Research / requirement

Current research

Requirements — risk matrix

Type: requirement Updated: 2026-10-05 Scale: cross-jurisdictional Path: requirements/RISK-MATRIX.md

requirementsrisk

Sources: SRC-0030SRC-0031SRC-0032SRC-0038SRC-0047SRC-0070SRC-0078SRC-0100SRC-0101SRC-0102SRC-0106SRC-0109SRC-0111SRC-0115SRC-0118SRC-0120SRC-0122SRC-0123SRC-0125

Risk matrix

CA-R2 requirements record. Updated: 2026-10-05. Risks drawn from the accepted corpus — each cites its evidence basis. No numerical risk scores are assigned: the corpus does not justify quantification. “Impact” is qualitative and refers to institutional legitimacy damage.

RiskStageImpactMitigation requirementResidual/open questionEvidence basis
Manipulated selection parametersSelectioncatastrophic — silent capture of membershipspecification is a signed governance artefact before any run; runs cannot proceed without it; full published methodologyhow unauthorised-parameter changes are detected post-hocSRC-0078, SRC-0109, SRC-0123
Biased invitation frameFramehigh — structural exclusion before selectionrecord frame type + coverage gap publicly; frame choice is a declared governance decisionno frame is complete; coverage-gap disclosure standard undefinedSRC-0120, SRC-0021
Self-selection distortionResponsehigh — respondent pool skews educated/engagedpublish funnel; stratification + overweighting to correct; never claim pool = populationcorrection is partial by nature; acceptable residue undefinedSRC-0120, SRC-0038
Inappropriate strataSelection spechigh — engineered compositionstrata + rationale + legal basis in published specification; contested criteria flagged (attitudinal)who adjudicates strata legitimacy disputesSRC-0109, SRC-0038, SRC-0032
Privacy leakageAll stageshigh — member/respondent harm + chilling effectdata minimisation; special-category handling per jurisdiction; small-cell suppression in publicationsmall-cell publication risk unresolved corpus-wideSRC-0032, SRC-0122, SRC-0031
Participant targeting / lobbyingMembership, deliberationhigh — members steered by interestsmember contact protection; lobbying disclosure rules; protected deliberation option (Brussels model)balance transparency vs protection per institutionSRC-0030, SRC-0101
Duplicate participationRegistration/selectionmedium — gaming, seat double-occupancydeduplication rules across pool/reserves/members; identity-verification policy declaredverification strength vs privacy trade-offcorpus methods records
Replacement manipulationReplacementhigh — backdoor membership controlreplacement only under declared protocol; every event logged with provenance; reserve ordering rules publishedmax replacement rate before re-draw needed is undefinedSRC-0073, SRC-0074
Agenda captureAgenda/remithigh — output steered before deliberation startsagenda authority declared; proposer/approver/exclusion-reasons recorded; member/public channels where the model grants themleast-studied power in corpus; no settled safeguardSRC-0070, SRC-0100
Expert/evidence captureEvidencehigh — evidence framing steers judgementexpert selection rationale + disclosures + member-requested evidence rights + contestability procedurebalance assessment remains a human judgement — by whom?SRC-0038, SRC-0100, SRC-0129
Facilitator biasDeliberationhigh — least auditable corpus stagefacilitation method + assignment recorded; oversight mechanism; session structure documentedmeasurement of influence unsolvedSRC-0106, SRC-0115
Inaccessible participationMembership/deliberationhigh — seating ≠ participationprovisions per INCLUSION-ACCESSIBILITY.md; exclusion events recordedenforcement of provision deliverySRC-0038, SRC-0101
Digital exclusionAll member-facing stagesmedium-high — excludes offline populationoffline/postal/telephone parity for every required interactioncost of offline parity at scaleSRC-0088, SRC-0120
Opaque votingDecisionhigh — unverifiable outputsdecision record contract per DECISION-AND-DISSENT.md; method + electorate + result preservedsecret-ballot verification without identity linkageSRC-0038, SRC-0116
Suppression of dissentDecision/publicationmedium-high — false consensus impressiondissent/minority objects are first-class; support levels published per recommendationminority-statement trigger thresholdsSRC-0017, SRC-0047, SRC-0111
Misleading claims of representativenessPublication/communicationhigh — legitimacy fraud (esp. open assemblies presented as random)legitimacy basis is a declared property; mass input carries bias label; self-selection never renderable as sortitionenforcement is reputational not technicalSRC-0066, SRC-0106
Ignored recommendationsResponsehigh — hollow institutionresponse obligation declared + tracked; no response is a visible state; deadlines computabletracking is not enforceabilitySRC-0027, SRC-0043
Misleading implementation claimsImplementationhigh — declared ≠ deliveredresponse/commitment/implementation are separate objects; progress evidence linked, self-report labelledwho verifies implementation evidenceSRC-0027, SRC-0043
Concentration of power in permanent membersPermanent institutionsmedium-high — incumbency capturerotation + bounded terms + repeat-service rules; memory in records not tenureoptimal rotation fraction unknownSRC-0021, SRC-0118, SRC-0109
Capture in delegated/nested systemsMulti-levelmedium-high — delegate class emergesdelegation events auditable; fresh-draw fractions preserved; descriptive-claims not composed upwarddelegation vs fresh-sortition ratio is constitutional, not technicalSRC-0111, SRC-0125
Confusion between AI suggestions and human decisionsAny AI-assisted surfacemedium-high — authority launderingAI artefacts labelled and provenanced as advisory; never actions of authority (CA-D007)interface conventions for advisory vs authoritativeCA-D007, THEORY-TO-PRACTICE

Cross-cutting observations

Sources