How Moral Trade structures reasoning

Offer structure

Each offer separates what one side will do from what it asks another person to do. This keeps pledge swaps, donation offsets, and payment-mediated action offers legible.

Participation modes

A participant can join as an individual, a collective, or an institution. The app also distinguishes between a passive mode, where you record delegate rules and possible source connections, and a proactive mode, where you state explicit wishes, offers, asks, constraints, and verification preferences directly.

Wish profiling without AI

The current synthesis layer is deterministic. It summarizes user-entered fields, captured excerpts, manual source notes, and structured constraints into a private profile of hopes, intent, capabilities, and uncertainty. Clarification questions are generated from missing or underspecified fields rather than from an LLM interviewer.

Matching

Current match suggestions are rule-based. They use stated cause areas, compatibility with payment or pledges, shared terms, and consent-gated previews rather than AI inference.

The governance profile keeps current decisioning deterministic and schema-bound. Any future model used beyond explanation rendering must first publish model cards, datasheets, benchmark slices, intended-use limits, fairness audits, and change logs.

Fairness documentation must cover subgroup surfacing parity, false match rate, human overrule rate, privacy leakage incidents, appeal overturn rate across trade format, cause area pair, geography bucket, privacy stage, optional governed sensitive attribute.

Wish registry and staged disclosure

The wish registry indexes broad previews only. Searches surface just enough information to decide whether a counterparty seems worth exploring further. Exact wishes, constraints, identity details, and contact data remain behind consent and privacy-grant stages.

Moral public goods and distributed coordination

Some compromise destinations matter because many different moral views can value them at once. Global health, anti-poverty work, climate resilience, and other broadly shared public goods can make donation offsets more credible by giving opposed donors a named destination that is not merely a thin bilateral settlement. The platform therefore highlights moral-public-goods compromise destinations and treats coordination power as something that should be distributed, reviewable, and hard to weaponize through coercive threats.

Follow-through after a promising match

Background scans can open notifications, saved-search results, match reports, network invite drafts, brokerage bounties, and introduction plans. The goal is to take the first bounded steps toward a real conversation without auto-sending messages or pretending the system already has trustworthy autonomy.

Verification

Agreement events let participants record evidence, counterproposals, disputes, and payment updates. The goal is disciplined review rather than engagement-maximizing discourse.

Public validator evidence

The core proposal contract, review workflow, factor-code vocabulary, API route catalog, and provenance schema are published as validator-backed technical evidence. These contracts are the inspectable version of the methodology: they show which fields, privacy classes, fallbacks, and review states the product is allowed to rely on.

Centralized first, portable later

The present implementation is centralized for simplicity, but the data model includes export, import, and schema endpoints so wish profiles and source summaries can move if a more interoperable or decentralized registry becomes preferable later.

FAQ

Moral Trade records structured proposals; it does not hold money, provide legal or tax advice, or claim escrow. Public examples are worked examples unless a signed-in participant publishes a live offer, and evidence must be reviewed before anyone relies on a trade record as fulfilled.

Sources

The public product language draws on Toby Ord's “Moral Trade” and Forethought's essays on convergence, compromise, and moral public goods. The site summarizes these ideas without claiming legal enforceability, custody, or evaluator status.