Wiki source code of Federation & Decentralization
Last modified by Robert Schaub on 2026/02/08 08:26
Hide last authors
| author | version | line-number | content |
|---|---|---|---|
 |
1.1 | 1 | = Federation & Decentralization = |
| 2 | |||
| 3 | FactHarbor is designed to operate as a **federated network of nodes** rather than a single central server. | ||
| 4 | |||
| 5 | Decentralization provides: | ||
| |
2.4 | 6 | |
| |
1.1 | 7 | * **Resilience** against censorship or political pressure |
| 8 | * **Autonomy** for local governance and moderation | ||
| 9 | * **Scalability** across many independent communities | ||
| 10 | * **Trust** without centralized control | ||
| 11 | * **Domain specialization** (health-focused nodes, energy-focused nodes, etc.) | ||
| 12 | |||
| 13 | FactHarbor draws inspiration from the Fediverse but uses stronger structure, versioning, and integrity guarantees. | ||
| 14 | |||
| 15 | |||
| 16 | == 1. Federation Architecture Diagram == | ||
| 17 | |||
| 18 | The following diagram shows the complete federated architecture with node components and communication layers. | ||
| 19 | |||
| |
2.12 | 20 | {{include reference="Archive.FactHarbor 2026\.01\.20.Specification.Diagrams.Federation Architecture.WebHome"/}} |
| |
1.1 | 21 | |
| 22 | |||
| 23 | == 2. Federated FactHarbor Nodes == | ||
| 24 | |||
| 25 | Each FactHarbor instance ("node") maintains: | ||
| |
2.4 | 26 | |
| |
1.1 | 27 | * Its own database |
| 28 | * Its own AKEL instance | ||
| 29 | * Its own reviewers, experts, and contributors | ||
| 30 | * Its own governance rules | ||
| 31 | |||
| 32 | Nodes exchange structured information: | ||
| |
2.4 | 33 | |
| |
1.1 | 34 | * Claims |
| 35 | * Scenarios | ||
| 36 | * Evidence metadata (not necessarily full files) | ||
| 37 | * Verdicts (optional) | ||
| 38 | * Hashes and signatures for integrity | ||
| 39 | |||
| 40 | Nodes choose which external nodes they trust. | ||
| 41 | |||
| 42 | |||
| 43 | == 3. Global Identifiers == | ||
| 44 | |||
| 45 | Every entity receives a globally unique, linkable identifier. | ||
| 46 | |||
| |
2.1 | 47 | **Format**: |
| |
1.1 | 48 | `factharbor://node_url/type/local_id` |
| 49 | |||
| |
2.1 | 50 | **Example**: |
| |
1.1 | 51 | `factharbor://factharbor.energy/claim/CLM-55812` |
| 52 | |||
| 53 | **Supported types**: | ||
| |
2.4 | 54 | |
| |
1.1 | 55 | * `claim` |
| 56 | * `scenario` | ||
| 57 | * `evidence` | ||
| 58 | * `verdict` | ||
| 59 | * `user` (optional) | ||
| 60 | * `cluster` | ||
| 61 | |||
| 62 | **Properties**: | ||
| |
2.4 | 63 | |
| |
1.1 | 64 | * Globally consistent |
| 65 | * Human-readable | ||
| 66 | * Hash-derived | ||
| 67 | * Independent of database internals | ||
| 68 | * URL-resolvable (future enhancement) | ||
| 69 | |||
| 70 | This allows cross-node references and prevents identifier collisions in federated environments. | ||
| 71 | |||
| 72 | |||
| 73 | == 4. Trust Model == | ||
| 74 | |||
| 75 | Each node maintains a **trust table** defining relationships with other nodes: | ||
| 76 | |||
| 77 | === 4.1 Trust Levels === | ||
| 78 | |||
| 79 | **Trusted Nodes**: | ||
| |
2.4 | 80 | |
| |
1.1 | 81 | * Claims auto-imported |
| 82 | * Scenarios accepted without re-review | ||
| 83 | * Evidence considered valid | ||
| 84 | * Verdicts displayed to users | ||
| 85 | * High synchronization priority | ||
| 86 | |||
| 87 | **Neutral Nodes**: | ||
| |
2.4 | 88 | |
| |
1.1 | 89 | * Claims imported but flagged for review |
| 90 | * Scenarios require local validation | ||
| 91 | * Evidence requires re-assessment | ||
| 92 | * Verdicts shown with "external node" disclaimer | ||
| 93 | * Normal synchronization priority | ||
| 94 | |||
| 95 | **Untrusted Nodes**: | ||
| |
2.4 | 96 | |
| |
1.1 | 97 | * Claims quarantined, manual import only |
| 98 | * Scenarios rejected by default | ||
| 99 | * Evidence not accepted | ||
| 100 | * Verdicts not displayed | ||
| 101 | * No automatic synchronization | ||
| 102 | |||
| 103 | === 4.2 Trust Affects === | ||
| 104 | |||
| 105 | * **Auto-import**: Whether claims/scenarios are automatically added | ||
| 106 | * **Re-review requirements**: Whether local reviewers must validate | ||
| 107 | * **Verdict display**: Whether external verdicts are shown to users | ||
| 108 | * **Synchronization frequency**: How often data is exchanged | ||
| 109 | * **Reputation signals**: How external reputation is interpreted | ||
| 110 | |||
| 111 | === 4.3 Local Trust Authority === | ||
| 112 | |||
| 113 | Each node's governance team decides: | ||
| |
2.4 | 114 | |
| |
1.1 | 115 | * Which nodes to trust |
| 116 | * Trust level criteria | ||
| 117 | * Trust escalation/degradation rules | ||
| 118 | * Dispute resolution with partner nodes | ||
| 119 | |||
| 120 | Trust is **local and autonomous** - no global trust registry exists. | ||
| 121 | |||
| 122 | |||
| 123 | == 5. Data Sharing Model == | ||
| 124 | |||
| 125 | === 5.1 What Nodes Share === | ||
| 126 | |||
| 127 | **Always shared** (if federation enabled): | ||
| |
2.4 | 128 | |
| |
1.1 | 129 | * Claims and claim clusters |
| 130 | * Scenario structures | ||
| 131 | * Evidence metadata and content hashes | ||
| 132 | * Integrity signatures | ||
| 133 | |||
| 134 | **Optionally shared**: | ||
| |
2.4 | 135 | |
| |
1.1 | 136 | * Full evidence files (large documents) |
| 137 | * Verdicts (nodes may choose to keep verdicts local) | ||
| 138 | * Vector embeddings | ||
| 139 | * Scenario templates | ||
| 140 | * AKEL distilled knowledge | ||
| 141 | |||
| 142 | **Never shared**: | ||
| |
2.4 | 143 | |
| |
1.1 | 144 | * Internal user lists |
| 145 | * Reviewer comments and internal discussions | ||
| 146 | * Governance decisions and meeting notes | ||
| 147 | * Access control data | ||
| 148 | * Private or sensitive content marked as local-only | ||
| 149 | |||
| 150 | === 5.2 Large Evidence Files === | ||
| 151 | |||
| 152 | Evidence files are: | ||
| |
2.4 | 153 | |
| |
1.1 | 154 | * Stored locally by default |
| 155 | * Referenced via global content hash | ||
| 156 | * Optionally served through IPFS | ||
| 157 | * Accessible via direct peer-to-peer transfer | ||
| 158 | * Can be stored in S3-compatible object storage | ||
| 159 | |||
| 160 | == 6. Synchronization Protocol == | ||
| 161 | |||
| 162 | Nodes exchange data using multiple synchronization methods: | ||
| 163 | |||
| 164 | === 6.1 Push-Based Synchronization === | ||
| 165 | |||
| 166 | **Mechanism**: Webhooks | ||
| 167 | |||
| 168 | When local content changes: | ||
| |
2.4 | 169 | |
| |
1.1 | 170 | 1. Node builds signed bundle |
| 171 | 2. Sends webhook notification to subscribed nodes | ||
| 172 | 3. Remote nodes fetch bundle | ||
| 173 | 4. Remote nodes validate and import | ||
| 174 | |||
| 175 | **Use case**: Real-time updates for trusted partners | ||
| 176 | |||
| 177 | === 6.2 Pull-Based Synchronization === | ||
| 178 | |||
| 179 | **Mechanism**: Scheduled polling | ||
| 180 | |||
| 181 | Nodes periodically: | ||
| |
2.4 | 182 | |
| |
1.1 | 183 | 1. Query partner nodes for updates |
| 184 | 2. Fetch changed entities since last sync | ||
| 185 | 3. Validate and import | ||
| 186 | 4. Store sync checkpoint | ||
| 187 | |||
| 188 | **Use case**: Regular batch updates, lower trust nodes | ||
| 189 | |||
| 190 | === 6.3 Subscription-Based Synchronization === | ||
| 191 | |||
| 192 | **Mechanism**: WebSub-like protocol | ||
| 193 | |||
| 194 | Nodes subscribe to: | ||
| |
2.4 | 195 | |
| |
1.1 | 196 | * Specific claim clusters |
| 197 | * Specific domains (medical, energy, etc.) | ||
| 198 | * Specific scenario types | ||
| 199 | * Verdict updates | ||
| 200 | |||
| 201 | Publisher pushes updates only to subscribers. | ||
| 202 | |||
| 203 | **Use case**: Selective federation, domain specialization | ||
| 204 | |||
| 205 | === 6.4 Large Asset Transfer === | ||
| 206 | |||
| 207 | For files >10MB: | ||
| |
2.4 | 208 | |
| |
1.1 | 209 | * S3-compatible object storage |
| 210 | * IPFS (content-addressed) | ||
| 211 | * Direct peer-to-peer transfer | ||
| 212 | * Chunked HTTP transfer with resume support | ||
| 213 | |||
| 214 | == 7. Federation Sync Workflow == | ||
| 215 | |||
| 216 | Complete synchronization sequence for creating and sharing new content: | ||
| 217 | |||
| 218 | === 7.1 Step 1: Local Node Creates New Versions === | ||
| 219 | |||
| 220 | User or AKEL creates: | ||
| |
2.4 | 221 | |
| |
1.1 | 222 | * New claim version |
| 223 | * New scenario version | ||
| 224 | * New evidence version | ||
| 225 | * New verdict version | ||
| 226 | |||
| 227 | All changes tracked with: | ||
| |
2.4 | 228 | |
| |
1.1 | 229 | * VersionID |
| 230 | * ParentVersionID | ||
| 231 | * AuthorType | ||
| 232 | * Timestamp | ||
| 233 | * JustificationText | ||
| 234 | |||
| 235 | === 7.2 Step 2: Federation Layer Builds Signed Bundle === | ||
| 236 | |||
| 237 | Federation layer packages: | ||
| |
2.4 | 238 | |
| |
1.1 | 239 | * Entity data (claim, scenario, evidence metadata, verdict) |
| 240 | * Version lineage (ParentVersionID chain) | ||
| 241 | * Cryptographic signatures | ||
| 242 | * Node provenance information | ||
| 243 | * Trust metadata | ||
| 244 | |||
| 245 | Bundle format: | ||
| |
2.4 | 246 | |
| |
1.1 | 247 | * JSON-LD for structured data |
| 248 | * Content-addressed hashes | ||
| 249 | * Digital signatures for integrity | ||
| 250 | |||
| 251 | === 7.3 Step 3: Bundle Includes Required Data === | ||
| 252 | |||
| 253 | Each bundle contains: | ||
| |
2.4 | 254 | |
| |
1.1 | 255 | * **Claims**: Full claim text, classification, domain |
| 256 | * **Scenarios**: Definitions, assumptions, boundaries | ||
| 257 | * **Evidence metadata**: Source URLs, hashes, reliability scores (not always full files) | ||
| 258 | * **Verdicts**: Likelihood ranges, uncertainty, reasoning chains | ||
| 259 | * **Lineage**: Version history, parent relationships | ||
| 260 | * **Signatures**: Cryptographic proof of origin | ||
| 261 | |||
| 262 | === 7.4 Step 4: Bundle Pushed to Trusted Neighbor Nodes === | ||
| 263 | |||
| 264 | Based on trust table: | ||
| |
2.4 | 265 | |
| |
1.1 | 266 | * Push to **trusted nodes** immediately |
| 267 | * Queue for **neutral nodes** (batched) | ||
| 268 | * Skip **untrusted nodes** | ||
| 269 | |||
| 270 | Push methods: | ||
| |
2.4 | 271 | |
| |
1.1 | 272 | * Webhook notification |
| 273 | * Direct API call | ||
| 274 | * Pub/Sub message queue | ||
| 275 | |||
| 276 | === 7.5 Step 5: Remote Nodes Validate Lineage and Signatures === | ||
| 277 | |||
| 278 | Receiving node: | ||
| |
2.4 | 279 | |
| |
1.1 | 280 | 1. Verifies cryptographic signatures |
| 281 | 2. Validates version lineage (ParentVersionID chain) | ||
| 282 | 3. Checks for conflicts with local data | ||
| 283 | 4. Validates data structure and required fields | ||
| 284 | 5. Applies local trust policies | ||
| 285 | |||
| 286 | Validation failures → reject or quarantine bundle | ||
| 287 | |||
| 288 | === 7.6 Step 6: Accept or Branch Versions === | ||
| 289 | |||
| 290 | **Accept** (if validation passes): | ||
| |
2.4 | 291 | |
| |
1.1 | 292 | * Import new versions |
| 293 | * Maintain provenance metadata | ||
| 294 | * Link to local related entities | ||
| 295 | * Update local indices | ||
| 296 | |||
| 297 | **Branch** (if conflict detected): | ||
| |
2.4 | 298 | |
| |
1.1 | 299 | * Create parallel version tree |
| 300 | * Mark as "external branch" | ||
| 301 | * Allow local reviewers to merge or reject | ||
| 302 | * Preserve both version histories | ||
| 303 | |||
| 304 | **Reject** (if validation fails): | ||
| |
2.4 | 305 | |
| |
1.1 | 306 | * Log rejection reason |
| 307 | * Notify source node (optional) | ||
| 308 | * Quarantine for manual review (optional) | ||
| 309 | |||
| 310 | === 7.7 Step 7: Local Re-evaluation Runs if Required === | ||
| 311 | |||
| 312 | After import, local node checks: | ||
| |
2.4 | 313 | |
| |
1.1 | 314 | * Does new evidence affect existing verdicts? |
| 315 | * Do new scenarios require re-assessment? | ||
| 316 | * Are there contradictions with local content? | ||
| 317 | |||
| 318 | If yes: | ||
| |
2.4 | 319 | |
| |
1.1 | 320 | * Trigger AKEL re-evaluation |
| 321 | * Queue for reviewer attention | ||
| 322 | * Update affected verdicts | ||
| 323 | * Notify users following related content | ||
| 324 | |||
| 325 | == 8. Cross-Node AI Knowledge Exchange == | ||
| 326 | |||
| 327 | Each node runs its own AKEL instance and may exchange AI-derived knowledge: | ||
| 328 | |||
| 329 | === 8.1 What Can Be Shared === | ||
| 330 | |||
| 331 | **Vector embeddings**: | ||
| |
2.4 | 332 | |
| |
1.1 | 333 | * For cross-node claim clustering |
| 334 | * For semantic search alignment | ||
| 335 | * Never includes training data | ||
| 336 | |||
| 337 | **Canonical claim forms**: | ||
| |
2.4 | 338 | |
| |
1.1 | 339 | * Normalized claim text |
| 340 | * Standard phrasing templates | ||
| 341 | * Domain-specific formulations | ||
| 342 | |||
| 343 | **Scenario templates**: | ||
| |
2.4 | 344 | |
| |
1.1 | 345 | * Reusable scenario structures |
| 346 | * Common assumption patterns | ||
| 347 | * Evaluation method templates | ||
| 348 | |||
| 349 | **Contradiction alerts**: | ||
| |
2.4 | 350 | |
| |
1.1 | 351 | * Detected conflicts between claims |
| 352 | * Evidence conflicts across nodes | ||
| 353 | * Scenario incompatibilities | ||
| 354 | |||
| 355 | **Metadata and insights**: | ||
| |
2.4 | 356 | |
| |
1.1 | 357 | * Aggregate quality metrics |
| 358 | * Reliability signal extraction | ||
| 359 | * Bubble detection patterns | ||
| 360 | |||
| 361 | === 8.2 What Can NEVER Be Shared === | ||
| 362 | |||
| 363 | **Model weights**: No sharing of trained model parameters | ||
| 364 | |||
| 365 | **Training data**: No sharing of full training datasets | ||
| 366 | |||
| 367 | **Local governance overrides**: AKEL suggestions can be overridden locally | ||
| 368 | |||
| 369 | **User behavior data**: No cross-node tracking | ||
| 370 | |||
| 371 | **Internal review discussions**: Private content stays private | ||
| 372 | |||
| 373 | === 8.3 Benefits of AI Knowledge Exchange === | ||
| 374 | |||
| 375 | * Reduced duplication across nodes | ||
| 376 | * Improved claim clustering accuracy | ||
| 377 | * Faster contradiction detection | ||
| 378 | * Shared scenario libraries | ||
| 379 | * Cross-node quality improvements | ||
| 380 | |||
| 381 | === 8.4 Local Control Maintained === | ||
| 382 | |||
| 383 | * Nodes accept or reject shared AI knowledge | ||
| 384 | * Human reviewers can override any AKEL suggestion | ||
| 385 | * Local governance always has final authority | ||
| 386 | * No external AI control over local content | ||
| 387 | * Privacy-preserving knowledge exchange | ||
| 388 | |||
| 389 | == 9. Decentralized Processing == | ||
| 390 | |||
| 391 | Each node independently performs: | ||
| |
2.4 | 392 | |
| |
1.1 | 393 | * AKEL processing |
| 394 | * Scenario drafting and validation | ||
| 395 | * Evidence review | ||
| 396 | * Verdict calculation | ||
| 397 | * Truth landscape summarization | ||
| 398 | |||
| 399 | Nodes can specialize: | ||
| |
2.4 | 400 | |
| |
1.1 | 401 | * Health-focused node with medical experts |
| 402 | * Energy-focused node with domain knowledge | ||
| 403 | * Small node delegating scenario libraries to partners | ||
| 404 | * Regional node with language/culture specialization | ||
| 405 | |||
| 406 | Optional data sharing includes: | ||
| |
2.4 | 407 | |
| |
1.1 | 408 | * Embeddings for clustering |
| 409 | * Claim clusters for alignment | ||
| 410 | * Scenario templates for efficiency | ||
| 411 | * Verdict comparison metadata | ||
| 412 | |||
| 413 | == 10. Scaling to Thousands of Users == | ||
| 414 | |||
| 415 | Nodes scale independently through: | ||
| |
2.4 | 416 | |
| |
1.1 | 417 | * Horizontally scalable API servers |
| 418 | * Worker pools for AKEL tasks | ||
| 419 | * Hybrid storage (local + S3/IPFS) | ||
| 420 | * Redis caching for performance | ||
| 421 | * Sharded or partitioned databases | ||
| 422 | |||
| 423 | Federation allows effectively unlimited horizontal scaling by adding new nodes. | ||
| 424 | |||
| 425 | Communities may form: | ||
| |
2.4 | 426 | |
| |
1.1 | 427 | * Domain-specific nodes (epidemiology, energy, climate) |
| 428 | * Language or region-based nodes | ||
| 429 | * NGO or institutional nodes | ||
| 430 | * Private organizational nodes | ||
| 431 | * Academic research nodes | ||
| 432 | |||
| 433 | Nodes cooperate through: | ||
| |
2.4 | 434 | |
| |
1.1 | 435 | * Scenario library sharing |
| 436 | * Shared or overlapping claim clusters | ||
| 437 | * Expert delegation between nodes | ||
| 438 | * Distributed AKEL task support | ||
| 439 | * Cross-node quality audits | ||
| 440 | |||
| 441 | == 11. Federation and Release 1.0 == | ||
| 442 | |||
| 443 | **POC**: Single node, optional federation experiments | ||
| 444 | |||
| 445 | **Beta 0**: 2-3 nodes, basic federation protocol | ||
| 446 | |||
| 447 | **Release 1.0**: Full federation support with: | ||
| |
2.4 | 448 | |
| |
1.1 | 449 | * Robust synchronization protocol |
| 450 | * Trust model implementation | ||
| 451 | * Cross-node AI knowledge exchange | ||
| 452 | * Federated search and discovery | ||
| 453 | * Distributed audit collaboration | ||
| 454 | * Inter-node expert consultation | ||
| 455 | |||
| 456 | == 12. Related Pages == | ||
| 457 | |||
| |
2.9 | 458 | * [[AKEL (AI Knowledge Extraction Layer)>>Archive.FactHarbor 2026\.01\.20.Specification.AI Knowledge Extraction Layer (AKEL).WebHome]] |
| |
2.11 | 459 | * [[Data Model>>Archive.FactHarbor 2026\.01\.20.Specification.Data Model.WebHome]] |
| |
2.10 | 460 | * [[Architecture>>Archive.FactHarbor 2026\.01\.20.Specification.Architecture.WebHome]] |
| |
2.13 | 461 | * [[Workflows>>Archive.FactHarbor 2026\.01\.20.Specification.Workflows.WebHome]] |