Musing - Fair Witnessing

--- title: Musing - Fair Witnessing robots: noindex, nofollow tags: musings --- # Fair Witnessing === ## Draft #2 ### Fair Witnessing in a Decentralized World by Christopher Allen > ABSTRACT: "Fair Witnessing" is a new approach for asserting and interpreting digital claims in a way that mirrors real-world human trust: through personal observation, contextual disclosure, and progressive validation. It can be implemented with the decentralized architecture of Gordian Envelopes to allow individuals to make verifiable statements while balancing privacy, accountability, and interpretability. At its core, fair witnessing is not about declaring truth, it's about showing your work. In the early days of decentralized identity, we referred to what we were working on as "Verifiable Claims." The idea was simple: let people make cryptographically signed statements and allow others to verify them. But something unexpected happened. People assumed these claims would settle arguments or stop disinformation. They saw the term “verifiable” and equated it with “truth.” The reality was more modest: we could verify the source of a claim but not its accuracy. We could assert that a claim came from a specific person or organization (or even camera or other object) but not whether that claim was unbiased, well-observed, or contextually complete. This misunderstanding revealed a deeper problem: how do we represent what someone actually saw and how they saw it, in a way that honors the complexity of human trust? ### A Heinleinian Inspiration <img src="https://www.blockchaincommons.com/images/posts/fw-stranger.jpg" style="float: right" width=350"> Iin _Stranger in a Strange Land_, Robert Heinlein described a special profession: the Fair Witness. A Fair Witness would be trained to observe carefully, report precisely, make no assumptions, and avoid bias. If asked what color a house was, a Fair Witness would respond, “It appears to be white on this side.” It is this spirit we want to capture to fulfill the promise of the original verifiable claims. A Fair Witness in our digital era is someone who not only asserts a claim but also shares the conditions under which it was made, including context, methodology, limitations, and bias: - What were the physical conditions of the observation? - Was the observer physically present? - Did they act independently? - What interests or leanings might have shaped their perception? - How did they minimize those biases? These are not just nice-to-haves. They are necessary components of evaluating a claim's credibility. ### Beyond Binary Trust Fair witnessing challenges binary notions of trust. Traditional systems ask a "yes" or "no" question: do you trust this certificate? This issuer? But trust is rarely binary like this in the real world. It is layered, contextual, and progressive. The claim made by a pseudonymous environmental scientist might start out with low trust but could grow in credibility as: - They reveal their professional history. - Others endorse their work. - They disclose how they mitigated their potential biases. Trust builds over time, not in a single transaction. That's [progressive trust](https://www.blockchaincommons.com/musings/musings-progressive-trust-lifecycle/). ### Trust as a Nested Statement To marry a fair witness claim to the notion of progressive trust requires the nesting of information. As shown in the example of the environmental scientist, the witnessing of an observation gains weight as the context is added: turning the scientist's claims into a fair-witness statement required collecting together information about who the scientist is, what their training is, and what their peers think of them. But as noted, progressive trust isn't something that occurs in a single transaction: it's revealed over time. We don't _want_ it to all be revealed at once, because that could result in information overload for someone consulting a claim and could have privacy implications for the witness. A progressive trust model of fair witnessing requires that you show what you must and that you withhold what’s not needed—until it is. ### Privacy and Accountability, Together This model strikes a crucial balance. On one hand, it empowers individuals (fair witnesses) to speak from experience without needing permission from a centralized authority. On the other hand, it allows others to verify the integrity of the claim **without requiring total exposure**. There are numerous use cases: * You can prove you were trained without revealing your name. * You can demonstrate personal observation without revealing your exact location. * You can commit to a fact today and prove you knew it later. ### Fair Witnessing with Gordian Envelope The demands of Fair Witnessing go beyond the capabilities of traditional verifiable credentials (VCs), primarily because VCs can't remove signed information but maintain its validation—and the ability to do so is critically important if you want to nest information for revelation over time. Fortunately, a technology already exists that provides this precise capability: Blockchain Commons' [Gordian Envelope](https://developer.blockchaincommons.com/envelope/), which allows for: the organized storage of information; the validation of that information through signatures; the elision of that information; the continued validation of the information after elision; and the provable restoration of that information. Any subject, predicate, or object in Gordian Envelope can itself be a claim, optionally encrypted or elided. This enables a deeply contextual, inspectable form of expression. For example: - Alice could make a fair-witness observation, which would be an envelope. - Information on the context of Alice's assertion can be a sub-envelope. - A credential for fair witness training can be a sub-envelope. - Endorsements of Alice's work as a fair witness can be sub-envelopes. - Endorsements, credentials, and even the entire envelope can be signed by the appropriate parties. - Any envelope or sub-envelope can be elided, without affecting these signatures and without impacting the ability to provably restore the data later. It's progressive trust appropriate for use with fair witnessing in an existing form! ### Toward a New Epistemology Being a Fair Witness isn’t about declaring truth. It’s about saying what's known, with context, so others can assess what's truth. Truth, in this model, is _interpreted_, not imposed. A verifier—or a jury—decides if a claim is credible, not because a central authority says so, but because the Fair Witness has provided information with sufficient context and endorsements. In other words, fair witnessing is not about **what is true**, but about **how we responsibly say what we believe to be true—and what others can do with that**. This is epistemology (the theory of knowledge) that's structured as a graph. It's cryptographically sound, privacy-respecting, and human-auditable. It reflects real-world trust: messy, contextual, and layered. By modeling that complexity rather than flattening it, we gain both rigor and realism. ### Conclusion In a world of machine-generated misinformation, ideological polarization, and institutional distrust, we must return to the foundations: observation, context, and human responsibility. Fair witnessing offers a new path forward—one that is verifiable, privacy-respecting, and grounded in how humans actually trust. Learn more: [ [Progressive Trust](https://developer.blockchaincommons.com/progressive-trust/) | [Gordian Envelope](https://github.com/BlockchainCommons/Research/blob/master/papers/bcr-2024-006-envelope-graph.md) ] === ## Oldest draft (2022?) In today's world, where information is readily available at the touch of a button, it's more important than ever to ensure that the information we receive is accurate, unbiased, and objective. However, determining the truth can be challenging due to our own knowledge, our cognitive biases, or institutional powers with financial and/or emotional incentives to spread misinformation or even malicious disinformation. An insight from my work in the Decentralized Identity community is that at one point we named one part of our effort "Verifiable Claims". However, we soon realized that people were joining the effort because they believed they could use it to fight disinformation, when in reality, it merely authenticated who the author of the claim was, or the provenance of the entity (sensor, camera, organzation) that stands behind the claim. Thus we ultimately renamed the specification "Verifiable Credentials" but we still see people and organization desiring to use them to define some form of truth (in particular as a "trustful" source of truth, see my article on Progressive Trust). With the rise of powerful artificial intelligence tools like ChatGPT, there is also growing concern about their ability to manipulate the truth or spread misinformation. Some have proposed codifying "truth" in their models, but this also carries significant risks. Instead, I propose that we explore enhancing the authentication and provenance of Verifiable Claims through the use of the concept called "fair witnessing." Science fiction author Robert Heinlein first wrote about fair witnessing in a book title Stranger in a Strange Land. In it, he suggested a profession where people are trained to be absolutely impartial in their assessments, and to speak only from their direct experience, without inference or speculation. He called these people Fair Witnesses. If you point to a distant house and ask one of Heinlein’s Fair Witness what color it is, he or she will say, “It appears to be painted white on this side.” Thus a fair witness would be someone (a person, not a machine) who has been trained to observe and report on events in an objective and unbiased manner. It's important to understand that fair witnessing is not just about reporting on events as they happen, but also about the mindset and approach a fair witness takes when observing and reporting. This includes being aware of and documenting their own biases, and actively working to eliminate them, as well as understanding the cultural, social, and political factors that may influence their observation and how it is reported. Thus a Verifiable Claim made by a fair witness would include not only authentication of their identity as the author of the claim, but also information about their training and reputation as a fair witness, as well as statements such as "personally observed on date and location," "observation conducted in partnership with Y organization", "observation conducted independently, without outside influence", "observation conducted with possible personal biases of X, and efforts were made to minimize them", "observation conducted with a specific understanding of any the following possible financial, political or personal interests that could influence the observation or report, and methodology to attempt to avoid them". "observation conducted with a specific understanding of these possible conflicts of interest, and efforts made to minimize them", "Observation conducted with a specific understanding of these possible ethical concerns, and efforts made to address them", etc. I'm not quite sure exactly what a good ontology is for fair witness claims, but I think it is worthy of discussion. Another advantage of fair witness claims is that they put the power of interpreting the "truth" of the claim by the holder, rather than imposing them by a centralized issuer. I can decide if you've sufficiently sourced your observations, mitigated your biases, and addressed conflicts of interest, and ask others of their opionions on your claim. An ethical challenge with fair witness claims is that they can violate personal privacy of the witness. After all, the whole point of their claim is to correlate for truth. But I believe that with elision and proof of inclusion tools, that you can balance the risks of publicy making a claim against the risks of making it. A fair witness can demonstrate that they were trained, without necessarily revealing their identity, or putting their identity information under escrow. The need for accurate and unbiased information is more critical than ever. To combat the spread of misinformation and disinformation, we need to take a proactive approach and focus on enhancing going beyond authentication of the source of a claim, and move toward greater provenance and detailed criteria about the claim. === ## Fair Witness Ontology: A Graph Model for Trustworthy Assertions In today’s world of ubiquitous claims and AI-generated content, how do we share and verify statements in a way that protects privacy, supports nuance, and grows trust over time? The answer may lie in a concept from science fiction: **fair witnessing**. Originally coined by Robert Heinlein, a fair witness is someone who sees and reports facts with impartiality and transparency, offering not truth, but _how_ they saw what they saw. We now have the technical tools to bring that ideal into the digital world. Using [Gordian Envelopes](https://datatracker.ietf.org/doc/draft-mcnally-envelope/09/) — a format that supports deeply nested, selectively disclosed, signed assertions — we can build a new kind of graph-based ontology for claims: one that supports **progressive trust**, **peer-to-peer attestations**, and **claims about claims**. ## How This Is Different From Existing Approaches Current digital credential and claim systems face significant limitations: - **Binary Trust Models**: Systems like Verifiable Credentials or TLS certificates often require an all-or-nothing trust stance. You either trust the issuer, or you don't. There’s no way to gradually increase confidence in a claim over time. - **Overexposure of Data**: Traditional approaches often force individuals to over-disclose personal information (e.g., showing your full birthdate to prove you’re over 18), undermining privacy. - **Opaque Reasoning**: Most existing claims are not easily inspectable. Why should a verifier trust the claim? What evidence or reasoning supports it? These details are typically buried or unavailable. - **Centralized Issuer Dependence**: Many models rely heavily on a trusted issuing institution. This excludes decentralized communities or individuals and fails when institutions are biased, corrupt, or unavailable. - **Lack of Context or Provenance**: Claims are often decontextualized. There’s no record of how they were made, what biases were present, or under what conditions the claim is valid. The **Fair Witness Ontology** with Gordian Envelopes addresses all of these: - It supports **non-binary, progressive trust** using layered disclosures, endorsements, and optional fuzzy logic scores. - It enables **selective disclosure** with cryptographic elision and Merkle inclusion proofs. - It offers a **graph-based model** where each claim is a navigable, interpretable network of nested assertions. - It embraces **decentralized, peer-to-peer assertion**. Anyone can claim; trust is built through corroboration, not central control. - It records **meta-context** like observation conditions, bias mitigation efforts, and the identity (or pseudonym) of the claimant. - It supports **claims about claims**, enabling recursive context, dispute resolution, and transparency. ## Core Principles The **Fair Witness Ontology** enables: - **Selective Disclosure**: Claims can be partially revealed. Like a matryoshka doll, the outer assertion is visible while inner details (e.g., identity, evidence) remain encrypted or redacted until needed. Gordian Envelopes enable Merkle-based elision and proof of inclusion. - **Progressive Trust**: Trust grows in stages, not a binary yes/no. With each additional disclosure or endorsement, a claim's credibility increases. This supports a fuzzy logic approach where confidence levels range on a spectrum (e.g., 0.3 to 0.9), rather than pass/fail. - **Human-Comprehensible, Machine-Interpretable**: Graphs of who-said-what are easily understood by humans (like a jury), but structured enough for software agents to evaluate. - **Decentralized, Peer-to-Peer**: Claims don't require a central authority. Peers can make statements about themselves or others: "@ShannonA co-authored _Meeples Together_ with me." These statements can stand independently or be supported through endorsements. - **Contextual Reasoning**: Every claim can embed bias disclosures, conflict of interest flags, methodology notes, or ethical considerations. Verifiers can interpret these according to their own policies. ## Building Blocks of the Ontology ### Node Types - **Person/Agent**: Anyone making or being the subject of a claim. Identified with pseudonyms or keys. Identity envelopes can be selectively disclosed to balance privacy and verification. - **Claim/Assertion**: A nested graph of (subject, predicate, object) triples. Every claim is itself an envelope, optionally including context, evidence, and endorsements. - **Meta-Assertion**: A claim about another claim (e.g., "Bob asserts that Alice's claim is unverified", or "Carol believes claim X was made under duress"). These are fully supported via nested triples. - **Evidence**: A piece of data or another claim backing a statement. May be disclosed later. This can include hashes, timestamps, documents, and multimedia. - **Endorsement**: A peer's signed support for another claim or person. Endorsements are first-class envelopes, allowing them to carry their own evidence or assertions. - **Context/Meta-Claim**: Who, when, and how a claim was made. Useful for assessing bias, capacity, and scope. Examples include "observation conducted independently" or "with possible biases mitigated." ### Predicate Types - **asserts / claims**: Links a person to a claim: "Alice asserts [Claim]." This helps explicitly model provenance. - **observed / witnessed**: First-hand observation: "Alice witnessed Bob sign the contract." This predicate signals direct experiential knowledge. - **endorses**: Peer vouching: "Bob endorses Alice as a Fair Witness." This models reputational flows and supports trust bootstrapping. - **co-authored / participated / graduatedFrom**: Domain-specific predicates used in nested factual claims. These enrich the content with precise semantics. - **supports / backedBy**: Ties a claim to evidence or another supporting statement. These can also be recursively nested. - **conflictedBy / mitigatedBy**: Indicate where personal, institutional, or systemic biases might exist, and what steps were taken to address them. ## Example: Peer-to-Peer Authorship 1. **Inner Claim**: "ShannonA co-authored _Meeples Together_." 2. **Outer Envelope**: "ChristopherA asserts [Inner Claim]." 3. **Endorsement**: "Bob endorses [ChristopherA's claim]." 4. **Meta-Claim**: "Carol asserts that Bob's endorsement was influenced by a conflict of interest." Each of these is an envelope that can be signed and independently verified. Only the outer layer needs to be revealed initially. If more validation is needed (e.g., proof of co-authorship), inner layers can be revealed with permission or under legal obligation. The integrity of elided parts can be confirmed via Merkle proofs. ## Progressive Trust in Action This ontology supports the principles described in [Progressive Trust](https://developer.blockchaincommons.com/progressive-trust/). Trust starts low, and increases as: - The claimant reveals more about themselves - Supporting evidence is disclosed - Peers add endorsements Rather than requiring absolute belief in a single certificate, the relying party (human or machine) gradually builds confidence. For example, a pseudonymous claim may start at 0.3 confidence, grow to 0.7 with peer support, and hit 0.9 after document hashes are verified. Trust is _graded_, not binary. Evaluation engines can apply community-defined policies to weigh types of evidence, prior reputation, and independent endorsements. Multiple communities may interpret the same claim differently, depending on their own thresholds, histories, or trust policies. ## Use Case: Personally Observed Event _Alice witnesses Bob sign a contract._ - Alice asserts: "I witnessed Bob sign contract_X at 5 PM, Jan 10." - Nested: Bob → signed → contract_X - Optional encrypted layer: Video of the event A judge could request Alice to reveal the video hash, or confirm her identity as a notary. If the judge accepts her credibility, the claim gains weight. If challenged, more layers can be peeled. The Gordian Envelope structure ensures that unrevealed elements can later be disclosed with verifiable integrity. Alice might have disclosed biases (e.g., "Bob is my business partner") and documented mitigation efforts ("I stood apart during signing; an auditor was present"). ## Use Case: Endorsing a Fair Witness _Bob and Carol endorse Alice as a reliable observer._ - Bob → endorses → Alice (as Fair Witness) - Carol → endorses → Alice Endorsements may themselves contain sub-claims ("I observed Alice testify accurately in Case123") or be contested by others ("I believe Carol is biased toward Alice due to prior collaboration"). Verifiers who trust Bob or Carol may elevate their trust in Alice’s assertions without needing centralized credentials. This creates a rich, interpretable web-of-trust-like model with a memory of relationships, contexts, and history. ## A Living Web of Assertions This ontology turns the internet into a **living graph of claims**, where: - Each envelope carries its own cryptographic integrity - Anyone can make statements about events, themselves, or others - Trust is accumulated through context, support, and community Nested statements like "observation conducted with mitigation of political biases" or "claim evaluated by community X" can be recorded without revealing sensitive data. Software agents and human verifiers can **traverse this graph**, following trails of assertions, endorsements, and evidence. The path of reasoning is inspectable, explainable, and contestable. ## Privacy and Accountability, Together Because of the envelope structure: - A fair witness can prove their training or credibility without revealing their name - Sensitive claims can remain hashed or encrypted - Inner data can be proven to exist and be unchanged (via Merkle proofs) This balances **accountability** with **privacy**: a person can make strong public claims without doxxing themselves unless absolutely required. Elision techniques provide flexible control over disclosure based on evolving needs and audiences. Claim presentation can be tailored to different audiences. A jury may see a fully expanded claim with all supporting evidence, while a casual observer sees only a high-level assertion with minimal disclosure. ## Towards a Decentralized Epistemology The Fair Witness Ontology isn't about declaring truth. It’s about: - Saying exactly what you know - Providing context for your observation - Enabling others to assess and build trust in your claims Truth, in this model, is _interpreted_, not imposed. A verifier (or jury) decides if a claim is credible, not because a central authority signed it, but because the envelope contains: - A clear statement - Supporting context and evidence - Endorsements from trusted peers - Meta-claims that explain or critique its creation This is **epistemology as a graph** — structured, cryptographically sound, privacy-respecting, and human-auditable. It reflects real-world trust: messy, contextual, and layered. By modeling that complexity rather than flattening it, we gain both rigor and realism. ## Beyond the Graph: Refinements and Future Directions As the Fair Witness Ontology matures from concept to implementation, several advanced themes and future opportunities emerge. These refinements push the ontology beyond a static technical model into a dynamic sociotechnical foundation for decentralized trust. ### Naming and Framing: Evolving Language for Evolving Roles The term **"Fair Witness"** is inspired by Heinlein's depiction of trained observers who report only what they see. While evocative and conceptually powerful, it may not be ideal in every cultural, legal, or technical context. Some alternatives include: - **Verifiable Observer** — emphasizes independently checkable experience. - **Trustful Notary** — captures both signing and impartiality roles. - **Decentralized Reporter** — evokes peer-generated, non-institutional information. - **Progressive Witness** — aligns with the "progressive trust" lifecycle. A broader ontology can support multiple role terms and semantic tags that reflect different use cases or communities, allowing adaptable and brandable identities that still plug into a shared structural foundation. ### Privacy Without Identity: Anonymous Trust and Zero-Knowledge Inclusion The ontology assumes that some claims must be made **pseudonymously or anonymously**, especially in high-risk, whistleblower, or adversarial environments. Still, the underlying data must be verifiable. This is where privacy-enhancing technologies strengthen fairwitnessing: - **Zero-Knowledge Proofs (ZKPs)** allow a witness to prove they have received a certain training or hold a credential **without revealing its contents**. - **Selective disclosure** via Gordian Envelope allows structured elision: encrypted elements of an envelope can be exposed only to authorized verifiers. - **Escrow mechanisms** support identity recovery or legal release when appropriate, without defaulting to exposure. - **Verifiable pseudonyms** (e.g., DIDs backed by strong signing histories) enable building reputations without linking to real-world names. Together, these methods ensure that **truthful claims do not require total identity surrender**, enabling safe participation in adversarial information ecosystems. ### Social and Interpretive Trust: From Data to Dialogue Trust is not a score—it's a conversation. Fairwitnessing supports: - **Community-driven interpretations** of claims, where multiple parties debate the sufficiency of evidence or the presence of conflict. - **Jury-style interfaces** that expose just enough context for a group of verifiers to evaluate a layered claim, piece by piece. - **Disputable claims**, where counterclaims or meta-assertions can be recorded transparently and without erasure. This creates a **dialogic model of truth**: multiple fair witnesses can observe the same event differently, and both perspectives can coexist in the graph until contextually adjudicated. ### Tooling: Interfaces, Agents, and Auditability For this ontology to be actionable at scale, the infrastructure must support: - **Claim Authoring Tools**: Mobile or desktop apps for fair witnesses to easily issue assertions with embedded context and metadata. - **Graph Explorers**: Visual, interactive tools to navigate layered, nested envelopes—especially useful for jurors, auditors, or researchers. - **Policy Engines**: Configurable trust logic interpreters that allow verifiers to define "sufficient proof" thresholds tailored to their domain or application. - **Community Annotation Interfaces**: Mechanisms to endorse, contest, or discuss claims within social networks or governance bodies. This tooling must support both **machine automation** and **human reasoning**, surfacing the right amount of detail to the right audience, at the right time. ### Defense Against Disinformation: A Constructive Countermeasure Disinformation spreads through ambiguity, virality, and opacity. Fairwitnessing addresses these weaknesses directly: - **Provenance Tracking**: By embedding who, where, and how a claim was made, verifiers can trace origins and filter manipulations. - **Bias Disclosure**: Claims can explicitly document known conflicts and mitigation efforts, signaling epistemic integrity. - **Contestability and Plurality**: Competing claims can coexist in the graph, but each must present its backing. - **Trusted Endorsements**: Peer endorsements, like citations, show who else stands behind a claim, forming webs of corroboration. - **Layered Evidence Trails**: Instead of requiring blind faith, claims gradually unfold with verifiable backing—if challenged, more can be revealed. This model doesn’t assume perfect objectivity. It assumes **disciplined transparency**, enabling participants to build and defend claims in a shared epistemological space. ### Policy-Driven Disclosure and Audience-Specific Views Not all verifiers need the same level of information. The ontology supports **policy-driven disclosure**: - A casual reader may see a top-level summary claim. - A moderator or peer-review board may require contextual details. - A legal process may unlock encrypted identities or location data under due process. With audience-specific presentation layers, each envelope becomes a **responsive disclosure object** that adapts to the verifier's role and risk threshold. ## Learn More ### Naming and Framing While "Fair Witness" is rooted in Heinlein's fictional vision of objective observation, the term also evokes civic duty, impartiality, and human presence. However, it may be worth exploring alternative or companion terms such as *Verifiable Observer*, *Decentralized Reporter*, or *Trustful Notary* to suit different cultural or technological contexts. The ontology remains flexible to be rebranded while preserving its core semantics. ### Privacy Without Identity One of the powerful affordances of Gordian Envelopes is the ability to prove qualifications (e.g., "I am a trained fair witness") without revealing the identity of the witness. This can be achieved through: - **Anonymous credentials** or **Zero-Knowledge Proofs (ZKPs)** - **Selective disclosure of training certificates or organizational affiliations** - **Escrow-based identity reveal** (under warrant, legal order, or consent) This allows fair witnesses to assert their credibility without sacrificing their personal privacy. ### Social and Interpretive Trust Trust in fairwitness claims is not just computed—it is negotiated. Verifiers can: - Cross-check evidence - Evaluate the bias disclosures and mitigation methods - Discuss claims within communities or juries This models trust as a **social process**, not merely a cryptographic function. The ontology supports community vetting, rebuttals, and the layering of interpretations over time. ### Tools for Fairwitnessing For this model to succeed in practice, it needs usable tools: - **Apps for recording, signing, and issuing claims** - **Interfaces for exploring trust graphs visually** - **Review panels or juries who can traverse and debate claims** These tools could resemble social media but with integrity, or legal testimony but without the courtroom. ### A Defense Against Disinformation Fairwitnessing also has a broader purpose: to counter the spread of misinformation and disinformation. By embedding: - Verifiable provenance (who, when, how) - Transparent bias and conflict documentation - Layered, inspectable evidence trails ...this ontology supports a **culture of transparency**, even under pseudonymity. It gives people a way to rebut false claims, contextualize dubious information, and hold each other to higher standards of informational integrity. ## Learn More - [Gordian Envelope Specification](https://datatracker.ietf.org/doc/draft-mcnally-envelope/09/) - [Gordian Envelope Graph Model](https://github.com/BlockchainCommons/Research/blob/master/papers/bcr-2024-006-envelope-graph.md) - [Progressive Trust](https://developer.blockchaincommons.com/progressive-trust/) The future of claims isn't centralized. It's fairwitnessed.