plumb

Generate alternative representations, APIs, and algorithms by changing the way a problem is described. Use for design exploration at code scale, not a compulsory code-quality checklist.

Starting point

A representation can make a hard operation disappear. Before improving an algorithm, play with what the program considers a thing, a relation, or an event.

Bring a question, fragment, reference, or tension. Follow whichever association becomes useful; these invitations are not a sequence to complete. Speculation should remain distinguishable from an observed fact.

Thought experiments

Transpose noun and verb

Describe the same feature as objects with methods, then as a stream of events, then as a transformation of values. Which awkward case becomes ordinary in each description?

Invent a tiny world

Replace production with three objects and one surprising event. Let a concrete counterexample generate a new representation rather than a list of abstract rules.

Borrow a mathematical shape

Try a graph, interval, algebra, or state machine as a metaphor. Ask what operation becomes compositional. Keep the analogy only as far as it illuminates the actual problem.

A possible turn

A calendar conflict checker looks like nested comparisons between bookings. Redrawing bookings as occupied intervals suggests a sweep; redrawing them as constraints suggests a solver; redrawing them as requests suggests negotiation. The useful output may be this fork itself.

Leave the space open

Offer a few genuinely different possibilities, a sketch, an analogy, or a sharper question when useful. Do not turn every prompt into a scored diagnostic, a completion gate, or a demand for an action plan. The human may choose a direction, combine them, or leave the question open.

Once an interface or representation is selected, assay can retain the tiny counterexample as a regression test.

Optional references

Read selectively when a thread needs another perspective or a concrete technique. The notes supply possibilities, examples, and limits; none is a prerequisite or a requirement to close the exploration.

• Share a relation rather than merely lines
• Representations that change the problem
• Clarity through a concrete boundary
• Choose a small representation experiment
• Expose transformations, decisions, and effects
• Use words to distinguish concepts
• Follow an assumption across representations
• Let a small observable example inform the design