Desire path
Description
A desire-path is the route people actually take, worn into the environment by repeated use, standing in visible contrast to the route the design prescribed. The dirt shortcut across a lawn is the namesake: dozens of pedestrians, each independently choosing the straightest line to a goal, aggregate into a trail the official walkways failed to provide. The concept’s load-bearing claim is that the wear is a diagnostic signal — a free usability study written in dirt. It shows precisely where the designed path and the desired path diverge, and it does so without anyone being asked. The same shape recurs wherever a system prescribes a route and its users vote with their feet. In software, users lean on whatever behavior is convenient — an incidental output ordering, an undocumented endpoint, a config side effect — wearing paths through the interface the designers never sanctioned. In organizations, the real process (who actually gets called to unstick an order) diverges from the documented one. In all of these the trace is the aggregate of many independent local choices, and reading it tells you something the prescribed design cannot: what people, in fact, prefer to do. The prescription that famously attaches to desire-paths — “pave the cowpaths,” follow the natural line of movement (Christopher Alexander’s A Pattern Language, Pattern 120 “Paths and Goals,” lays paths along how people actually navigate between goals) — is not automatic. The same trace warrants opposite responses depending on why it exists. A shortcut that reveals a genuinely better line should be accommodated; an entrenched habit that persists only through inertia may be worth obliterating, not paving (Michael Hammer’s reengineering warning against “paving the cow paths” is the business-process inversion of the urban-design advice). The trace states the mismatch; diagnosing why it is worn is what decides the response.Aliases
The route-worn-by-use goes by many names across domains: desire line (the common British term), cowpath or cow path, goat track, elephant path (Dutch olifantenpad), social trail and game trail in ecology. The “desire line / desire path” label is popularly traced to Gaston Bachelard’s The Poetics of Space (1958), but that attribution is disputed — scholars have failed to locate the phrase in Bachelard, and the term may instead originate in early-twentieth-century civil-engineering / transit-planning usage. The catalog treats the etymology as unsettled and does not lean on the Bachelard attribution.Triggers
User-initiated: User contrasts how something is used with how it was meant to be used, or points at a worn-in shortcut, workaround, or shadow process. Vocabulary cues: “desire path,” “cowpath,” “where people actually walk,” “real vs intended flow,” “the workaround everyone uses,” “how it’s actually done around here.” Agent-initiated: Engine notices a persistent divergence between a prescribed route and observed behavior, and treats the divergence as data. Candidate inference: “there’s a desire-path here — the observed usage is telling you where the design’s model of use is wrong; the question is whether to accommodate it or block it.” Situation-shape signals: Unofficial shortcuts worn into a physical space; users depending on unintended-but-convenient system behaviors; documented process vs actual process gaps; feature-usage analytics that reveal flows the designers never intended; repeated support tickets that all route around the same sanctioned path.Exclusions
- One-off deviation with no persistent trace — a desire-path is aggregated, repeated use worn into a durable signal. A single person cutting across once reveals individual whim, not a stable revealed preference; without persistence there is no diagnostic signal to read.
- Deliberately-designed “organic” routes — a path a landscape architect intentionally curves to look natural is not a desire-path. The concept requires the route to emerge against the prescribed design, not to be designed to mimic emergence. Crediting a planned path as a desire-path mistakes intention for emergence.
- Workarounds around a load-bearing control — when users route around a gate that exists for a real reason (a safety interlock, a security check, a required audit step), the worn path is a policy-violation signal, not a “pave the cowpath” invitation. The concept’s prescription assumes the deviation reveals a legitimate unmet need; where the control is load-bearing, accommodating the shortcut inverts its purpose.
- Equivalent alternatives / preference ties — where the actual and prescribed routes are equally good and the deviation carries no information about an unmet need, the trace is noise, not signal; there is no design mismatch to diagnose.
Structure
Relationships
- emergence — a desire-path is a specialization of emergence: emergent collective structure (a pattern in the aggregate, not in any single crossing) narrowed to the case that deviates from a prescribed design and thereby reveals a mismatch.
- affordance — affordance is the offered possibility; a desire-path is the worn evidence that an afforded-but-unprescribed possibility was actually taken. The two are the offer and the revealed-preference trace of the offer.
- schema-anomaly — a desire-path is a schema-anomaly in physical (or process, or dependency) space: the route that shouldn’t exist per the plan is the noticed, load-bearing deviation. Both concepts are where-to-look heuristics — attend to the anomaly; it is telling you where your model is wrong.
Examples
"Desire paths: the unofficial footpaths that frustrate, captivate campus planners," University of Wisconsin–Madison News (April 24, 2019). · architecture-and-design
"Desire paths: the unofficial footpaths that frustrate, captivate campus planners," University of Wisconsin–Madison News (April 24, 2019). · architecture-and-design
The dirt shortcut worn diagonally across a campus lawn is the canonical desire-path: many pedestrians, each independently choosing the straightest line to a goal, aggregate into a visible trail the official walkways failed to provide. Campus planners read these traces as data. The proactive response is to locate “pedestrian generators” — parking, bike racks, lecture halls, housing — trace the likely routes between them, and lay the paths there before the grass wears through. The reactive responses split two ways: accommodate the trace (pave it) or resist it — a diagonal path on Bascom Hill was designed out with rain-water runnels, a humped green space that costs extra effort to cross, and post-and-chain barriers.Inference: The worn line is a free usability study written in dirt — it shows exactly where the designed path and the desired path diverge. The planner’s choice, pave versus resist, is the general desire-path decision: does the deviation reveal a legitimate unmet need worth accommodating, or a shortcut worth blocking for a reason the walkers don’t see? The trace states the mismatch; it does not, by itself, decide the response.
Hyrum's Law, as stated in Titus Winters, Tom Manshreck and Hyrum Wright, *Software Engineering at Google* (O'Reilly, 2020). · computer-science
Hyrum's Law, as stated in Titus Winters, Tom Manshreck and Hyrum Wright, *Software Engineering at Google* (O'Reilly, 2020). · computer-science
Hyrum’s Law states: “With a sufficient number of users of an API, it does not matter what you promise in the contract: all observable behaviors of your system will be depended on by somebody.” The documented API — the prescribed path — is the interface’s contract. But users find and lean on whatever behavior is convenient: an incidental output ordering, a particular error message’s text, a timing quirk, an undocumented endpoint. Those unintended-but-relied-upon behaviors are the digital desire-path, worn not into grass but into the dependency graph, and they constrain the maintainer exactly as a paved-over shortcut constrains a landscaper — you can no longer change what people have quietly come to depend on.Inference: Every observable behavior is a potential path, sanctioned or not; with enough users, all of them get walked. The maintainer’s real lever is upstream — narrow what is observable, so fewer unintended paths can form, rather than trying to forbid the paths after users have worn them in. That is the API-design analogue of laying the official walkway before the grass wears through.
Michael Hammer, "Reengineering Work: Don't Automate, Obliterate," *Harvard Business Review* 68(4): 104–112 (July–August 1990). · business
Michael Hammer, "Reengineering Work: Don't Automate, Obliterate," *Harvard Business Review* 68(4): 104–112 (July–August 1990). · business
Hammer’s reengineering manifesto turns the desire-path’s design lesson on its head for organizational processes. Warning against merely computerizing existing workflows, he wrote: “It is time to stop paving the cow paths. Instead of embedding outdated processes in silicon and software, we should obliterate them and start over.” Here the “cow path” is the entrenched actual process — the informal sequence of steps an organization has worn in over years — and Hammer’s claim is that automating it (paving it) locks in a route whose only justification is that it is the one everyone already walks. His Ford example: rather than automate an accounts-payable process staffed by some 400 clerks, Ford redesigned the process itself and ran it with about 5.Inference: The desire-path signal — “here is the route actually taken” — does not, by itself, say pave me. The same trace warrants opposite responses depending on why it exists: a shortcut that reveals a genuinely better line should be paved (good urban design, per Alexander’s “paths and goals”); an entrenched habit that persists only through inertia should be obliterated, not embedded in software. Reading the worn path is step one; diagnosing why it is worn decides whether to accommodate it or redesign around it.