> ## Documentation Index
> Fetch the complete documentation index at: https://docs.fundwork.app/llms.txt
> Use this file to discover all available pages before exploring further.

# Dispute Resolution Architecture

> How Fundwork thinks about dispute resolution for paid work, the categories of verification, and the planned dispute flow for escrowed opportunities.

Dispute resolution is one of the hardest parts of any paid-work marketplace. Software cannot objectively decide whether a piece of subjective work is good, and pretending otherwise leads to systems that mistreat either workers or clients. This page describes how Fundwork thinks about verification, what a planned dispute flow looks like, and what is explicitly not yet decided.

<Note>
  Fundwork is in **alpha**. The production dispute mechanism is not finalized. This page describes the architectural direction. For current dispute handling in alpha, see [Trust & Safety → Disputes](/trust-and-safety/disputes).
</Note>

## Different Work Needs Different Verification

The most important observation is that different kinds of work require different verification mechanisms. Treating them all the same produces bad outcomes. Fundwork categorizes verification along three broad axes.

### Objective verification

Some work can be verified against a deterministic criterion. Examples include:

* An onchain event, for example a contract deployment at a specific address.
* Passing a defined set of automated tests.
* A specific API output.
* A specific Git commit or pull request merged.
* A deterministic artifact matching a known hash.

Opportunities in this category may eventually support automated verification: the verification code itself decides whether the deliverable qualifies.

### Client approval

Most creative and knowledge work is inherently subjective. Examples include:

* Design
* Writing
* Video
* Strategy
* Research quality

For these categories the client needs to approve the deliverable manually. No mechanical check can substitute for judgement about whether the output is fit for purpose.

### Disputed work

When the client and the worker disagree about whether the deliverable meets the agreement, the opportunity enters a dispute path. This is where the architecture matters most.

## Planned Dispute Flow

For escrowed opportunities, the planned dispute flow is:

<Steps>
  <Step title="Submission">
    The worker submits the deliverable. The opportunity moves to a **Submitted** state.
  </Step>

  <Step title="Disagreement">
    The client rejects or challenges the submission, or the worker challenges a rejection they believe is unfair.
  </Step>

  <Step title="Dispute opened">
    The opportunity moves to a **Disputed** state. Escrow release is paused. See [Escrow Design](/technical/escrow-design).
  </Step>

  <Step title="Evidence frozen">
    Submitted files, messages, timestamps, and related records are captured so they cannot be altered while the dispute is under review.
  </Step>

  <Step title="Review">
    The dispute is reviewed according to the dispute policy in effect for that opportunity.
  </Step>

  <Step title="Decision">
    A decision is recorded: release, refund, or partial resolution.
  </Step>

  <Step title="Escrow acts on the decision">
    Escrow releases or refunds according to the recorded outcome. State transitions to **Released** or **Refunded**.
  </Step>
</Steps>

## What Evidence Matters

A dispute is only as fair as the evidence available to review. The intended evidence set includes:

* The original opportunity description as posted.
* Any milestone requirements attached to the opportunity.
* The submission files themselves.
* Timestamps for submission, approval, and revision events.
* Messages exchanged between the parties on Fundwork.
* Any revision requests recorded during the engagement.
* Onchain records for funding, milestone events, and prior partial payments.

Keeping communication and submissions on-platform is what makes this evidence set complete. Off-platform work is largely invisible to any dispute process.

## What Fundwork Is Not Promising

Dispute resolution is where marketplaces most often overstate their capabilities. To be explicit:

* Fundwork does not currently offer decentralized arbitration.
* Fundwork does not currently offer automated arbitration.
* Fundwork does not currently commit to Fundwork-controlled arbitration.
* Fundwork does not claim smart contracts can decide whether subjective work is good.

Fundwork will define and publish the dispute rules that apply to escrowed work before enabling production escrow. Those rules will describe who makes the decision, what evidence is considered, what timelines apply, and how the outcome is enforced.

## Directions Under Evaluation

Future versions of the platform may explore any of the following, depending on which categories of work they best fit:

* **Human reviewers** who evaluate submissions against the agreed criteria.
* **Designated third-party arbitrators** selected by both parties in advance.
* **Multi-party review** where several independent reviewers each vote.
* **Deterministic verification** for objectively verifiable opportunity types.
* **Agent-assisted verification** where software agents surface evidence for human decision-makers.

These are exploratory directions, not commitments. Any one of them raises real design questions around cost, latency, incentive alignment, and correctness that will be worked through before it is offered as a production dispute path.

## Status

<CardGroup cols={2}>
  <Card title="Currently available" icon="circle-check">
    Direct support for dispute-related issues in alpha via Fundwork support. See [Trust & Safety → Disputes](/trust-and-safety/disputes).
  </Card>

  <Card title="Planned before public launch" icon="clock">
    Published dispute rules and evidence policy for escrowed opportunities. Integration between the dispute state and the escrow state machine.
  </Card>

  <Card title="Under evaluation" icon="flask">
    Human reviewers, third-party arbitrators, multi-party review, and deterministic verification for eligible categories.
  </Card>

  <Card title="Future" icon="telescope">
    Agent-assisted evidence gathering and automated verification for objectively verifiable work.
  </Card>
</CardGroup>
