Skip to Content

Changes for page FAQ

Last modified by Robert Schaub on 2025/12/23 18:00
From version 1.3
edited by Robert Schaub
on 2025/12/22 14:32
Change comment: Renamed back-links.
To version 1.1
edited by Robert Schaub
on 2025/12/22 14:22
Change comment: Imported from XAR

Summary

Details

Page properties
Content
... ... @@ -1,20 +1,13 @@
1 1  = Frequently Asked Questions (FAQ) =
2 -
3 3  Common questions about FactHarbor's design, functionality, and approach.
4 -
5 5  == 1. How do claims get evaluated in FactHarbor? ==
6 -
7 7  === 1.1 User Submission ===
8 -
9 9  **Who**: Anyone can submit claims
10 10  **Process**: User submits claim text + source URLs
11 11  **Speed**: Typically <20 seconds to verdict
12 -
13 13  === 1.2 AKEL Processing (Automated) ===
14 -
15 15  **What**: AI Knowledge Extraction Layer analyzes claim
16 16  **Steps**:
17 -
18 18  * Parse claim into testable components
19 19  * Extract evidence from provided sources
20 20  * Score source credibility
... ... @@ -23,12 +23,9 @@
23 23  * Publish automatically
24 24  **Authority**: AKEL makes all content decisions
25 25  **Scale**: Can process millions of claims
26 -
27 27  === 1.3 Continuous Improvement (Human Role) ===
28 -
29 29  **What**: Humans improve the system, not individual verdicts
30 30  **Activities**:
31 -
32 32  * Monitor aggregate performance metrics
33 33  * Identify systematic errors
34 34  * Propose algorithm improvements
... ... @@ -36,11 +36,8 @@
36 36  * Test changes before deployment
37 37  **NOT**: Reviewing individual claims for approval
38 38  **Focus**: Fix the system, not the data
39 -
40 40  === 1.4 Exception Handling ===
41 -
42 42  **When AKEL flags for review**:
43 -
44 44  * Low confidence verdict
45 45  * Detected manipulation attempt
46 46  * Unusual pattern requiring attention
... ... @@ -49,19 +49,14 @@
49 49  * Takes action on abuse/manipulation
50 50  * Proposes detection improvements
51 51  * Does NOT override verdicts
52 -
53 53  === 1.5 Why This Model Works ===
54 -
55 55  **Scale**: Automation handles volume humans cannot
56 56  **Consistency**: Same rules applied uniformly
57 57  **Transparency**: Algorithms can be audited
58 58  **Improvement**: Systematic fixes benefit all claims
59 -
60 60  == 2. What prevents FactHarbor from becoming another echo chamber? ==
61 -
62 62  FactHarbor includes multiple safeguards against echo chambers and filter bubbles:
63 63  **Mandatory Contradiction Search**:
64 -
65 65  * AI must actively search for counter-evidence, not just confirmations
66 66  * System checks for echo chamber patterns in source clusters
67 67  * Flags tribal or ideological source clustering
... ... @@ -82,12 +82,9 @@
82 82  * Multiple independent nodes with different perspectives
83 83  * No single entity controls "the truth"
84 84  * Cross-node contradiction detection
85 -
86 86  == 3. How does FactHarbor handle claims that are "true in one context but false in another"? ==
87 -
88 88  This is exactly what FactHarbor is designed for:
89 89  **Scenarios capture contexts**:
90 -
91 91  * Each scenario defines specific boundaries, definitions, and assumptions
92 92  * The same claim can have different verdicts in different scenarios
93 93  * 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?)
... ... @@ -99,11 +99,8 @@
99 99  * Every scenario states its assumptions clearly
100 100  * Users can compare how changing assumptions changes conclusions
101 101  * Makes context-dependence visible, not hidden
102 -
103 103  == 4. What makes FactHarbor different from traditional fact-checking sites? ==
104 -
105 105  **Traditional Fact-Checking**:
106 -
107 107  * Binary verdicts: True / Mostly True / False
108 108  * Single interpretation chosen by fact-checker
109 109  * Often hides legitimate contextual differences
... ... @@ -115,11 +115,8 @@
115 115  * **Version history**: Shows how understanding evolves
116 116  * **Contradiction search**: Actively seeks opposing evidence
117 117  * **Federated**: No single authority controls truth
118 -
119 119  == 5. How do you prevent manipulation or coordinated misinformation campaigns? ==
120 -
121 121  **Quality Gates**:
122 -
123 123  * Automated checks before AI-generated content publishes
124 124  * Source quality verification
125 125  * Mandatory contradiction search
... ... @@ -142,12 +142,9 @@
142 142  * No single point of control
143 143  * Cross-node contradiction detection
144 144  * Trust model prevents malicious node influence
145 -
146 146  == 6. What happens when new evidence contradicts an existing verdict? ==
147 -
148 148  FactHarbor is designed for evolving knowledge:
149 149  **Automatic Re-evaluation**:
150 -
151 151  1. New evidence arrives
152 152  2. System detects affected scenarios and verdicts
153 153  3. AKEL proposes updated verdicts
... ... @@ -155,7 +155,6 @@
155 155  5. New verdict version published
156 156  6. Old versions remain accessible
157 157  **Version History**:
158 -
159 159  * Every verdict has complete history
160 160  * Users can see "as of date X, what did we know?"
161 161  * Timeline shows how understanding evolved
... ... @@ -167,12 +167,9 @@
167 167  * Users following claims are notified of updates
168 168  * Can compare old vs new verdicts
169 169  * Can see which evidence changed conclusions
170 -
171 171  == 7. Who can submit claims to FactHarbor? ==
172 -
173 173  **Anyone** - even without login:
174 174  **Readers** (no login required):
175 -
176 176  * Browse and search all published content
177 177  * Submit text for analysis
178 178  * New claims added automatically unless duplicates exist
... ... @@ -183,7 +183,6 @@
183 183  * Suggest scenarios
184 184  * Participate in discussions
185 185  **Workflow**:
186 -
187 187  1. User submits text (as Reader or Contributor)
188 188  2. AKEL extracts claims
189 189  3. Checks for existing duplicates
... ... @@ -192,12 +192,9 @@
192 192  6. Generates scenarios (draft)
193 193  7. Runs quality gates
194 194  8. Publishes as AI-Generated (Mode 2) if passes
195 -
196 196  == 8. What are "risk tiers" and why do they matter? ==
197 -
198 198  Risk tiers determine review requirements and publication workflow:
199 199  **Tier A (High Risk)**:
200 -
201 201  * **Domains**: Medical, legal, elections, safety, security, major financial
202 202  * **Publication**: AI can publish with warnings, expert review required for "AKEL-Generated" status
203 203  * **Audit rate**: Recommendation 30-50%
... ... @@ -216,11 +216,8 @@
216 216  * AKEL suggests tier based on domain, keywords, impact
217 217  * Moderators and Trusted Contributors can override
218 218  * Risk tiers reviewed based on audit outcomes
219 -
220 220  == 9. How does federation work and why is it important? ==
221 -
222 222  **Federation Model**:
223 -
224 224  * Multiple independent FactHarbor nodes
225 225  * Each node has own database, AKEL, governance
226 226  * Nodes exchange claims, scenarios, evidence, verdicts
... ... @@ -232,7 +232,6 @@
232 232  * **Specialization**: Domain-focused nodes (health, energy, etc.)
233 233  * **Trust diversity**: Multiple perspectives, not single truth source
234 234  **How Nodes Exchange Data**:
235 -
236 236  1. Local node creates versions
237 237  2. Builds signed bundle
238 238  3. Pushes to trusted neighbor nodes
... ... @@ -240,16 +240,12 @@
240 240  5. Accept or branch versions
241 241  6. Local re-evaluation if needed
242 242  **Trust Model**:
243 -
244 244  * Trusted nodes → auto-import
245 245  * Neutral nodes → import with review
246 246  * Untrusted nodes → manual only
247 -
248 248  == 10. Can experts disagree in FactHarbor? ==
249 -
250 250  **Yes - and that's a feature, not a bug**:
251 251  **Multiple Scenarios**:
252 -
253 253  * Trusted Contributors can create different scenarios with different assumptions
254 254  * Each scenario gets its own verdict
255 255  * Users see *why* experts disagree (different definitions, boundaries, evidence weighting)
... ... @@ -266,12 +266,9 @@
266 266  * Different nodes can have different expert conclusions
267 267  * Cross-node branching allowed
268 268  * Users can see how conclusions vary across nodes
269 -
270 270  == 11. What prevents AI from hallucinating or making up facts? ==
271 -
272 272  **Multiple Safeguards**:
273 273  **Quality Gate 4: Structural Integrity**:
274 -
275 275  * Fact-checking against sources
276 276  * No hallucinations allowed
277 277  * Logic chain must be valid and traceable
... ... @@ -294,12 +294,9 @@
294 294  * Tier A marked as highest risk
295 295  * Audit sampling catches errors
296 296  * Community can flag issues
297 -
298 298  == 12. How does FactHarbor make money / is it sustainable? ==
299 -
300 300  [ToDo: Business model and sustainability to be defined]
301 301  Potential models under consideration:
302 -
303 303  * Non-profit foundation with grants and donations
304 304  * Institutional subscriptions (universities, research organizations, media)
305 305  * API access for third-party integrations
... ... @@ -306,37 +306,25 @@
306 306  * Premium features for power users
307 307  * Federated node hosting services
308 308  Core principle: **Public benefit** mission takes priority over profit.
309 -
310 310  == 13. Related Pages ==
311 -
312 312  * [[Requirements (Roles)>>Test.FactHarbor.Specification.Requirements.WebHome]]
313 -* [[AKEL (AI Knowledge Extraction Layer)>>Test.FactHarbor pre12 V0\.9\.70.Specification.AI Knowledge Extraction Layer (AKEL).WebHome]]
314 -* [[Automation>>Test.FactHarbor pre12 V0\.9\.70.Specification.Automation.WebHome]]
259 +* [[AKEL (AI Knowledge Extraction Layer)>>Test.FactHarbor.Specification.AI Knowledge Extraction Layer (AKEL).WebHome]]
260 +* [[Automation>>Test.FactHarbor.Specification.Automation.WebHome]]
315 315  * [[Federation & Decentralization>>Test.FactHarbor.Specification.Federation & Decentralization.WebHome]]
316 316  * [[Mission & Purpose>>Test.FactHarbor.Organisation.Core Problems FactHarbor Solves.WebHome]]
317 -
318 318  == 20. Glossary / Key Terms ==
319 -
320 320  === Phase 0 vs POC v1 ===
321 -
322 322  These terms refer to the same stage of FactHarbor's development:
323 -
324 324  * **Phase 0** - Organisational perspective: Pre-alpha stage with founder-led governance
325 325  * **POC v1** - Technical perspective: Proof of Concept demonstrating AI-generated publication
326 326  Both describe the current development stage where the platform is being built and initially validated.
327 -
328 328  === Beta 0 ===
329 -
330 330  The next development stage after POC, featuring:
331 -
332 332  * External testers
333 333  * Basic federation experiments
334 334  * Enhanced automation
335 -
336 336  === Release 1.0 ===
337 -
338 338  The first public release featuring:
339 -
340 340  * Full federation support
341 341  * 2000+ concurrent users
342 342  * Production-grade infrastructure