research-document

Communication Engineering portfolio roadmap

Research Area: General Research
Status: draft
Version: 0.1
Priority: medium
Updated: 2026-07-22
Source: research/frontier/PORTFOLIO-ROADMAP.md

Communication Engineering portfolio roadmap

Portfolio version: 0.2.0 Updated: 2026-07-30
Authority: mutable frontier view governed by DF-COMM-2026-0001
Program objective: develop and validate a context-sensitive framework that helps humans and AI choose communication strategies for specified audiences, objectives, and constraints.

Portfolio thesis

The program should advance from measurable, low-risk communication tasks toward broader explanatory and selection models. It must not begin with an exhaustive style catalog or AI optimizer: both depend on validated constructs, outcomes, and boundary conditions.

Prioritization method

Priority is a judgment based on five 0–3 factors:

decision leverage;
uncertainty reduction;
dependency unlock;
feasibility;
risk reduction.

Scores support review but do not replace reasoning. Safety, ethics, missing authority, or decisive evidence can override a total. Ties favor work that creates discriminating evidence and reusable measurement infrastructure.

Current portfolio graph

flowchart TD
    RQ001["RQ-COMM-001<br/>Select pilot task and audience"]
    RQ003["RQ-COMM-003<br/>Define outcome measures"]
    RQ004["RQ-COMM-004<br/>Define constraints"]
    RQ006["RQ-COMM-006<br/>Run feasibility pilot"]
    RQ007["RQ-COMM-007<br/>Run controlled comparison"]
    RQ008["RQ-COMM-008<br/>Build functional taxonomy"]
    RQ009["RQ-COMM-009<br/>Design AI selection model"]
    RQ010["RQ-COMM-010<br/>Validate across contexts"]
    RQ001 --> RQ003
    RQ004 --> RQ001
    RQ001 --> RQ006
    RQ003 --> RQ006
    RQ004 --> RQ006
    RQ006 --> RQ007
    RQ007 --> RQ008
    RQ008 --> RQ009
    RQ007 --> RQ010
    RQ010 --> RQ009

Portfolio items

RQ-COMM-001 — Select a concrete pilot task and audience

Parent: program objective
Children: RQ-COMM-006
Maturity: Scoping
Priority: P0; score 13/15 (leverage 3, uncertainty 3, unlock 3, feasibility 2, risk reduction 2)
Supporting evidence: EV-COMM-2026-0001 through 0003; HY-COMM-2026-0001
Contradicting evidence: none direct
Dependencies: RQ-COMM-003, RQ-COMM-004
Related research: RQ-COMM-005, RQ-COMM-011
Produced artifacts: RP-COMM-2026-0001; MS-COMM-2026-0001
Engineering enabled: experimental harness and reusable task fixtures
Confidence: Medium that procedural documentation is the right class; Low for any specific task
Largest uncertainty: stakeholder value and representative recruitment
Next action: interview or otherwise obtain evidence from at least one intended user group; compare three task candidates against a frozen rubric
Exit evidence: named audience, observed need, safe synthetic task, measurable success/failure, feasible recruitment, and no floor/ceiling result
Review trigger: task candidate requires sensitive data or specialist harm review

RQ-COMM-002 — Test the disciplinary boundary

Parent: program objective
Children: RQ-COMM-008
Maturity: Idea
Priority: P2; score 8/15
Supporting evidence: none
Contradicting evidence: repository charter warns against assuming a new discipline
Dependencies: two completed empirical cycles
Related research: communication theory, HCI, rhetoric, information architecture, technical communication, systems engineering
Produced artifacts: none
Engineering enabled: vocabulary and repository/product boundary decisions
Confidence: Low
Largest uncertainty: whether the proposed field adds predictive or operational value beyond integration of existing fields
Next action: defer comparative review until the program has concrete constructs and outcomes
Exit evidence: explicit comparison of scope, mechanisms, methods, predictions, and incremental utility
Review trigger: any external-facing discipline claim

RQ-COMM-003 — Define a purpose-to-outcome measurement model

Parent: program objective
Children: RQ-COMM-001, RQ-COMM-006, RQ-COMM-007
Maturity: Deep Investigation
Priority: P0; score 14/15
Supporting evidence: EV-COMM-2026-0001 through 0005 and 0007 through 0010; HY-COMM-2026-0003
Contradicting evidence: EV-COMM-2026-0004 and 0009 show metric divergence; EV-COMM-2026-0007 contradicts universal use of time
Dependencies: none for scoping; task selection for validation
Related research: psychometrics, usability, learning science, decision science
Produced artifacts: preliminary matrix below; RP-COMM-2026-0001 and 0002; active mission MS-COMM-2026-0002
Engineering enabled: scoring rubric, instrumentation, acceptance tests
Confidence: Medium
Largest uncertainty: task-specific reliability, sensitivity, burden, and accessibility feasibility
Next action: apply the RP-COMM-2026-0002 scoring specification to three RQ-COMM-001 candidate tasks, then run an independent-rater and floor/ceiling feasibility check
Exit evidence: reliability, validity, sensitivity, burden, and failure-mode assessment for each selected measure
Review trigger: primary outcome chosen or pilot produces metric conflict

Preliminary outcome architecture:

Communication purpose	Primary candidate outcome	Guardrails	Common invalid proxy
Inform	accurate comprehension	calibration, accessibility	reading level
Instruct	unassisted correct task completion	critical errors, assistance, accessibility	preference or speed alone
Teach	delayed transfer to a new problem	misconceptions, effort	immediate recall
Support a decision	decision quality relative to stated values/evidence	uncertainty understanding, harm	conversion
Persuade ethically	informed, autonomous attitude/action change	truth, coercion, subgroup harm	engagement
Coordinate	shared state and correct next actions	latency, repair cost	message count
Build warranted trust	trust calibration to actual reliability	skepticism, correction response	trust score alone

RQ-COMM-004 — Establish ethical, privacy, safety, and accessibility constraints

Parent: program objective
Children: RQ-COMM-001, RQ-COMM-006, RQ-COMM-009
Maturity: Scoping
Priority: P0; score 13/15
Supporting evidence: EV-COMM-2026-0006; GV-ENG-001
Contradicting evidence: none
Dependencies: task/data-flow detail for validation
Related research: RQ-COMM-012
Produced artifacts: synthetic-data constraint in MS-COMM-2026-0001
Engineering enabled: recruitment/data plan, accessibility acceptance criteria, safe optimization boundaries
Confidence: Medium
Largest uncertainty: participant-data classification and applicable review
Next action: write a task-specific data flow and harm analysis after RQ-COMM-001 selects candidates
Exit evidence: data inventory, retention, consent/review determination, accessibility scope, abuse cases, and stop conditions
Review trigger: real user data, vulnerable groups, or high-stakes content

RQ-COMM-005 — Pre-register the ROS comparison baseline

Parent: repository pilot
Children: none
Maturity: Scoping
Priority: P1; score 11/15
Supporting evidence: PILOT-MEASUREMENT-communication-engineering
Contradicting evidence: no comparable prior-project data
Dependencies: RQ-COMM-001 scope freeze
Related research: all first-slice work
Produced artifacts: draft pilot measurement plan
Engineering enabled: ROS continue/simplify/reject decision
Confidence: Medium
Largest uncertainty: credible counterfactual baseline
Next action: freeze a lightweight non-ROS workflow before implementation
Exit evidence: comparable inputs/outcomes, time accounting, and known confounders
Review trigger: scope freeze

RQ-COMM-006 — Feasibility-test the procedural experiment

Parent: RQ-COMM-001
Children: RQ-COMM-007
Maturity: Idea
Priority: P1; score 10/15
Supporting evidence: HY-COMM-2026-0001 and 0003
Contradicting evidence: none
Dependencies: RQ-COMM-001, 003, 004
Related research: MS-COMM-2026-0001
Produced artifacts: planned EX-COMM-2026-0001
Engineering enabled: frozen experimental harness
Confidence: Low before task selection
Largest uncertainty: scoring reliability, effect sensitivity, and participant burden
Next action: remain blocked from activation until dependencies exit
Exit evidence: executable protocol, usable scoring, acceptable burden, no floor/ceiling effect
Review trigger: task selection

RQ-COMM-007 — Estimate effects of controlled communication variants

Parent: RQ-COMM-006
Children: RQ-COMM-008, RQ-COMM-010
Maturity: Idea
Priority: P2; score 8/15
Supporting evidence: none generated by this repository
Contradicting evidence: none
Dependencies: successful RQ-COMM-006 and analysis plan
Related research: HY-COMM-2026-0002
Produced artifacts: planned RP-COMM-2026-0003
Engineering enabled: evidence-based document transformation
Confidence: Low
Largest uncertainty: causal separation of wording, structure, and visuals
Next action: design only after feasibility
Exit evidence: uncertainty intervals, failure analysis, subgroup limits, replication decision
Review trigger: feasibility exit

RQ-COMM-008 — Build a functional communication taxonomy

Parent: RQ-COMM-002, RQ-COMM-007
Children: RQ-COMM-009
Maturity: Idea
Priority: P2; score 7/15
Supporting evidence: TH-COMM-2026-0001 predicts function/context matter
Contradicting evidence: no exhaustiveness evidence
Dependencies: stable constructs and at least one empirical cycle
Related research: genres, rhetoric, discourse, information design
Produced artifacts: none
Engineering enabled: strategy retrieval and comparison
Confidence: Low
Largest uncertainty: classification dimensions and boundary criteria
Next action: collect candidate dimensions without claiming completeness
Exit evidence: explicit inclusion rules, multi-label structure, inter-rater test, counterexamples
Review trigger: completion of RQ-COMM-007

RQ-COMM-009 — Design an AI strategy-selection model

Parent: RQ-COMM-008, RQ-COMM-010
Children: none
Maturity: Idea
Priority: P3; score 5/15
Supporting evidence: none
Contradicting evidence: missing validated inputs/outcomes
Dependencies: RQ-COMM-004, 008, 010
Related research: AI instruction following, uncertainty calibration
Produced artifacts: none
Engineering enabled: adaptive communication system
Confidence: Very Low
Largest uncertainty: safe inference under missing audience/context data
Next action: define misuse cases only; do not implement optimizer
Exit evidence: validated input schema, abstention/fallback rules, offline evaluation, harm analysis
Review trigger: supported taxonomy and cross-context findings

RQ-COMM-010 — Test generalization across audiences and contexts

Parent: RQ-COMM-007
Children: RQ-COMM-009
Maturity: Idea
Priority: P2; score 8/15
Supporting evidence: TH-COMM-2026-0001 predicts moderation
Contradicting evidence: none
Dependencies: replicated within-context effect
Related research: cross-cultural pragmatics, expertise reversal, accessibility
Produced artifacts: none
Engineering enabled: applicability boundaries and adaptive rules
Confidence: Low
Largest uncertainty: which moderators are decision-relevant
Next action: preregister replication contexts after first effect
Exit evidence: heterogeneous-effects model and failed-generalization cases
Review trigger: replicated RQ-COMM-007 effect

RQ-COMM-011 — Explain documentation longevity and decay

Parent: program objective
Children: none
Maturity: Idea
Priority: P2; score 7/15
Supporting evidence: none
Contradicting evidence: none
Dependencies: corpus access and longevity definition
Related research: maintenance, information architecture, software evolution
Produced artifacts: none
Engineering enabled: durable documentation architecture and decay monitors
Confidence: Very Low
Largest uncertainty: survivorship bias and measurable usefulness over time
Next action: scope longitudinal corpus and falsifiable longevity signals
Exit evidence: sampling method, outcome definition, confounder model
Review trigger: first pilot completes or a suitable corpus becomes available

RQ-COMM-012 — Define anti-manipulation boundaries

Parent: RQ-COMM-004
Children: RQ-COMM-009
Maturity: Idea
Priority: P1; score 10/15
Supporting evidence: constitutional safety and ethical constraints
Contradicting evidence: none
Dependencies: representative persuasion/decision use cases
Related research: dark patterns, informed consent, autonomy, rhetoric
Produced artifacts: none
Engineering enabled: optimization constraints and refusal rules
Confidence: Low
Largest uncertainty: operational distinction between ethical persuasion and manipulation across power asymmetries
Next action: comparative normative and empirical scoping before any persuasion engineering
Exit evidence: abuse cases, red lines, review thresholds, contested cases
Review trigger: any persuasion or behavior-change project

Stage gates

Transition	Minimum evidence
Idea → Scoping	decision relevance, owner candidate, explicit uncertainty
Scoping → Background Research	bounded question, inclusion criteria, evidence plan
Background Research → Deep Investigation	credible hypotheses and discriminating method
Deep Investigation → Validation	reusable result plus explicit falsification target
Validation → Synthesis	replication/triangulation and bounded applicability
Synthesis → Engineering Ready	accepted theory/decision, requirements, constraints, acceptance tests
Engineering Ready → Monitoring	implementation validated and telemetry/review triggers defined

Items may move backward when assumptions fail, sources prove dependent, measures are invalid, or boundary conditions narrow materially.

Portfolio review cadence

Review after every REP, failed experiment, material contradiction, or engineering decision.
Review dormant P0/P1 items every 30 days while the portfolio is active.
Review P2/P3 items quarterly or when dependencies change.
Record deletions only as supersession or archival; never erase prior rationale.

Current portfolio decision

Continue MS-COMM-2026-0002 at Deep Investigation. Its desk-research phase selected an outcome vector, but task-specific validity remains blocked on RQ-COMM-001 candidate tasks. The highest-value next input is evidence from a named audience and comparison of three candidate tasks. Do not activate the participant experiment until RQ-COMM-001, 003, and 004 meet their exit evidence.