources or on nothing, in priority order. Each blocks something.

#	Claim	Current basis	Needed	Blocks
1	The arnona rule — that rates cannot be raised without joint Interior and Finance approval	Working assumption, uncited	The current statutory provision, the approval criteria, and Jerusalem's record of seeking deviations	Every costed proposal, and now the renting area directly
2	The bylaw-making process and the Interior Minister's confirmation requirement	Municipalities Ordinance, chapter identified but not read	The operative sections, the confirmation criteria, typical elapsed time, Jerusalem's recent record	Every class-B proposal
3	The 2024 election results table	English Wikipedia, accessed 2026-08-10	Ministry of the Interior published results	Council arithmetic in stage 4
4	The next election date	Inferred from the five-year cycle	The Local Authorities (Elections) Law, and whether the 2024 postponements moved the cycle	Nothing structural — the project is not timed to it
5	East Jerusalem franchise and turnout	Commonly repeated figures, not verified	A primary or reputable statistical source	Any city-wide prevalence claim
6	Community administrations (מנהלים קהילתיים) — formal status and powers	Named only	Their statutory or contractual basis, budget and relationship to the municipality	Any proposal routed through them
7	The balancing grant and Jerusalem-specific funding	Named only	Amounts, formula, conditionality	Fiscal envelope in stage 4
8	Who is statutorily liable for arnona — the mahzik (מחזיק) rule — and whether the municipality has any discretion over whose name a bill sits in	Testimony only	The operative sections of the Municipalities Ordinance, and Jerusalem's registration practice	The whole renting area
9	Jerusalem's arnona is the highest in the country	Testimony only	A like-for-like comparison on a stated measure, across a stated comparator set, for a stated year	The renting problem statement
10	Jerusalem is consistently the poorest city in Israel	Testimony only	CBS socio-economic index and poverty measures; the ranking differs by measure and the measure has to be named	The housing-supply problem statement
11	Which bodies actually deliver the light rail works, and the municipality's formal role	Nothing	The delivering entities, their ownership, and the permit relationship with the city	The construction-disruption jurisdiction map — and probably its headline finding
12	Construction noise measured at 110 dB	Testimony only, with no method recorded	Instrument, distance, weighting and duration — a decibel figure without them cannot be cited	The whole construction-noise problem statement
13	Whether any instrument caps how long a site may operate noisily, as distinct from daily hours	Testimony, reported second-hand and flagged there as unverified	The noise regulations, any Jerusalem bylaw, and the noise conditions on a building permit	construction-noise, and its likeliest class-A finding
14	"Countless" city-centre businesses have closed because of the works	Testimony only	A count on a stated measure, a baseline, and something separating works-caused closure from ordinary retail attrition	The small-businesses problem statement
15	Whether arnona relief is available to a business harmed by public works, and whether any Israeli municipality has granted it	Nothing	The relief instrument and its approval route, plus the precedent scan that settles feasibility fastest	Every proposal in small-businesses
16	Jerusalem's air quality is "significantly lower than it should be for a Western capital"	Testimony only	Monitoring-network measurements by station and year, a named comparator set, and the regional dust baseline controlled for	The environmental-quality problem statement
17	Who enforces unnecessary horn use	Nothing	The Traffic Regulations provision and the enforcing body — expected to be Israel Police, which would make the headline grievance class D	Whether environmental-quality has a municipal proposal in it at all
18	The funding and delivery of Jerusalem's tourism and prestige capital projects, including the Old City walkway	Nothing	The development budget by programme, the split between municipal and national/development-authority money, and whether any of it was ever fungible	The capital-priorities jurisdiction map, and the viability of its whole argument
19	That the loudest municipal argument is not the one with the largest effect on residents	Framing claim from the account, unsourced	Resident priority survey data, council and committee agenda analysis, and Hebrew-language local press coverage by subject	The argument the city is having, and a good deal of this platform's emphasis
20	Bus frequency on the affected corridors has fallen far below its previous level	Testimony only	Published timetable data before and during the works, and whether the cause is the works, a driver shortage or a specification change	The public-transport problem statement
21	Who owns and maintains Jerusalem's bus stops, shelters and shade	Nothing	Ownership, any advertising concession, and shelter coverage by neighbourhood	The likeliest class-A finding in public-transport
22	Whether small commercial stock in Talpiot is being replaced by residential high-rise as a planning outcome the city chose	Testimony only	Permits granted, zoning changes, and whether the municipality had discretion	small-businesses, and a second mechanism distinct from works disruption
23	Jerusalem has "good fibre penetration"	Account only	Penetration by neighbourhood against national figures, and whether the city held any lever over deployment	The remote-work argument in what is good about the city
24	Fast, reliable rail makes the Tel Aviv/Jerusalem opposition outdated	Account only	Journey times, frequency and the reliability record on the line	The same argument, and the economic thread generally
25	The city has coupled its plans to diplomatic events and embassy openings	Account only	What has actually been committed on that basis, as against public messaging about it	capital-priorities, where it is the mirror of the tourism objection
26	Who employs and directs fare inspectors on Jerusalem buses and the light rail, and under what statutory power	Nothing	The operator/concession structure, the inspection power, and whether any conduct standard or complaint route exists	Whether the enforcement half of public-transport has a municipal hook at all
27	The documented cases of inspector conduct described as a national media phenomenon	Recollection only	Sourced, dated reporting, characterised accurately rather than from memory	The public-transport problem statement — and the claim most likely to be overstated if left unchecked
28	Whether what changed in the bus market was deregulation or competitive tendering	Testimony, hedged by the speaker	The reform's actual shape, because it determines who can change the outcome	public-transport, and a likely wrong-attribution finding
29	The fare and the fine, in shekels	The account reports the single fare at about ₪5.90 per ride; the penalty is unknown, and the dollar figures were flagged by the speaker as approximate	The published tariff and penalty, plus how effective per-journey cost differs between a monthly pass and single fares	Any use of the disproportion argument; a resident-reported figure is not a citation

Next actions

Blocking everything:

1. Run /map-powers on the eight open areas. Nothing may be researched, drafted or proposed in any of them until the jurisdiction gate has been cleared. Suggested order:

   1. construction-disruption — its attributions are the most likely to be wrong, which makes it the sharpest first test of the skill.
   2. construction-noise — shares the permit question with the one above, so running them together does the statutory reading once.
   3. renting — the clearest candidate for an unused class-A power.
   4. small-businesses — shares the arnona constraint with renting; same argument.
   5. housing-supply — depends most on the standing powers primer being done.
   6. capital-priorities — the jurisdiction question and the evidence question are the same question here, so it is the most expensive of the seven.
   7. public-transport — expected C/D on both the service and the fare-enforcement regime, but the split it forces between a nationally franchised service and the municipal street furniture around it is reusable across the transit material.
   8. environmental-quality — last on purpose. A D on the headline grievance is a live possibility, and finding that out cheaply is what the gate is for.

Runnable now, independent of the threads:

2. Build the verified powers primer — items 1, 2, 6 and 7 above. This is the largest piece of standing research in the project and every policy area will draw on it, so doing it once properly beats doing it six times partially.

   Add one item to it that the spine surfaced: what the municipality must publish, may publish, and can compel a contractor or delivering body to tell it. Construction disruption, construction noise and public transport all turn on that same question, as do two observations. Researched once as standing material it is cheap; researched three times inside three jurisdiction maps it is not.

3. Research the City Hall baseline — the coalition as constituted, portfolio and committee allocation, any changes since February 2024, and the administration's record. Coalition agreements first; they are the highest-value document in that section.

4. Re-verify the 2024 results against the Ministry of the Interior, and settle the next election date.

Deferred:

5. Cover image and Open Graph image for the site — the landing page and social cards currently have neither.

6. First run of /map-powers will be the plugin's real test. Expect the command and the skill to need revision afterwards; that revision is part of the experiment, not a defect.

7. Broaden the testimony base. Eight areas resting on one household is the project's biggest methodological weakness right now, and no amount of stage-2 research fully fixes it. It got four times worse on 10 August, not better.

   The observations now say what to go and ask for, which is the one useful thing they do. Each names the testimony that would promote it: someone who has tried to run a business here, someone who used a formal objection route, someone who could not complete a journey on foot, someone who missed a notice because of the language it was in. Those are four specific asks rather than a general wish for more accounts.

8. Go and read the Hebrew. The account's own closing point is that dissatisfaction here is contained in offline, Hebrew-language spaces — the argument the city is having. An English-language repository kept by an anglophone immigrant will systematically miss exactly the evidence that would confirm or refute most of what is above. Neighbourhood groups, Hebrew local media, community administration minutes and the municipality's own Hebrew publications are where stage 2 has to go.

9. Recover the end of the 10 August account. The transcript stops mid-sentence on the containment point; whatever followed was not captured.

Candidate threads

Not commitments, and not chosen — these are areas where a Jerusalem resident plausibly has a first-person account, listed so the blank page is less blank. The user picks; an area only opens when there is testimony behind it.

Cleanliness and refuse collection · street and pavement condition away from the works · parking and enforcement · the light rail and bus network as experienced rather than as planned · building permits and the local planning committee · parks, playgrounds and shade · the community administrations · what the municipality publishes and what it does not · services in neighbourhoods that get less of them · animal control · signage and wayfinding · what happens when you call the municipal hotline · Shabbat and public space — named in the account as the argument the city is actually having, and deliberately not opened as a policy area on the strength of that; it opens if there is testimony about it rather than about its salience · the Jerusalem economy — raised twice now and still not opened, and now written up as an observation rather than left as a line in this paragraph. The opening account gives the constraint side: fewer jobs, lower pay, a cost of living close to the centre's. What is good about the city gives the opportunity side: remote work, fibre and rail connectivity, and the argument that the Tel Aviv/Jerusalem opposition is outdated. Together those are probably enough testimony to open an area. It stays shut for now because eight areas already sit behind an unrun jurisdiction gate, and because it is still not clear the municipality holds a lever over an employment market — which is exactly the question /map-powers exists to answer. Open it once the gate has been run on something.

The ones worth doing are the ones where the resident's assumption about who is responsible turns out to be wrong, or where a class-A power has been sitting unused. Neither is knowable in advance of the testimony.