Changes for page FAQ

Last modified by Robert Schaub on 2025/12/24 20:33

From 1.1 to 2.1 From 3.1 to 3.2

From version 2.1

edited by Robert Schaub
on 2025/12/14 23:02

Change comment: There is no comment for this version

To version 3.1

edited by Robert Schaub
on 2025/12/15 16:56

Change comment: Imported from XAR

Raw
Rendered

Summary

Page properties (1 modified, 0 added, 0 removed)

Details

Page properties

Content

@@ -6,24 +6,70 @@
  == How do facts get input into the system? ==
--FactHarbor uses a hybrid model:
++FactHarbor uses a hybrid model combining three complementary approaches:
--**~1. **AI-Generated (scalable)**: System dynamically researches claims—extracting, generating structured sub-queries, performing mandatory contradiction search (actively seeking counter-evidence, not just confirmations), running quality gates. Published with clear "AI-Generated" labels.**
++=== 1. AI-Generated Content (Scalable) ===
--**2. Expert-Authored (authoritative)**: Domain experts directly author, edit, and validate content—especially for high-risk domains (medical, legal). These get "Human-Reviewed" status and higher trust.
++**What**: System dynamically researches claims using AKEL (AI Knowledge Extraction Layer)
--**3. Audit-Improved (continuous quality)**: Sampling audits (30-50% high-risk, 5-10% low-risk) where expert reviews systematically improve AI research quality.
++**Process**:
++* Extracts claims from submitted text
++* Generates structured sub-queries
++* Performs **mandatory contradiction search** (actively seeks counter-evidence, not just confirmations)
++* Runs automated quality gates
++* Publishes with clear "AI-Generated" labels
--**Why both matter**:
++**Publication**: Mode 2 (public, AI-labeled) when quality gates pass
--* AI research handles scale—emerging claims, immediate responses with transparent reasoning
--* Expert authoring provides authoritative grounding for critical domains
--* Audit feedback ensures AI quality improves based on expert validation patterns
++**Purpose**: Handles scale — emerging claims get immediate responses with transparent reasoning
--Experts can author high-priority content directly, validate/edit AI outputs, or audit samples to improve system-wide performance—focusing their time where expertise matters most.
++=== 2. Expert-Authored Content (Authoritative) ===
--POC v1 demonstrates the AI research pipeline (fully automated with transparent reasoning); full system supports all three pathways.
++**What**: Domain experts directly author, edit, and validate content
++**Focus**: High-risk domains (medical, legal, safety-critical)
++
++**Publication**: Mode 3 ("Human-Reviewed" status) with expert attribution
++
++**Authority**: Tier A content requires expert approval
++
++**Purpose**: Provides authoritative grounding for critical domains where errors have serious consequences
++
++=== 3. Audit-Improved Quality (Continuous) ===
++
++**What**: Sampling audits where experts review AI-generated content
++
++**Rates**:
++* High-risk (Tier A): 30-50% sampling
++* Medium-risk (Tier B): 10-20% sampling
++* Low-risk (Tier C): 5-10% sampling
++
++**Impact**: Expert feedback systematically improves AI research quality
++
++**Purpose**: Ensures AI quality evolves based on expert validation patterns
++
++=== Why All Three Matter ===
++
++**Complementary Strengths**:
++* **AI research**: Scale and speed for emerging claims
++* **Expert authoring**: Authority and precision for critical domains
++* **Audit feedback**: Continuous quality improvement
++
++**Expert Time Optimization**:
++
++Experts can choose where to focus their time:
++* Author high-priority content directly
++* Validate and edit AI-generated outputs
++* Audit samples to improve system-wide AI performance
++
++This focuses expert time where domain expertise matters most while leveraging AI for scale.
++
++=== Current Status ===
++
++**POC v1**: Demonstrates the AI research pipeline (fully automated with transparent reasoning and quality gates)
++
++**Full System**: Will support all three pathways with integrated workflow
++
  ----
  == What prevents FactHarbor from becoming another echo chamber? ==
@@ -31,7 +31,6 @@
  FactHarbor includes multiple safeguards against echo chambers and filter bubbles:
  **Mandatory Contradiction Search**:
--
  * AI must actively search for counter-evidence, not just confirmations
  * System checks for echo chamber patterns in source clusters
  * Flags tribal or ideological source clustering
@@ -38,25 +38,21 @@
  * Requires diverse perspectives across political/ideological spectrum
  **Multiple Scenarios**:
--
  * Claims are evaluated under different interpretations
  * Reveals how assumptions change conclusions
  * Makes disagreements understandable, not divisive
  **Transparent Reasoning**:
--
  * All assumptions, definitions, and boundaries are explicit
  * Evidence chains are traceable
  * Uncertainty is quantified, not hidden
  **Audit System**:
--
  * Human auditors check for bubble patterns
  * Feedback loop improves AI search diversity
  * Community can flag missing perspectives
  **Federation**:
--
  * Multiple independent nodes with different perspectives
  * No single entity controls "the truth"
  * Cross-node contradiction detection
@@ -68,23 +68,20 @@
  This is exactly what FactHarbor is designed for:
  **Scenarios capture contexts**:
--
  * Each scenario defines specific boundaries, definitions, and assumptions
  * The same claim can have different verdicts in different scenarios
  * Example: "Coffee is healthy" depends on:
--** Definition of "healthy" (reduces disease risk? improves mood? affects specific conditions?)
--** Population (adults? pregnant women? people with heart conditions?)
--** Consumption level (1 cup/day? 5 cups/day?)
--** Time horizon (short-term? long-term?)
++  ** Definition of "healthy" (reduces disease risk? improves mood? affects specific conditions?)
++  ** Population (adults? pregnant women? people with heart conditions?)
++  ** Consumption level (1 cup/day? 5 cups/day?)
++  ** Time horizon (short-term? long-term?)
  **Truth Landscape**:
--
  * Shows all scenarios and their verdicts side-by-side
  * Users see *why* interpretations differ
  * No forced consensus when legitimate disagreement exists
  **Explicit Assumptions**:
--
  * Every scenario states its assumptions clearly
  * Users can compare how changing assumptions changes conclusions
  * Makes context-dependence visible, not hidden
@@ -94,7 +94,6 @@
  == What makes FactHarbor different from traditional fact-checking sites? ==
  **Traditional Fact-Checking**:
--
  * Binary verdicts: True / Mostly True / False
  * Single interpretation chosen by fact-checker
  * Often hides legitimate contextual differences
@@ -101,7 +101,6 @@
  * Limited ability to show *why* people disagree
  **FactHarbor**:
--
  * **Multi-scenario**: Shows multiple valid interpretations
  * **Likelihood-based**: Ranges with uncertainty, not binary labels
  * **Transparent assumptions**: Makes boundaries and definitions explicit
@@ -114,7 +114,6 @@
  == How do you prevent manipulation or coordinated misinformation campaigns? ==
  **Quality Gates**:
--
  * Automated checks before AI-generated content publishes
  * Source quality verification
  * Mandatory contradiction search
@@ -121,13 +121,11 @@
  * Bubble detection for coordinated campaigns
  **Audit System**:
--
  * Stratified sampling catches manipulation patterns
  * Expert auditors validate AI research quality
  * Failed audits trigger immediate review
  **Transparency**:
--
  * All reasoning chains are visible
  * Evidence sources are traceable
  * AKEL involvement clearly labeled
@@ -134,13 +134,11 @@
  * Version history preserved
  **Moderation**:
--
  * Moderators handle abuse, spam, coordinated manipulation
  * Content can be flagged by community
  * Audit trail maintained even if content hidden
  **Federation**:
--
  * Multiple nodes with independent governance
  * No single point of control
  * Cross-node contradiction detection
@@ -153,7 +153,6 @@
  FactHarbor is designed for evolving knowledge:
  **Automatic Re-evaluation**:
--
 . New evidence arrives
 . System detects affected scenarios and verdicts
 . AKEL proposes updated verdicts
@@ -162,19 +162,16 @@
 . Old versions remain accessible
  **Version History**:
--
  * Every verdict has complete history
  * Users can see "as of date X, what did we know?"
  * Timeline shows how understanding evolved
  **Transparent Updates**:
--
  * Reason for re-evaluation documented
  * New evidence clearly linked
  * Changes explained, not hidden
  **User Notifications**:
--
  * Users following claims are notified of updates
  * Can compare old vs new verdicts
  * Can see which evidence changed conclusions
@@ -186,7 +186,6 @@
  **Anyone** - even without login:
  **Readers** (no login required):
--
  * Browse and search all published content
  * Submit text for analysis
  * New claims added automatically unless duplicates exist
@@ -193,7 +193,6 @@
  * System deduplicates and normalizes
  **Contributors** (logged in):
--
  * Everything Readers can do
  * Submit evidence sources
  * Suggest scenarios
@@ -200,7 +200,6 @@
  * Participate in discussions
  **Workflow**:
--
 . User submits text (as Reader or Contributor)
 . AKEL extracts claims
 . Checks for existing duplicates
@@ -217,7 +217,6 @@
  Risk tiers determine review requirements and publication workflow:
  **Tier A (High Risk)**:
--
  * **Domains**: Medical, legal, elections, safety, security, major financial
  * **Publication**: AI can publish with warnings, expert review required for "Human-Reviewed" status
  * **Audit rate**: Recommendation 30-50%
@@ -224,7 +224,6 @@
  * **Why**: Potential for significant harm if wrong
  **Tier B (Medium Risk)**:
--
  * **Domains**: Complex policy, science causality, contested issues
  * **Publication**: AI can publish immediately with clear labeling
  * **Audit rate**: Recommendation 10-20%
@@ -231,7 +231,6 @@
  * **Why**: Nuanced but lower immediate harm risk
  **Tier C (Low Risk)**:
--
  * **Domains**: Definitions, established facts, historical data
  * **Publication**: AI publication default
  * **Audit rate**: Recommendation 5-10%
@@ -238,7 +238,6 @@
  * **Why**: Well-established, low controversy
  **Assignment**:
--
  * AKEL suggests tier based on domain, keywords, impact
  * Moderators and Experts can override
  * Risk tiers reviewed based on audit outcomes
@@ -248,7 +248,6 @@
  == How does federation work and why is it important? ==
  **Federation Model**:
--
  * Multiple independent FactHarbor nodes
  * Each node has own database, AKEL, governance
  * Nodes exchange claims, scenarios, evidence, verdicts
@@ -255,7 +255,6 @@
  * No central authority
  **Why Federation Matters**:
--
  * **Resilience**: No single point of failure or censorship
  * **Autonomy**: Communities govern themselves
  * **Scalability**: Add nodes to handle more users
@@ -263,7 +263,6 @@
  * **Trust diversity**: Multiple perspectives, not single truth source
  **How Nodes Exchange Data**:
--
 . Local node creates versions
 . Builds signed bundle
 . Pushes to trusted neighbor nodes
@@ -272,7 +272,6 @@
 . Local re-evaluation if needed
  **Trust Model**:
--
  * Trusted nodes → auto-import
  * Neutral nodes → import with review
  * Untrusted nodes → manual only
@@ -284,19 +284,16 @@
  **Yes - and that's a feature, not a bug**:
  **Multiple Scenarios**:
--
  * Experts can create different scenarios with different assumptions
  * Each scenario gets its own verdict
  * Users see *why* experts disagree (different definitions, boundaries, evidence weighting)
  **Parallel Verdicts**:
--
  * Same scenario, different expert interpretations
  * Both verdicts visible with expert attribution
  * No forced consensus
  **Transparency**:
--
  * Expert reasoning documented
  * Assumptions stated explicitly
  * Evidence chains traceable
@@ -303,7 +303,6 @@
  * Users can evaluate competing expert opinions
  **Federation**:
--
  * Different nodes can have different expert conclusions
  * Cross-node branching allowed
  * Users can see how conclusions vary across nodes
@@ -315,7 +315,6 @@
  **Multiple Safeguards**:
  **Quality Gate 4: Structural Integrity**:
--
  * Fact-checking against sources
  * No hallucinations allowed
  * Logic chain must be valid and traceable
@@ -322,7 +322,6 @@
  * References must be accessible and verifiable
  **Evidence Requirements**:
--
  * Primary sources required
  * Citations must be complete
  * Sources must be accessible
@@ -329,13 +329,11 @@
  * Reliability scored
  **Audit System**:
--
  * Human auditors check AI-generated content
  * Hallucinations caught and fed back into training
  * Patterns of errors trigger system improvements
  **Transparency**:
--
  * All reasoning chains visible
  * Sources linked
  * Users can verify claims against sources
@@ -342,7 +342,6 @@
  * AKEL outputs clearly labeled
  **Human Oversight**:
--
  * Tier A requires expert review for "Human-Reviewed" status
  * Audit sampling catches errors
  * Community can flag issues
@@ -354,7 +354,6 @@
  [ToDo: Business model and sustainability to be defined]
  Potential models under consideration:
--
  * Non-profit foundation with grants and donations
  * Institutional subscriptions (universities, research organizations, media)
  * API access for third-party integrations
@@ -371,4 +371,5 @@
  * [[AKEL (AI Knowledge Extraction Layer)>>FactHarbor.Specification.AI Knowledge Extraction Layer (AKEL).WebHome]]
  * [[Automation>>FactHarbor.Specification.Automation.WebHome]]
  * [[Federation & Decentralization>>FactHarbor.Specification.Federation & Decentralization.WebHome]]
--* [[Mission & Purpose>>FactHarbor.Organisation.Mission & Purpose.WebHome]]
++* [[Mission & Purpose>>FactHarbor.Organisation.Core Problems FactHarbor Solves.WebHome]]
++

Changes for page FAQ

Summary

Details

Applications

Navigation

Need help?