Use this skill when an OPEN/DIRECTION question must move toward a defensible public position, especially because:

• journalists or voters will reasonably demand a concrete answer;
• current law or evidence has changed;
• a valmanifest, EU-program or shadow budget needs a parameter decision;
• the same unresolved question keeps recurring;
• the party risks using more analysis as avoidance rather than epistemic humility.

Core rule:

OPEN is a queue, not a home.

The goal is not to force false certainty. The goal is to know what remains uncertain, what is a value choice, when a choice must be made and what evidence would justify changing it later.

Before closing a question:

1. memory/01-core.md;
2. relevant domain memory/03–10;
3. memory/14-citizen-coverage.md when distribution/rights/life situations matter;
4. memory/15-deprecated-and-open-questions.md;
5. program/12-besluts-och-granskningsram.md;
6. program/14-hard-questions-and-decision-queue.md when the question is a recurring public/press attack;
7. latest relevant audit/source note.

For economic questions also read program/02-ekonomisk-strategi.md and any current tax/budget architecture.

Do not start from the party's preferred framing. Start from the strongest public challenge.

Examples:

• Should Sweden adopt the euro — yes or no?
• Are you restoring the property tax?
• Should police be allowed to identify people with real-time facial recognition?
• Why are teeth not treated like the rest of the body?

Then write the best hostile follow-up. If the answer only survives a friendly first question, it is not closed.

Separate:

• normative/value choice;
• empirical mechanism;
• legal constraint/current law;
• professional judgment;
• fiscal/resource constraint;
• implementation/transition;
• distribution and rights;
• reversible vs irreversible components.

Do not let research pretend to answer a value question, and do not let a value slogan pretend to settle an empirical question.

Use systemic detail understanding:

whole → system → dependency → mechanism → detail → power → intervention → feedback

Ask:

• what outcome are we actually trying to change?
• which institutions/markets/people produce it?
• what depends on what?
• where is the actual bottleneck or power point?
• who can object, appeal or exit?
• what second-order effect appears elsewhere?

For time-sensitive questions, use current primary sources first:

1. Swedish law, government inquiries, Riksdag, authorities, SCB, Riksbank, KI, official EU/NATO/OECD sources where relevant;
2. systematic reviews, established research institutions and peer-reviewed work;
3. serious professional objections and minority views;
4. contemporary public debate to identify the questions citizens/journalists will actually ask.

Do not treat the debate as evidence. Use it to reveal hidden value conflicts, implementation questions and missing alternatives.

Store reusable provenance in sources/ with date and links/citations. A source file never creates policy by itself.

Include status quo unless obviously impossible.

For each alternative show:

• strongest argument for it;
• strongest argument against it;
• expected mechanism;
• rights/intrusion;
• fiscal cost/revenue where material;
• distribution;
• implementation risk;
• reversibility;
• what evidence is genuinely uncertain.

Avoid fake balance: if one alternative violates a core right or has evidence of serious harm, say so.

At minimum check:

• low/median/high income;
• children and youth;
• elderly;
• disabled/chronic illness;
• rural/urban;
• renters/owners where relevant;
• workers/small firms/large actors;
• people with weak digital or administrative capacity;
• minorities and those likely to bear enforcement;
• future generations for long-lived commitments.

Ask separately who gains money, agency, risk, voice, exit and burden of proof.

COMMITMENT

Use when the question follows directly from rights/core values or is a principle the party will defend even at political cost.

DIRECTION

Use when the target and broad mechanism are chosen but parameters/implementation remain.

PILOT

Use when the mechanism is plausible, reversible and evidence can reasonably be produced through controlled trial.

OPEN

Keep only when a real unresolved choice remains. Every strategically important OPEN question must have:

• decision deadline/event;
• missing evidence or value choice;
• named next research product;
• a truthful interim press answer.

More research is needed without these fields is not a valid terminal state.

10 seconds

One direct sentence. Answer the actual question before adding nuance.

60 seconds

State choice/rule + why + key safeguard/tradeoff.

Full answer

System map, evidence, alternatives, cost, distribution, implementation, uncertainty and change conditions.

The short answer may omit detail but must never contradict the full answer.

A closed position must say what could reopen it.

Examples:

• an independent evaluation shows no effect;
• error rates exceed a predefined threshold;
• the fiscal base changes materially;
• a new legal constraint applies;
• distribution becomes much worse than modeled;
• a new technology removes the original tradeoff.

This is not weakness. It is the party's public anti-dogma contract.

When a position changes:

1. update relevant canonical memory/ file;
2. update relevant program//Q&A/decision queue;
3. update memory/15 if OPEN/deprecated status changed;
4. add source provenance if useful;
5. add an audit for a major/controversial closure;
6. update status/index only when materially needed.

Do not keep contradictory versions alive in multiple canonical files. Historical arguments belong in audits/sources, not parallel policy.

Before calling a question closed, verify:

• Would a hostile interviewer say we answered the question?
• Is the 10-second answer true?
• Is the value judgment visible rather than hidden as science?
• Did we model the whole system rather than one headline metric?
• Did we analyze who loses?
• Is funding/administrative capacity real?
• Does the position preserve voice, exit and meaningful appeal where relevant?
• Is the change condition clear?
• Would the same method yield this answer if the politically popular side were reversed?

If not, keep working.