# 06 — The Investigation Market

> *Parties who care about contested claims pay for investigations. Validators compete to execute them. The protocol routes real money to where real disputes are hottest — and investigative journalism finds a new business model.*

## The mechanism

When a claim on the Veritas log is contested — meaning at least one attestation is `DISPUTED`, or a party requests investigation before attestation exists — the investigation market lets anyone *commission* a deeper verification.

**Flow:**

1. **Commissioning.** A party (individual, organisation, NGO, media outlet, state actor, opposing party in a dispute) posts an investigation commission on the protocol. The commission specifies:
   - Target claim(s) by hash.
   - Consensus domain(s) under which investigation is requested.
   - Minimum validator quorum.
   - Escrow amount (in VRT, USDT, or fiat).
   - Deadline.
   - Optional: required validator criteria (credentials, jurisdictions, independence from commissioner).

2. **Escrow.** Commission funds are held in the treasury's escrow contract until conditions are met.

3. **Validator bidding or auto-assignment.** Validators who match criteria and have available capacity bid or are auto-assigned. The protocol's reputation math favours high-reputation, domain-credentialed validators with independence from commissioners.

4. **Execution.** Assigned validators investigate the claim — read primary sources, reproduce evidence, document chain of reasoning, sign a verdict.

5. **Publication.** Investigation outputs — evidence, reasoning, verdict — are published on the protocol log with the commissioner's name (or explicit anonymity marker if permitted).

6. **Payment release.** On completion + minimum quorum, the escrow releases to validators per the protocol's fee-split policy. Foundation takes a small fee (5–10%) for treasury operations.

7. **Dispute and appeal.** If the commissioner or any party disputes the investigation output, a published appeals process handles it (the dispute panel; see `09-refusals-and-panel.md`).

## Why this mechanism has real promise

### Real money routed to real disputes

Existing fact-check infrastructure operates on roughly uniform editorial attention: every claim gets roughly the same amount of investigator time, regardless of how contested it is or how much is at stake. Veritas's investigation market inverts this: claims that matter to parties with resources get more investigation, because those parties pay for it.

This is not a defect. Contested claims — especially ones where multiple parties dispute outcomes — are exactly the claims where more investigation produces more public value. A claim nobody disputes does not need an investigation; a claim that is fought over does.

### Investigative journalism finds a business model

The core business model of investigative journalism (advertising + subscription) collapsed with the internet. Attempts to replace it (patronage, foundations, paywalls, Substack) have produced survivors but not a robust sector.

The investigation market offers something different: **investigators get paid per investigation, directly from parties who care about the outcome.** A newsroom that can execute a contested-claim investigation well earns revenue per case. A retired investigative journalist can operate as a solo validator. A university investigative-journalism programme can train students while executing real commissions.

This is not sufficient to save the entire sector, but it is a genuinely new revenue flow that did not exist before.

### Mutually-hostile frames produce investigations for each other

If an Israeli-aligned party and a Palestinian-aligned party both believe the other side's claims are false, they can both pay for investigations. The protocol routes the money to validators who execute. The result: **both sides of a contested historical event produce high-quality, documented, signed investigations** — and the plural-verdict substrate records both.

This is the design thesis in `01-mutually-hostile-validators.md` made operational: adversaries fund the mechanism that produces the evidence. The protocol does not pick a side; it lets each side pay for its verdict to be taken seriously, and surfaces disagreement honestly.

### Price discovery on epistemic attention

For the first time, there is an explicit market price for *"I want an expert to verify this claim to this rigour."* Existing fact-check labour is mostly unpriced — donated, grant-funded, or a side-product of journalism. A pricing mechanism forces explicit acknowledgement of what verification actually costs and lets commissioners decide how much verification they want.

## Pricing model

Starting price schedule (Phase II pilot):

| Commission type | Minimum validators | Starting price (USDT-equivalent) |
|---|---|---|
| Quick verification (1 claim, single domain) | 2 | $300 |
| Standard investigation (1 claim, multi-domain) | 3 | $1,000 |
| Deep investigation (complex claim, primary-source work) | 5 | $3,000 |
| Extended investigation (novel primary-source acquisition, multi-week) | 5+ | $10,000+ |
| Adversarial cross-investigation (party A commissions re-investigation of party B's claim) | 5 | $2,500 |

These are starting figures. Real market price will emerge from supply / demand in the first 6–12 months of Phase II operation.

Pricing is published and transparent. Commissioners see the fee schedule before committing. Validators see the commission before bidding or accepting.

## Critical analysis — failure modes

**1 — Bad-faith muddying.** A well-funded actor repeatedly commissions investigations on claims they want to de-legitimise, flooding the log with faint-praise / uncertainty verdicts that collectively muddy a well-established fact. Mitigations:
- Commissioner patterns are public; pattern-detection flags repeat-commissioners of specific claim clusters.
- Price of additional investigation rises for claims that are already well-verified (diminishing-returns fee curve).
- Investigation output is public + signed; commissioners cannot buy *private* investigation to produce suppressed findings.

**2 — Sybil commissioners.** Many wallets from one beneficial owner commission investigations to simulate broad public interest. Mitigations:
- Chain-analysis flags suspicious commissioner patterns.
- Reputation score for commissioners; new / anonymous commissioners pay higher fees until history is built.
- Optional verified-commissioner badge (DID-linked to real-world identity) reduces fees; anonymous commissioning costs more.

**3 — Rich parties drown out poor parties.** In an asymmetric dispute, the well-funded side commissions 10 investigations, the under-funded side commissions one. Verdicts are dominated by the well-funded perspective. Mitigations:
- NGO / foundation funds specifically target under-resourced claims — "public-interest investigation fund" with independent board.
- Pro-bono validator commitments — validators commit a percentage of capacity to public-interest investigations at no fee.
- Commissioner identity is public; adversarial-party-one-sided commissioning is visible and context-providing.
- Protocol does not claim the investigation market produces "balanced" verdicts. It claims verdicts are clearly attributed to their commissioners + executing validators, and let consumers judge.

**4 — Investigator-commissioner collusion.** Commissioner pays validator privately, validator returns the verdict the commissioner wants, escrow pays as well. Mitigations:
- Validator payment comes from escrow, not from commissioner — commissioner cannot selectively reward.
- Validators cannot accept commissions from parties they have financial relationships with (published disclosure rules).
- Reputation-weighted quorum — a single compromised validator cannot move the verdict.
- Retrospective audit — patterns of validator-commissioner affinity are detectable statistically.

**5 — Investigation output is weaponised.** A signed investigation verdict becomes ammunition in public disputes; losing parties discredit the protocol rather than their narrative. Mitigations:
- Investigation outputs include explicit uncertainty bounds and contested-frame notation.
- Public communications teach audiences how to read investigation outputs — verdict is under a specific domain; other domains' verdicts may differ.
- Protocol is transparent about the limits of any single investigation.

**6 — Regulatory exposure for validators.** An investigation that concludes "claim X is false" and publishes evidence can expose validators to defamation claims from X's originator. Mitigations detailed in `72-legal-regulatory-landscape.md`:
- Validators frame outputs as opinion + procedural under editorial standard D.
- Foundation indemnifies validators acting within published standards.
- Insurance (D&O + E&O) carried at foundation level.
- DSA Art 22 Trusted Flagger pathway for EU operation.

**7 — Claim targeting for harassment.** Adversaries commission investigations of minor claims to harass specific content producers. Mitigations:
- Producer can opt their content out of investigation-market inclusion (protocol respects; but attestation-market may still produce public attestations without commissioner payment).
- Commissioning a claim at high volume (e.g., > N/year targeting same producer) triggers review.
- Dispute panel hears harassment complaints.

## Related work (to be enriched by quant agent)

- **Prediction markets as adjudication** — Polymarket's UMA-based dispute resolution; Augur; Kalshi.
- **Kleros court system** — case-fee-based dispute resolution; jurors earn fees.
- **UMA Optimistic Oracle** — dispute-triggered escalation pattern.
- **Expert-witness markets in law** — parties pay expert witnesses to support their case; court weighs all.
- **Academic peer review** — closest non-market analog; incentive structure is different but the "parties commission evaluation by independent experts" logic is comparable.
- **Investigative-journalism funding innovations** — ProPublica, Bellingcat, Distributed Denial of Secrets, The Intercept. Each has distinct revenue flow; none at investigation-market scale.

## Open questions

- How does the market handle claims where no side has money to commission? (Public-interest fund answer is partial but not complete.)
- Is there a minimum viable volume below which the market is too thin to price efficiently? Phase II data will tell us.
- Should commissioners see validator bids before accepting, or be auto-assigned to preserve independence? Probably auto-assignment with commissioner ability to reject on disclosed conflict.
- How does the protocol handle a commissioner who refuses to pay after validator execution? Escrow protects validators; commissioner forfeits stake.

## What we'd build

- **`veritas-investigation-market-contract`** — on-chain escrow, commissioning, auto-assignment, fee-split, payment release.
- **`veritas-validator-bidding`** — auction / assignment mechanics for validators.
- **`veritas-commissioner-reputation`** — reputation tracking for commissioners; pattern detection; suspicious-pattern flagging.
- **`veritas-investigation-workflow`** — operational tooling for validators executing commissions: evidence capture, reasoning trail, verdict signing, publication.
- **`veritas-public-interest-fund`** — foundation-operated pool that commissions investigations on behalf of under-resourced claims.
- **`veritas-investigation-audit`** — retrospective analysis of market outputs for collusion, asymmetry, and bad-faith patterns.
- **Published fee schedule + pricing policy** — transparent, updated quarterly based on market data.