Type: Node / Advisory Proposal
Placement note: Recommendation belongs in Division II because recommending is an attributable occurrence within a Process that proposes what should happen next without resolving the matter. Its practical content connects it to Claim and Interpretation, but its defining function is advisory action directed toward a possible Decision.
Working Definition: A recommendation is an attributable advisory proposal, addressed to a person or constituted forum that retains decisional authority, identifying a course, disposition, or response for a stated matter and scope. An accountable Recommendation also discloses its basis, conditions, and time, but it neither selects the proposed course as governing nor produces institutional effect by itself.
Function in the Grammar: Recommendation makes advice about action traceable without confusing influence with authority. It identifies who proposed what, to whom, concerning which matter, on what basis, within what scope, and under what conditions. A later Decision may adopt, modify, reject, defer, or ignore the proposal. Preserving those relations keeps the recommender's contribution visible while leaving responsibility for resolution with the actual decider.
The act of recommending, the Recommendation produced, its content, and its Record are distinct. RECOMMENDS(a, q, to: b) names the Action or Act through which the proposal is issued. Recommendation(r) names the attributable advisory proposal available for later consideration. RecommendationContent(q) names the proposed course. A report, message, or form may Record the Recommendation without being the recommending act or proving that the addressee considered it.
A Recommendation may result from Review, Interpretation, Evidence analysis, consultation, or ordinary practical reasoning. None is required in every case. Review can conclude without proposing a course, and a person can recommend a course without re-examining an existing object. Recommendation therefore composes with Review without being its subtype or necessary output.
Recommendations can have substantial practical influence even while remaining formally advisory. Expertise, control of information, organizational dependence, or technical automation may make refusal difficult. That influence is a question of Power and institutional design. It does not silently convert Recommendation into Decision or transfer the decider's accountability to the recommender.
Formal Pattern
Recommendation(
r,
recommender: Agent(a),
addressee: Agent(b) | Forum(f),
matter: Matter(m),
content: RecommendationContent(q),
basis: G,
conditions: C,
scope: S,
time: t
)
RECOMMENDS(
Agent(a),
RecommendationContent(q),
to: Agent(b) | Forum(f),
for: Matter(m),
basis: G,
within: Scope(S),
at: t
)
ISSUES(Agent(a), Recommendation(r), in: Process(p))
CONSIDERS(Agent(b), Recommendation(r), in: Process(p),
for: Forum(f), through: Action(alpha))
ADOPTS(
Agent(b),
Recommendation(r),
through: Decision(d),
as: DecisionContent(q2),
for: Forum(f)
)
DECLINES(Agent(b), Recommendation(r), through: Decision(d), for: Forum(f))
DEFERS(Agent(b), Recommendation(r), until: Condition(c), for: Forum(f))
q2 may equal the recommended content, narrow it, combine it with another proposal, or substitute a different course. Adoption is therefore a disclosed relation between a Recommendation and a Decision, not identity between them. CONSIDERS, ADOPTS, DECLINES, and DEFERS require attributable acts where an institution gives them formal significance.
A Forum may receive a Recommendation or serve as the setting in which it is considered, but it does not literally deliberate or decide. The Agents, Roles, and aggregation Rule through which consideration or adoption occurs must remain identifiable. The for: Forum(f) argument is omitted when the addressee acts only in a personal capacity.
If a governing Rule makes an output automatically controlling, that output performs more than Recommendation. The additional Decision, constitutive trigger, or other institutional Act must be represented explicitly so that binding force is not hidden inside advisory vocabulary.
Core Relations
| Relation | Notes |
|---|---|
| Agent and Role locate | A Recommendation requires an attributable recommender. A Role may grant standing, access, expertise, or authority to recommend without granting authority to decide. |
| Process and Matter situate | A Recommendation occurs within a Process and concerns a stated Matter. It may redirect attention or supply an option without resolving either. |
| Claim and Interpretation may support | A Recommendation may include claims about consequences and rely on interpretations of circumstances. Those claims and interpretations remain separately contestable. |
| Evidence and Standard may ground | Evidence and Standards may support the proposed course. Their use does not make the Recommendation correct or binding. |
| Review may produce | A governed re-examination may produce one or more Recommendations, but it may instead produce findings, questions, or an assessment without proposing action. |
| Decision may adopt | A Decision may adopt, modify, reject, defer, or ignore a Recommendation. The Decision, not the Recommendation, selects what will govern next. |
| Authority separates | Authority to recommend and authority to decide are distinct grants. An expert adviser may have the first while an editor, court, board, client, or electorate retains the second. |
| Power distorts | A formally advisory proposal may exert practical pressure through expertise, dependency, control of options, or automation. Practical influence does not erase the formal distinction. |
| Record preserves | A Record may preserve the Recommendation, its basis, and its later disposition. It is not the recommending Action or the Decision that may follow. |
Typical Questions
- Who issued the Recommendation, in what Role, and to whom?
- What course, disposition, or response is proposed?
- What Matter, scope, conditions, and time does it address?
- What Claims, Interpretations, Evidence, Standards, or Values support it?
- Does the recommender have disclosed interests, dependencies, or conflicts?
- Who retains authority and accountability for the Decision?
- Was the Recommendation considered, adopted, modified, rejected, deferred, or ignored?
- If adopted, which Decision performed the adoption and what content did it select?
- Did a formally advisory proposal become practically coercive?
- Is a Record of the Recommendation being mistaken for proof of consideration or adoption?
Examples
| Domain | Recommendation | Recommender and addressee | What remains undecided |
|---|---|---|---|
| Ordinary life | Take the earlier train because severe weather is expected | One traveler advising another | Whether the addressee changes the trip |
| Law | Impose a particular sentence within the authorized range | Probation officer or party addressing the court | The court's sentencing Decision |
| Science | Repeat an experiment with a larger sample before drawing the stated conclusion | Peer reviewer addressing researchers or editor | Revision, acceptance, and the truth of the hypothesis |
| Publishing | Return an encyclopedia entry for two conceptual revisions | Reviewer addressing the human author | Whether to revise, publish, defer, or reject |
| Markets | Approve an acquisition subject to specified risk controls | Advisory committee addressing the board | Whether the board authorizes the transaction |
| Community governance | Adopt a traffic-calming design after public consultation | Working group addressing a municipal forum | Which design, if any, the authorized forum adopts |
| Digital systems | Reject a proposed automated transition because confidence is below the governing threshold | Model or rules engine addressing a human reviewer | Whether the transition is rejected, retried, escalated, or approved |
Distinctions
Recommendation ≠ Decision. A Recommendation proposes what another Agent or constituted Forum should select. A Decision resolves among live possibilities and makes a course governing within scope. An editor may follow a reviewer's recommendation to reject a paper, but the reviewer did not thereby make the editorial Decision.
Recommendation ≠ Review. Review is governed examination of an existing object for a stated purpose against disclosed grounds or criteria. Recommendation is an advisory proposal about what should be done. A safety Review may identify unresolved risks without recommending a course, while a friend may recommend a restaurant without reviewing anything.
Recommendation ≠ Claim. A Claim puts forward a proposition that can be supported, contested, accepted, or rejected. A Recommendation proposes a practical course to a decider. "Option A is cheaper" is a Claim; "choose Option A" is a Recommendation. The proposal may rely on the Claim without being identical to it.
Recommendation ≠ Interpretation. Interpretation states what an input is taken to mean. Recommendation states what course is proposed in response. Interpreting a draft as conceptually incomplete does not itself recommend revision, rejection, or deferral.
Recommendation ≠ Instruction or order. An instruction or order purports to direct action under some authority or relationship. A Recommendation preserves the addressee's authority to decide. Calling an order a Recommendation can hide coercion; calling a Recommendation an order can falsely attribute authority to the recommender.
Recommendation ≠ Record. A Recommendation is the advisory proposal produced through an attributable Action or Act. A Record represents that proposal. A report may accurately record a Recommendation that was never considered, or falsely state that a Recommendation was issued.
Common Failure Modes
| Mode | Description |
|---|---|
| Recommendation laundering | An advisory proposal is treated as the operative Decision, allowing downstream action without an attributable decider. Where to look: automated or expert outputs that directly change status while the record names no adoption Decision. |
| Orphaned recommendation | The proposal lacks an identifiable addressee, Matter, scope, expiry condition, or decision owner. Where to look: reports that say "it is recommended" without identifying who can act, what exactly is proposed, or when the advice ceases to apply. |
| Basis or conflict concealment | The Recommendation withholds its Evidence, Standard, assumptions, incentives, or conflicts. Where to look: proposals presented as expert conclusions whose supporting analysis or interested relationships cannot be inspected. |
| Advisory capture | The recommender controls the evidence, available options, implementation path, or practical consequences of refusal so completely that the authorized decider becomes nominal. Where to look: repeated rubber-stamping, menus containing only one feasible option, or refusal followed by undisclosed penalties. |
Minimum Viable Test Case
r1 = Recommendation(
recommender: model_assistant,
addressee: human_author,
matter: editorial treatment of entry_x,
content: return_for_revision,
basis: rubric findings f1 and f2,
conditions: reassess after revision,
scope: entry_x version 3,
time: t1
)
RECOMMENDS(
model_assistant,
return_for_revision,
to: human_author,
for: entry_x,
basis: {f1, f2},
within: editorial_review,
at: t1
)
d1 = Decision(
decider: human_author,
matter: editorial treatment of entry_x,
content: return_for_revision,
time: t2
)
ADOPTS(human_author, r1, through: d1, as: return_for_revision)
Result:
r1 made a proposed course attributable and reviewable.
d1 selected the course that governs what happens next.
The shared content does not make r1 and d1 the same act.
Failure case:
If an automated workflow changes entry_x to revision_required at t1
without a governing Rule that constitutes a separate automated Decision,
r1 has been laundered into a Decision.
Cross-References
Existing
- Agent: the attributable recommender or decider
- Role: may confer authority to recommend without authority to decide
- Claim: may state propositions supporting the proposed course
- Interpretation: may characterize the situation to which the Recommendation responds
- Evidence: may support or undermine the Recommendation's basis
- Standard: may supply disclosed criteria for the Recommendation
- Decision: may adopt, modify, reject, defer, or ignore the proposal
- Authority: distinguishes advisory authority from decisional authority
- Power: explains how formally advisory proposals may become practically coercive
- Process: situates issuance, consideration, and disposition through time
- Record: preserves the Recommendation and any later disposition without becoming either act
- Review: may produce a Recommendation through governed examination, but remains independently defined