PSD.8 - Generate Decision Alternatives

Type: DPF pattern body Status: Candidate

Primary working result: a finite candidate set for one decision-support question, with materially different alternatives, their supported and unresolved premises, and the exact candidate gaps that still matter.

PSD.8:1 - Problem frame

Use this when an inquiry has clarified what matters but the proposed decision still contains only the incumbent, a sponsor’s preferred project, or two labels such as “buy or build”. Also use it when a request for development advice names a course, provider, reorganization, or AI tool before the possible directions have been constructed.

Start by asking what different action could address each live formulation or concern. Change the mechanism, arrangement, combination, timing, or commitment being considered, then state enough of each resulting alternative for a fair consequence comparison. The practical gain is a choice worth examining, including an honest gap when no supported alternative is available.

The governed object is the candidate set, not a selected action. A candidate describes a possible intervention or commitment; its description does not establish feasibility, permission, performed Work, or benefit. Alternatives may address competing formulations rather than obtain exactly the same result. Preserve that difference instead of inventing parity.

Do not use this pattern when an adequate finite set already exists and only comparison or an authorized choice remains. Use the direct domain practice when it already owns the complete alternative-generation question. A population or lineage without a population-local recipient or chooser is not a recipient of development advice; an authorized researcher’s intervention is a different decision-support case.

PSD.8:2 - Problem

An apparently open choice can be closed before analysis begins. A list of pump models omits deployment arrangements; a list of training courses omits changing the Work that prevents capability expression; a list of AI models omits the tool, data, or operating configuration responsible for the failure.

Unrestricted ideation creates the opposite difficulty. Hundreds of names can repeat the same intervention, depend on unsupported capacities, or combine benefits that cannot coexist. Novelty does not make an alternative useful, and an attractive candidate does not become admissible because it survived a workshop.

PSD.8:3 - Forces

Force	Tension
Breadth and relevance	A different mechanism may expose a better direction, while an unbounded search consumes the decision horizon.
Plural formulations	Alternatives should address material interpretations without pretending that every interpretation seeks the same result.
Imagination and qualification	A proposal may be worth investigating before feasibility is known, while unsupported proposals cannot be presented as ready commitments.
Combination and interaction	Hybrid or staged directions can preserve options, while shared resources, incompatibilities, and delay can remove their apparent advantage.
Generation and choice	Early preference helps identify concerns but can suppress candidates before their consequences are known.

PSD.8:4 - Solution

Construct alternatives from the live decision question and useful differences in the inquiry. Make each candidate complete enough for its next use, separate qualification gaps from rejected directions, and stop with a bounded set or a precise missing-alternative result.

PSD.8:4.1 - Recover the receiving question and useful inputs

Name the recipient, holder or subject, horizon, current arrangement, later choice owner, and the commitment being considered. Recover the protected conditions and resource bounds that can change which alternatives are live. If the request still has no recoverable question or authority boundary, return that missing premise rather than inventing a development programme.

Use a PSD.3 formulation set only where its differences suggest different interventions. A formulation is not itself an action. Use the PSD.5 model account only where a model discriminates candidate mechanisms or exposes an omitted consequence. Use PSD.7 inquiry results where a shared or contested claim changes the alternatives; agreement is not required.

These are conditional contributions. Use a qualified direct result when it supplies the same needed premise. If the needed input is absent, stale, or incompatible with this subject and horizon, retain the precise gap. Neither adjacent PatternIDs nor a completed workshop supplies an alternative by itself.

PSD.8:4.2 - Vary the decision-changing contents

Ask what the proposal is meant to achieve, why that result matters, and what other way could address the concern. Generate some ideas independently before group discussion when early advocacy could anchor the whole inquiry. Then use differences and objections to construct additional candidates.

Useful prompts include changing the intervention mechanism, obtaining arrangement, scale, timing, resource allocation, or commitment. Consider combining compatible actions, retaining the present arrangement, declining a proposed change, and obtaining information before a larger commitment. These are prompts, not a mandatory taxonomy or a required number of options.

Describe “no new action” against a real baseline. It may still include already authorized maintenance, contractual duties, continuing costs, deterioration, or exposure. It is not a cost-free, consequence-free empty world.

For a staged direction, say what happens first, what later choice remains open, which observation could inform it, and what delay or irreversibility the first commitment introduces. Do not call a probe reversible merely because it is small; obtain the direct feasibility, consent, safety, and recovery premises when they matter.

PSD.8:4.3 - Turn labels into sufficiently complete candidates

For each material alternative, state the proposed change from the current arrangement, affected subjects, intended result, enabling conditions, major burdens, evidence basis, and unresolved premise. Distinguish what is supported now from what is proposed or unknown.

Use C.38 only when several whole ways are intended to make the same receiving result available and labels still hide their different burdens. If the alternatives seek different results, retain the difference and let the decision-support comparison expose it. Do not force a same-result declaration merely to obtain a neat table.

A combination is another candidate when its interactions can change consequences. Check shared staff, facilities, money, dependencies, sequencing, and incompatible assumptions. Two independently attractive components do not establish an attractive combined arrangement. A provider’s advertised capability, a course title, or a benchmark score is not a qualified result for this holder.

PSD.8:4.4 - Distinguish exploration from a live commitment option

Keep a promising but unresolved proposal available for inquiry without presenting it as an admissible commitment. A short annotation is enough: “comparison-ready within these conditions”, “requires this qualification”, or “excluded for this stated reason”. These phrases describe the present use; they create no new universal status system.

Inspect an adequate existing specialist result before requesting more. Preserve useful directions and the exact limit on a stronger claim. A decision-changing cross-practice gap becomes a feasibility, allocation, transfer, security, or other request only when A.15.9 selects its obtainable contribution against the whole acquisition burden through C.11.DUA. A missing result may block one branch while others remain useful.

Use C.17 when a claim about novelty, diversity, or usefulness of the named candidates needs characterization. Use C.18 when open-ended generation, archive, or front stewardship is actually needed. Neither characterization nor archive membership chooses the decision alternative. Use C.19 only for an actual live-pool policy question; thin evidence alone supplies no default exploration policy.

PSD.8:4.5 - Challenge coverage and stop proportionately

Ask whether the set still reflects only one favored formulation, one available technology, or one powerful participant’s concern. Check whether a materially different mechanism, genuine no-change case, compatible combination, or bounded probe was omitted without reason. Merge labels that have the same decision-changing contents; retain differences that alter consequences or supported use.

Stop when the set is sufficient for the present comparison: material formulation differences have an addressed candidate or an explicit gap, each retained row is intelligible, protected-condition exclusions are visible, and additional search has no identified benefit worth the remaining effort. This is bounded sufficiency, not proof that all possible alternatives have been found.

If only one lawful direction survives, report that fact and its basis without manufacturing a rival. If none survives, return the missing qualification, infeasible requirement, or need to reopen the problem. A gap is a useful result; a decorative list is not.

PSD.8:4.6 - Return the set without making the choice

Return the question, subject and horizon; the exact candidate contents; their material differences; supported conditions and remaining gaps; excluded branches and reasons; and the observation that would add, remove, or change a candidate. Use the smallest readable form that lets another practitioner reconstruct those distinctions.

PSD.11 may use this set as the alternatives for its named consequence comparison. The comparison must still establish its own value, evidence, and scope basis. Use the set as an input to any further ranking or recommendation. The authorized chooser makes the choice. Any later programme or WorkPlan needs its own basis; support claims about performed work and achieved effects with evidence of those occurrences.

PSD.8:5 - Archetypal Grounding

PSD.8:5.1 - Flood-pump alternatives beyond a purchase contest

In this illustrative case, the board needs a pre-season return for the next flood season. The inquiry distinguishes nominal capacity, deployment reliability, unequal service, and downstream consequences. The model account says that nominal permanent and mobile capacity can meet the assumed inflow when the stated units are operating, but mobile arrival under road loss is not yet supported. Reachable assistance remains a separate service concern.

The inquiry turns those premises into three candidate descriptions and one unfinished proposal for staging:

Candidate	Different action being considered	Present use and gap
N — continue the present arrangement	Keep already authorized operation and maintenance; make no new capacity commitment.	A real baseline for comparison, not an assumption of adequate service or zero cost.
F — fixed-capacity change	Add fixed capacity with the necessary operating, staffing, and support arrangement.	A proposed whole change; direct engineering and service results must qualify its consequences.
M — mobile-reserve arrangement	Obtain mobile capacity together with its transport, deployment, staffing, and access arrangement.	Retained for inquiry; the road-loss deployment premise prevents an unconditional commitment claim.
S — unfinished staged proposal	The proposal would preserve a later capacity choice informed by an access or service inquiry, but its first intervention has not been specified.	Keep the construction question open: what should change first, and how would that permit the later choice? Specify that intervention and qualify its temporary provision, resource overlap, delay and recovery before comparing S as a distinct direction. Adding the benefits of F and M does not answer these questions.

The property-protection and reachable-assistance concerns remain distinct. They lead the team to ask whether each whole arrangement serves both; they do not establish an approved priority between them. No candidate is dropped because its proponent dissents from the majority.

Return N, F and M with their stated limits, and retain S as an unfinished proposal with a named construction question. The supported comparisons remain usable at their stated scope. If commitment-ready options are needed, identify the missing premise for M and the unfinished construction and qualification of S. Keep those gaps visible without claiming that the usable set exhausts all possibilities.

PSD.8:5.2 - A ninety-day development-direction question

An advisory team is asked which direction a service organization should consider for a ninety-day reliability horizon. “Train, outsource, or use AI” is not yet a set of comparable directions. The adviser resolves the labels into internal development with covered service duties, a qualified provider arrangement, and a mixed human–tool arrangement with specified oversight and fallback. A bounded mixed-arrangement probe is a separate candidate when its information and operating consequences differ from deployment.

Each branch needs its own supplier result. Human capability evidence does not establish transfer to the representative later Work; provider evidence does not establish interface or security fit; model evaluation does not establish reliability of the mixed arrangement. If the organization-allocation comparison is missing, return that exact gap rather than ranking the three labels.

For advice about one person, the same move generates directions only from qualified human and target-Work premises. For an AI arrangement, it distinguishes parameter, data, scaffold, tool, code, and environment changes only where the case makes those objects live. Neither application imports the other’s mechanism or evidence.

PSD.8:5.3 - Cheap non-use

A recipient already has two complete, qualified ways of obtaining the same repair result and asks which to choose. Use the existing set and the applicable comparison or C.11 choice guidance. Another brainstorming exercise adds no candidate-changing result.

PSD.8:6 - Bias-Annotation

Scope: alternatives for bounded decision-support engagements. Lenses: Prag tests useful differences; Onto/Epist separates proposed arrangements from facts; Arch exposes combinations and dependencies; Gov preserves protected conditions and later authority; Did replaces label lists with intelligible candidates.

Incumbency, sponsor anchoring, technology availability, and novelty bias can each shrink or distort the set. Challenge the omitted mechanism and baseline, not merely the number of ideas. Counter the opposite bias toward endless variety by asking what additional candidate could actually change the present comparison.

PSD.8:7 - Conformance Checklist

The recipient question, holder or subject, current arrangement, horizon, and separate choice boundary are recoverable.
Formulations, models, and inquiry claims contribute only where they change the candidates.
Alternatives differ in decision-changing contents, not merely names.
The no-change case includes its actual continuing obligations and consequences.
Combinations and staged directions expose interactions, delay, and unresolved enabling conditions.
Supported, proposed, and unknown premises remain distinguishable.
Qualification gaps and excluded branches have exact reasons; missing alternatives are not fabricated.
The stopping point states bounded sufficiency, not exhaustive search.
The returned set permits a later comparison without asserting selection, authority, Work, or benefit.

PSD.8:8 - Common Anti-Patterns and How to Avoid Them

Anti-pattern	Repair
Score the sponsor’s option against “do nothing”.	Recover a real baseline and a materially different mechanism or explain why none is live.
Treat buy, provider, AI, and internal as peer kinds.	Describe the proposed whole arrangements and their receiving results.
Exclude a proposal because feasibility is unknown.	Keep the inquiry candidate separate from commitment eligibility and name the missing result.
Add component benefits to justify a hybrid.	Qualify the combined configuration and its interactions.
Treat novelty or archive membership as selection.	Use the characterization or archive result only for its stated function.
Generate forever because the set is not exhaustive.	Stop at the bounded comparison need and retain a specific reopen condition.

PSD.8:9 - Consequences

The recipient gets a genuinely wider but finite decision question, with candidate gaps visible before scoring. The cost is some additional construction and qualification effort. Concentrate it on omitted differences that could change the decision; do not describe every possible arrangement in equal detail.

PSD.8:10 - Rationale

Structured inquiry becomes useful to choice when its differences change what can be done. Alternative construction preserves those differences while qualification prevents imagination from becoming an unsupported commitment. Keeping generation separate from comparison permits a promising proposal to remain investigable without calling it a winner.

PSD.8:11 - SoTA-Echoing

Practice question	Best-known line	Serious alternative or default	Defect overcome and pattern mutation	Source roles and limits	Reopen condition
How can an engagement escape the incumbent without producing a decorative idea list?	Use objectives and materially different mechanisms to guide alternative construction.	Unstructured group brainstorming is cheap; a supplied option menu is cheaper still.	Adapt: :4.1–:4.3 add concern-led prompts and independent generation where anchoring matters. The deliberate cost is a short construction pass before comparison; an adequate existing set takes the cheap exit.	Borgonovo et al.’s 2026 decision-analysis review, especially its value-focused-thinking discussion, is a current synthesis candidate. Keeney’s 2012 value-focused brainstorming supplies the specific objective-led and individual-before-group procedure. Its policy example is not universal effectiveness evidence.	Reopen if a simpler elicitation gives the same material breadth or the objectives themselves suppress a live formulation.
When is a staged alternative genuinely different?	Preserve options only through explicit dependencies, observations, and feasible later changes.	Compare only fixed end-state projects, or label an underspecified pilot “adaptive”.	Adapt: :4.2–:4.4 require a specified and qualified first commitment, later question, and delay or lock-in consequence. The unfinished S example shows why a staging label is insufficient. Additional description is justified only when timing changes the comparison.	The 2019 DAPP chapter is the best-known-line candidate for pathway dependencies and failure conditions; its water-policy applications do not supply PSD’s local thresholds, authority, or universal sequence.	Reopen when monitoring, lead time, or reversibility makes the staged direction infeasible or no longer distinct.
Does a novelty result already supply the alternatives?	Keep characterization, generation stewardship, and local choice separate.	Read a novelty profile or archive as a ready option set.	Adopt: :4.4 uses C.17 and C.18 only for their actual objects; :4.6 retains the PSD candidate result. This avoids reconstructing machinery when ordinary candidate descriptions suffice.	Current `C.17` characterizes bounded novelty or use; `C.18` governs generation, archive, and front stewardship; `C.38` forms same-result ways. Their exact boundaries defeat the broader reading, but none supplies holder feasibility.	Reopen only if a direct contribution or the receiving candidate question changes.

PSD.8:12 - Relations

PSD.3 supplies formulation differences for candidate formation, not selected actions. PSD.5 supplies discriminating model claims, not the obtaining conditions they represent. PSD.7 supplies shared and contested inquiry claims only where they change the set; dissent remains explicit.
PSD.11 may consume the candidate set for the same named comparison. It establishes its own comparison result; generation does not select.
C.17, C.18, C.19, and C.38 supply characterization, generation stewardship, live-pool policy, and same-result construction only under their own use conditions.
A.15.9 supplies bounded use of an outside-practice result and selection of a worthwhile further request. Direct holder and domain practices supply feasibility, mechanisms, protected conditions, and evidence.
C.11 governs the later local choice over an adequate option set. A missing or incompatible input returns to its qualified direct source or remains an exact gap; no neighboring body is a compulsory prerequisite.