Purpose

Determine whether the proposed “Communication Engineering” label identifies a distinct and useful engineering profile, or merely renames established bodies of knowledge. This roadmap is a mutable execution view, not a registry record and not evidence by itself.

Completion gate for every body of work

A stream may be marked complete only when it records:

• foundational constructs and theories;
• recent peer-reviewed or standards-based developments;
• at least one empirical or field result where available;
• competing interpretations and counterevidence;
• methods and outcome measures reusable by this repository;
• what the body already explains;
• what it does not explain;
• overlap with the proposed profile;
• at least one assumption or belief challenged;
• a discriminating prediction or experiment;
• confidence, limitations, and exact source provenance;
• ten nontrivial future research questions when the stream is treated as a completed topic.

Shared claims to challenge

• Communication failure is mainly an information-content problem. Rejected: coordination, trust, ownership, and authority can dominate.
• More complete documentation necessarily improves outcomes. Rejected: relevance and managed selectivity matter.
• Communication quality can be evaluated independently of recipient, purpose, authority, timing, and organizational context. Rejected by expertise reversal, interruption moderators, and organizational mediation.
• Tacit knowledge can usually be converted into an artifact. Rejected: transactive memory and knowledge integration remain social.
• Reduced clarification means better communication. Rejected: questions can be productive coordination; silence can be disengagement.
• A technically correct message causes correct action. Rejected: rhetorical framing, authority, closed-loop confirmation, and context mediate action.
• A reusable template works across expertise levels and risk classes. Rejected by expertise reversal; adaptation must itself be tested.
• AI summarization reduces total communication cost. Rejected as a general claim: factuality, verification, omission, and calibration costs apply.
• Integration of existing methods constitutes a new discipline. Rejected: incremental predictive or practical value must be demonstrated.
• The word “engineering” is justified without predictive or verification capability. Rejected: systems and decision science already provide the relevant burden of proof.

Execution checklist

Synthesis gate

After all streams:

• Build a claim-by-field coverage matrix. See RP-BOUNDARY-2026-5A7C.
• Identify any exclusive object, mechanism, method, or prediction. None found.
• Compare an integrated profile against the strongest adjacent method.
• Quantify added value and added process cost.
• Test whether the label improves retrieval and decisions or creates terminological confusion.
• Accept, narrow, rename, or reject the proposed profile through a Decision Record. DF-BOUNDARY-2026-3F72 rejects the discipline claim and leaves only an experimental profile.

The three unchecked items are empirical follow-ups, not incomplete literature streams. They require a preregistered experiment or terminology study and must not be checked from synthesis alone.

Stop and continuation rules

Stop a stream when its completion gate is met or when access limitations are recorded and the remaining evidence cannot change the current decision. Never mark a literature search exhaustive unless a reproducible systematic protocol was actually run. Every incomplete stream must state the next source class, query, or experiment needed.