Foray 246 / Reusable prompt · version 2 · 24 October 2026
Find the better arrangement nearby
This prompt preserves the original enquiry into professional opportunities hiding in plain sight. It broadens the original “never mentioned before” test into two explicit lanes and makes analysis and representation open to improvement. Supply either an invented scenario or a corpus you have permission to use; choose its disclosure boundary first.
Ends
Find professional opportunities, useful experiments, delivery improvements, better jobs or credible ways to earn that are hiding in plain sight around the subject’s current practice. Identify nearby arrangements that could sustain greater value or reward without merely moving the burden to someone else. Explain why the subject may be walking past them, what makes them reachable, and what sequence of changes could move from here to there.
Investigate “nearby partial equilibria” critically. Define the bounded system, actors, choices, incentives, constraints and time horizon. Decide whether equilibrium is the right concept; an opportunity frontier, coordination trap, option, transition or unstable niche may describe the evidence better. Do not manufacture precision or a formal solution to satisfy the metaphor.
Ways
- Reconstruct the subject’s present position from the supplied evidence: ends, work, capabilities, relationships, access, assets, recurring friction, actual buying routes, constraints and what they have already considered. Distinguish observed facts, interpretations and unknowns. Do not turn an old contact into a current opportunity.
- Search beyond the obvious next step. You may invent or combine methods of investigation, analogy, decomposition, causal analysis, search, modelling, simulation or field enquiry. Choose dimensions and boundaries because they reveal useful distinctions. Explain what the method can and cannot support; compare it with a simple baseline.
- Keep two novelty lanes: A — unmentioned candidates, with a bounded check against the supplied corpus; B — underexploited known ideas or combinations, naming the familiar ingredients and the proposed difference. “Not found in the supplied material” is not “never thought of” or “new in the world”. Include a rejection lane for relabelling, weak access, absent demand and poor fit.
- For promising candidates, identify a beneficiary, buyer or sponsor, a bounded useful result, why existing provision falls short, required permission and participation, reward distribution, transition cost, opportunity cost, failure modes and likely responses by others. Separate reachability from attractiveness, first delivery from repeatability, and an attractive destination from a credible path.
- Actively seek disconfirmation and competing explanations. Ask why the opportunity is unexploited, what makes the present arrangement persist, who might refuse or capture the benefit, and how it changes if demand, access, time or the environment changes. Prefer a small discriminating probe to an unsupported business case. State stop and revision conditions.
- Invent the representation that best serves the reasoning. You may use prose, tables, diagrams, small multiples, interactive projections, pathways, games, simulations, or a better form not anticipated here. Do not force the problem into two dimensions, a weighted score, a dashboard or an executive deck. Show material omitted dimensions, uncertainty and sensitivity; make any numeric assumptions inspectable.
- End with a few bounded next experiments and the evidence they would generate, not a compulsory task list. Preserve alternative futures, unresolved questions and the strongest reason to keep the current arrangement. Explain where a later, more capable model could challenge this run.
Means
Use the authorised corpus, available tools and current primary sources where they materially improve the answer. State what you actually inspected and what you could not. Search is permission to investigate, not permission to disclose private material, contact people, spend money, accept terms or change commitments.
For a public run, use a wholly fictional case and a separate disclosure check. For a private run, preserve evidence links privately and do not publish its personal map. Record the date, model identity if available, evidence boundary, assumptions and any comparisons with previous runs. Retain enough inputs and reasoning to distinguish a better model from richer evidence or a changed brief.
How to compare later runs
Evaluate evidence fidelity, useful novelty, actor realism, plausible access, discriminating tests and clarity of representation. Ask what changed and why. A more elaborate picture or more confident recommendation is not evidence of a better answer. Keep earlier conclusions visible as earlier conclusions; improvement remains to be demonstrated in use.