How this works

Source discipline

Three different kinds of support

The project uses research, political and community frameworks, and its own design judgments. Those are not interchangeable.

A citation beside a card does not mean the whole card was tested in a study. It means a source is relevant to a narrower mechanism or context. The card-level note says where the evidence stops.

Direct Care

How one recommendation gets chosen

1. Start with current answers. State, desired outcome, blocker, available time, social preference, capacity, access needs, and hard constraints are the main input.
2. Remove impossible cards first. A card that breaks time, social, capacity, access, hard-constraint, or current consequence limits is not eligible.
3. Score relevance inside the eligible set. Desired outcome carries the most weight, then current state, then blocker. Mutual Aid role hints are a small optional refinement.
4. Keep ordinary recommendations Green or Yellow. Yellow cards stop for an explicit consequence check. Saying no routes toward a lower-consequence option.
5. Return one option. “Another,” “gentler,” “stranger,” and “social” search the remaining eligible pool without relaxing current limits.

The weights are relevance weights, not psychological scores. The tool does not estimate radicalism, resilience, courage, diagnosis, or readiness for escalation.

Mutual Aid import

What “Use my results” actually does

The import is opt-in. After you choose it, Self Care reads the completed browser-local Mutual Aid Role Quiz state and derives a small summary from eligible Role and Exposure sections. It does not import raw answers, Role scores, Capacity, or Strengths. It never reads the Mutual Aid backup credential key.

The derived summary stays in the current Self Care session. Current Self Care answers remain authoritative. Historical Exposure can add context, but it cannot raise the current risk ceiling or override a current limit.

What this does not establish: there is no empirical claim here that importing a role summary improves well-being. This is a source-verified personalization and privacy contract.

Optional current mode

Higher-Consequence Mode

Ordinary Direct Care remains Green and Yellow. Higher-Consequence Mode is a separate option you can open manually. A compatible Mutual Aid Exposure result can make it easier to find, but imported history never selects the mode, reviews consequences, or accepts a recommendation.

Orange mode can return reviewed Orange actions only after you choose that mode, mark which consequences are off the table today, and accept the consequences of that specific suggestion. Red identifies a context where arrest, detention, serious injury, serious retaliation, or comparable consequences may be on the table. Red-context self-care may protect vulnerable parts of life, reduce concentrated targeting, refuse isolation, redistribute risk, cover material care costs, support recovery, or contest the system producing the harm.

The companion library separates context risk from action risk. Most Red-context recommendations are Green, Yellow, or Orange actions. A reviewed Red action can appear only after you explicitly include Red options, it fits every current limit, and you answer a consequence check for that particular suggestion. Higher exposure is not treated as healthier, braver, or more politically committed.

Privacy: the mode, current consequence limits, answers about a particular suggestion, and higher-consequence history are session-only. They are not written to browser storage, a URL, or your Mutual Aid results. Capacity, Strengths, and raw Mutual Aid answers are not imported.

Evidence boundary: this mode is Diffrint design synthesis. Research and community sources may support ideas such as autonomy, support, recovery, accessibility, privacy, or sustainable participation. They do not prove that a particular higher-consequence action is therapy, wise for a particular person, or legally safe.

Automated surveillance infrastructure is politically contestable. This activity can name the system as a target for collective opposition, but it does not provide operational instructions for damaging equipment, defeating security, concealing evidence, or evading law enforcement.

Group Care

A structural check, not an organizational score

Group Care asks about 14 structural domains through seven paired areas, then asks how much process the group can afford and which limits are hard. It returns one bounded experiment.

1. Notice recurring structure. The prompts cover workload, backup, boundaries, decisions, informal power, conflict, urgency, access, exposure, care labor, belonging, burnout, succession, and joy.
2. Apply hard limits. The matcher cannot increase public or legal exposure, add care labor to people already carrying it, or assign structural repair to the overloaded person. Meeting and asynchronous-access constraints stay hard.
3. Prefer the smallest useful test. Capacity caps burden. A real emergency excludes experiments that only make sense later. If nothing fits, the tool tells the group to keep its limits instead of manufacturing a task.

The 28 experiments are reviewed hypotheses for organizing practice. They are not a validated organizational treatment package, and the internal ranking is not exposed as a healthy-group score.

Claim map

What the research is used to support

These are deliberately narrower than the product's politics or the wording of any individual card.

Card and experiment review

How source status is attached

Each Direct Care card carries one or more basis tags. Those tags determine which source notes appear beside a result without changing how the card is matched. Group Care sources are connected to the structural areas each experiment addresses.

• Evidence-supported context
• Theory / community framing
• Diffrint synthesis
• Mixed basis
• Source review still needed

All ten Direct Care basis categories and all 28 Group Care experiments have a source-review rule. If a new, unknown tag appears, the page marks it as needing review instead of inventing a citation.

Limits

What these sources do not prove

• No study here proves that refusal, disobedience, mutual aid, or anti-authoritarian politics are morally correct.
• No source validates all 60 Direct Care cards as treatments, or all 28 Group Care experiments as organizational interventions.
• Behavioral activation evidence is clinical evidence for depression treatment. It is not used to quietly rebrand this tool as therapy.
• Workplace burnout and meeting research can illuminate mechanisms, but volunteer groups, activist formations, friendships, and collectives are not simply workplaces.
• Social-connection evidence does not make solitude a problem. Current social preference remains a hard product input.
• Historical or imported exposure never counts as permission to suggest higher-risk action.

Political and design influences

Where the framing comes from

Dean Spade's mutual-aid work informs the project's attention to collective responsibility, group process, burnout, and hierarchy. Disability-justice and autistic self-advocacy sources inform collective access and the refusal to make disclosure a prerequisite for participation. Kropotkin is included as historical anarchist theory, not current biological evidence.

Diffrint's “Illegalist Therapy” essay is a design influence on the language of agency and consequences. This tool does not present that essay as an established clinical modality. Its ordinary recommendation scope is deliberately narrower. Camover is documented only as an influence on Mischief's game-like attention to surveillance and rules; destructive conduct is not carried into the card library.

Sources

Source catalog and caveats

Each entry says what it is being used for and where its relevance stops.