Built by citizens who believe
journalism matters
Warrant is an open-source project created by people who are concerned with protecting and advancing journalism and the free press. We are not a media company. We are not a newsroom. We are a community building infrastructure that makes integrity economically rational for independent journalists.
Every line of code is public. Every moderation decision is auditable. Every rule is documented. If you care about the future of journalism, you can inspect how this works — or help build it.
Why this exists
The incentives of digital media are broken. Warrant is an attempt to realign them.
For readers
There is no easy way to distinguish well-sourced reporting from unsupported claims. Corrections are buried. Repeat inaccuracies carry no consequences. Readers who want high-signal investigative work have nowhere reliable to find it.
For journalists
Independent reporters lack credible distribution channels and fair monetization. Ad-driven platforms reward engagement over accuracy, and journalists who take the time to verify are outcompeted by those who don't.
For the ecosystem
Digital media rewards speed over verification, outrage over truth, and low accountability for repeat inaccuracies. The economic incentives of journalism are misaligned with the public interest.
How we approach it
We shift incentives by coupling distribution, revenue, and reputation to demonstrated integrity — without becoming a publisher.
Platform, not publisher
We do not write, commission, or edit journalists' claims. We enforce standards through gating, labeling, downranking, and removal. This preserves Section 230 protection and keeps editorial independence with the journalist.
Truth is a process, not a badge
We never claim omniscience. We never say "verified true." Instead, we use process-based states: Supported, Disputed, Needs Source, Correction Issued. Process language is defensible and honest.
Everything is auditable
Every claim, edit, label, and moderation action is attributable, versioned, and reversible. No silent memory-holing. Dispute resolution relies on a transparent trail.
Incentives over intentions
Revenue, distribution privilege, and reputation are tied to demonstrated integrity — not just engagement. Good intentions don't survive contact with scale. Incentives do.
Enforcement is distributed
Risk is distributed through the reputation system, not concentrated in a moderation team making editorial calls. The system self-corrects through economic and reputational consequences.
What the labels mean
Every article on Warrant carries integrity labels that tell you the current state of its sourcing and accuracy. Here is what each one means.
Multiple credible sources align with the claims in this article. Source documentation is complete and the author has a strong track record. This is the highest integrity state — it means the reporting process was rigorous, not that the platform guarantees absolute truth.
Credible counter-evidence has been submitted against one or more claims in this article. A dispute does not mean the article is wrong — it means there is a substantive challenge that has not yet been resolved. The article remains available with the label attached so readers can evaluate both sides.
The article contains factual claims that lack adequate citations or supporting documentation. This may mean sources are missing, incomplete, or unverifiable. The author has been notified and distribution may be reduced until sourcing improves.
The author has amended one or more claims in this article. Corrections range from minor (typos, clarifications) to major (material factual errors). The full correction history is preserved and visible. Prompt, transparent corrections are treated as a sign of integrity, not failure.
This article has been flagged for review by the integrity team, typically because it contains high-risk content (such as allegations against named individuals with weak sourcing). It remains visible but with reduced distribution until the review is complete.
How reputation works
Every contributor has a reputation score from 0 to 100 (starting at 50). Reputation is earned through demonstrated integrity and lost through repeated failures. It directly affects how widely your work is distributed and how much you earn.
Reputation increases
- Publishing well-sourced content+2.0
- Your work is cited by others+1.5
- Complete source documentation+1.0
- Dispute overturned in your favor+2.0
- Tenure on the platform+0.5
Reputation decreases
- Dispute upheld against your content-5.0
- Major correction (factual error)-3.0
- Flag upheld against you-2.0
- Minor correction (clarification)-0.5
How reputation affects distribution
Higher reputation means broader initial distribution — your articles reach more readers faster. Lower reputation means slower distribution and more scrutiny before content surfaces. Very low reputation can result in throttled publishing or suspension. This creates a direct economic incentive: integrity pays, because it earns you more readers and more revenue.
How distribution works
Articles are ranked by a transparent algorithm — not engagement metrics, not ad revenue, not editorial preference. The scoring formula is:
Author reputation
Up to 40 points based on track record
Source quality
Up to 25 points for complete, primary sourcing
Recency
Up to 20 points, decaying over 72 hours
Integrity penalties
Disputed (-10), Under Review (-8), Needs Source (-5)
The full algorithm is open source. You can read it in src/services/distribution.ts.
How revenue works
Subscription revenue flows to the people who do the work. The platform takes a small margin to cover infrastructure; the rest goes directly to journalists.
Distributed proportionally based on readership weighted by an integrity multiplier (0.1x — 1.5x) derived from reputation, corrections, and disputes.
Covers hosting, infrastructure, payment processing, and integrity operations. This margin is configurable and published transparently.
We monitor the Gini coefficient of revenue distribution to ensure the system doesn't collapse into winner-take-all dynamics. Fair distribution is a design goal.
Verification is publication
This is one of the most important principles in our integrity model, and it is what sets Warrant apart from other platforms.
The problem it solves
Most platforms create a liability gap: the original poster bears all the risk, while those who validate, amplify, or “fact-check” a claim bear far less. This produces a pattern where low-reputation actors post risky claims and high-reputation actors “verify” them cheaply — externalizing epistemic risk upward with no consequences.
Our rule
Any account that publicly validates a claim assumes the same responsibility and consequences as if it had published the claim itself.
Posting false information results in a reputation penalty. Validating false information results in an identical penalty. This makes endorsement a scarce, high-stakes action. You only validate what you are willing to stake your reputation on.
Endorsement becomes scarce
You only validate what you're willing to stake your reputation and revenue on.
Fact-checking becomes adversarial
Copy-paste validation is irrational when it carries real consequences.
False claims propagate backward
When a claim is proven false, penalties hit every validator in the chain.
Identity where it matters
We balance accountability with safety. Not everyone needs to be identified — but everyone who earns money does.
Readers
Readers can be fully pseudonymous. You need an email address to subscribe and log in, but your identity is never exposed publicly. You can flag content, suggest corrections, and vote on feature requests without revealing who you are.
Contributors
Revenue-earning journalists must complete identity verification (government ID + liveness check) to confirm they are a real human. This does not require using your real name publicly — pseudonymous bylines are fully supported. Verification ensures that bans have real consequences and prevents sockpuppet manipulation.
Open source, always
Warrant is licensed under the GNU Affero General Public License v3 (AGPL-3.0). Every algorithm, every scoring formula, every moderation rule is published in the open. We believe that a platform asking you to trust its integrity systems must let you verify them yourself.
Want to understand how articles are ranked? Read src/services/distribution.ts. Want to see how reputation is calculated? Check src/services/integrity.ts. Want to audit the revenue split? It's in src/services/revenue.ts.
Support real journalism
Whether you read, write, or build — there is a place for you here.