Skip to Content

Wiki source code of Governance

Last modified by Robert Schaub on 2025/12/24 20:29
Show last authors
1 = Governance =
2
3 FactHarbor is governed collaboratively with clear separation between **organizational policy and decisions** and **technical implementation**.
4
5 == 1. Governance Structure ==
6
7 {{include reference="FactHarbor.Organisation.Diagrams.Governance Structure.WebHome"/}}
8
9 * **Governing Team** – Sets high-level policy, organizational direction, funding priorities
10 * **Lead** – Coordinates execution, represents organization publicly
11 * **Core Maintainers** – Technical and specification decisions, code/spec review
12 * **Domain Experts** – Subject-matter authority in specialized areas
13 * **Community Contributors** – Feedback, proposals, and participation in decision-making
14
15
16 == 2. Decision-Making Levels ==
17
18 === 2.1 Technical Decisions (Maintainers) ===
19
20 **Scope**: Architecture, data model, AKEL configuration, quality gates, system performance
21
22 **Process**:
23 * Proposals discussed in technical forums
24 * Review by core maintainers
25 * Consensus-based approval
26 * Breaking changes require broader community input
27 * Quality gate adjustments require rationale and audit validation
28
29 **Examples**:
30 * Adding new quality gate
31 * Adjusting AKEL parameters
32 * Modifying audit sampling algorithms
33 * Database schema changes
34
35 === 2.2 Policy Decisions (Governing Team + Community) ===
36
37 **Scope**: Risk tier policies, publication rules, content guidelines, ethical boundaries
38
39 **Process**:
40 * Proposal published for community feedback
41 * Discussion period (recommendation: minimum 14 days for major changes)
42 * Governing Team decision with community input
43 * Transparency in reasoning
44 * Risk tier policy changes require Expert consultation
45
46 **Examples**:
47 * Defining Tier A domains
48 * Setting audit sampling rates
49 * Content moderation policies
50 * Community guidelines
51
52 === 2.3 Domain-Specific Decisions (Experts) ===
53
54 **Scope**: Domain quality standards, source reliability in specialized fields, Tier A content validation
55
56 **Process**:
57 * Expert consensus in domain
58 * Documented reasoning
59 * Review by other experts
60 * Escalation to Governing Team if unresolved
61 * Experts set domain-specific audit criteria
62
63 **Examples**:
64 * Medical claim evaluation standards
65 * Legal citation requirements
66 * Scientific methodology thresholds
67 * Tier A approval criteria by domain
68
69
70 == 3. AI and Human Roles in Governance ==
71
72 === 3.1 Human-Only Governance Decisions ===
73
74 The following can **never** be automated:
75
76 * **Ethical boundary setting** – What content is acceptable, what harm thresholds exist
77 * **Risk tier policy** – Which domains are Tier A/B/C (though AKEL can suggest)
78 * **Audit system oversight** – Quality standards, sampling strategies, auditor selection
79 * **Dispute resolution** – Conflicts between experts, controversial decisions
80 * **Community guidelines enforcement** – Bans, suspensions, conflict mediation
81 * **Organizational direction** – Mission, vision, funding priorities
82
83 === 3.2 AKEL Advisory Role ===
84
85 AKEL can **assist but not decide**:
86
87 * Suggest risk tier assignments (humans validate)
88 * Flag content for expert review (humans decide)
89 * Identify patterns in audit failures (humans adjust policy)
90 * Propose quality gate refinements (maintainers approve)
91 * Detect emerging topics needing new policies (Governing Team decides)
92
93 === 3.3 Transparency Requirement ===
94
95 All governance decisions must be:
96 * **Documented** with reasoning
97 * **Published** for community visibility
98 * **Reviewable** by community members
99 * **Reversible** if evidence of error or harm
100
101
102 == 4. Audit System Governance ==
103
104 === 4.1 Audit Oversight Committee ===
105
106 **Composition**: Maintainers, Domain Experts, and Governing Team member(s)
107
108 **Responsibilities**:
109 * Set quality standards for audit evaluation
110 * Review audit statistics and trends
111 * Adjust sampling rates based on performance
112 * Approve changes to audit algorithms
113 * Oversee auditor selection and rotation
114 * Publish transparency reports
115
116 **Meeting Frequency**: Recommendation: Regular meetings as needed
117
118 **Reporting**: Recommendation: Periodic transparency reports to community
119
120 === 4.2 Audit Performance Metrics ===
121
122 Tracked and published:
123 * Audit pass/fail rates by tier
124 * Common failure patterns
125 * System improvements implemented
126 * Time to resolution for audit failures
127 * Auditor performance (anonymized)
128
129 === 4.3 Feedback Loop Governance ===
130
131 **Process**:
132 1. Audits identify patterns in AI errors
133 2. Audit Committee reviews patterns
134 3. Maintainers propose technical fixes
135 4. Changes tested in sandbox
136 5. Community informed of improvements
137 6. Deployed with monitoring
138
139 **Escalation**:
140 * Persistent high failure rates → Pause AI publication in affected tier/domain
141 * Critical errors → Immediate system review
142 * Pattern of harm → Policy revision
143
144
145 == 5. Risk tier Policy Governance ==
146
147 === 5.1 Risk Tier Assignment Authority ===
148
149 * **AKEL**: Suggests initial tier based on domain, keywords, content analysis
150 * **Moderators**: Can override AKEL for individual content
151 * **Experts**: Set tier policy for their domains
152 * **Governing Team**: Approve tier policy changes, resolve tier disputes
153
154 === 5.2 Risk Tier Review Process ===
155
156 **Triggers for Review**:
157 * Significant audit failures in a tier
158 * New emerging topics or domains
159 * Community flags systematic misclassification
160 * Expert domain recommendations
161 * Periodic policy review
162
163 **Process**:
164 1. Expert domain review (identify if Tier A/B/C appropriate)
165 2. Community input period (recommendation: sufficient time for feedback)
166 3. Audit Committee assessment (error patterns in current tier)
167 4. Governing Team decision
168 5. Implementation with monitoring period
169 6. Transparency report on rationale
170
171 === 5.3 Current Tier Assignments (Baseline) ===
172
173 **Tier A**: Medical, legal, elections, safety/security, major financial decisions
174
175 **Tier B**: Complex science causality, contested policy, historical interpretation with political implications, significant economic impact
176
177 **Tier C**: Established historical facts, simple definitions, well-documented scientific consensus, basic reference info
178
179 **Note**: These are guidelines; edge cases require expert judgment
180
181
182 == 6. Quality Gate Governance ==
183
184 === 6.1 Quality Gate Modification Process ===
185
186 **Who Can Propose**: Maintainers, Experts, Audit Committee
187
188 **Requirements**:
189 * Rationale based on audit failures or system improvements
190 * Testing in sandbox environment
191 * Impact assessment (false positive/negative rates)
192 * Community notification before deployment
193
194 **Approval**:
195 * Technical changes: Maintainer consensus
196 * Policy changes (e.g., new gate criteria): Governing Team approval
197
198 **Examples of Governed Changes**:
199 * Adjusting contradiction search scope
200 * Modifying source reliability thresholds
201 * Adding new bubble detection patterns
202 * Changing uncertainty quantification formulas
203
204
205 == 7. Community Participation ==
206
207 === 7.1 Open Discussion Forums ===
208
209 * Technical proposals (maintainer-led)
210 * Policy proposals (Governing Team-led)
211 * Domain-specific discussions (Expert-led)
212 * Audit findings and improvements (Audit Committee-led)
213
214 === 7.2 Proposal Mechanism ===
215
216 Anyone can propose:
217 1. Submit proposal with rationale
218 2. Community discussion (recommendation: minimum timeframe for feedback)
219 3. Relevant authority reviews (Maintainers/Governing Team/Experts)
220 4. Decision with documented reasoning
221 5. Implementation (if approved)
222
223 === 7.3 Transparency ===
224
225 * All decisions documented in public wiki
226 * Audit statistics published periodically
227 * Governing Team meeting minutes published
228 * Expert recommendations documented
229 * Community feedback acknowledged
230
231
232 == 8. Dispute Resolution ==
233
234 === 8.1 Conflict Between Experts ===
235
236 1. Experts attempt consensus
237 2. If unresolved, escalate to Governing Team
238 3. Governing Team appoints neutral expert panel
239 4. Panel recommendation
240 5. Governing Team decision (final)
241
242 === 8.2 Conflict Between Maintainers ===
243
244 1. Discussion in maintainer forum
245 2. Attempt consensus
246 3. If unresolved, Lead makes decision
247 4. Community informed of reasoning
248
249 === 8.3 User Appeals ===
250
251 Users can appeal:
252 * Content rejection decisions
253 * Risk tier assignments
254 * Audit outcomes
255 * Moderation actions
256
257 **Process**:
258 1. Submit appeal with evidence
259 2. Reviewed by independent moderator/expert
260 3. Decision with reasoning
261 4. Final appeal to Governing Team (if warranted)
262
263
264 == 9. Related Pages ==
265
266 * [[AKEL (AI Knowledge Extraction Layer)>>FactHarbor.Specification.AI Knowledge Extraction Layer (AKEL).WebHome]]
267 * [[Automation>>FactHarbor.Specification.Automation.WebHome]]
268 * [[Requirements (Roles)>>FactHarbor.Specification.Requirements.WebHome]]
269 * [[Organisational Model>>FactHarbor.Organisation.Organisational-Model]]