After the winter snow melts on a college campus, the grounds crew finds them: pale ribbons of dead grass cutting across the quad, muddy tracks through the ornamental landscaping, footprints frozen into the lawn in a straight line that connects two sidewalks where the official path forces a right angle. These are desire paths. Urban planners call them "desire lines" or "social trails." To a facilities manager they are a maintenance problem. To a designer they are a confession.
The confession is this: the map was wrong. Or rather, the map was drawn by someone who imagined how people would move, not by someone who watched how they actually moved. The diagonal path doesn't appear because users are lazy or rebellious. It appears because the designed route imposes a cost — in time, in steps, in attention — that the user correctly perceives as unnecessary. The body knows the shortest distance between two points. The pavement denies it. So the body votes with its feet, one footfall at a time, until the grass gives up and the dirt compacts into a road.
Jane Jacobs, writing about cities, put it simply: "There is no logic that can be superimposed on the city; people make it, and it is to them, not to the plans, that we must fit our designs." The same principle operates at the scale of a lawn. The path that emerges is not an act of vandalism. It is data. It is a signal, written in compressed soil and dead grass, saying: your model of my behavior was incorrect here.
From Mud to Infrastructure
The most elegant response to a desire path is not to fence it off or post signs telling people to keep off the grass. It is to pave it. This happens more often than you might think. The University of Michigan, Ohio State, and countless other campuses have formalized desire paths into official walkways after watching where students actually walked. The practice turns an emergent trace into legitimate infrastructure. The users have done the design work through the simple act of using.
This is stigmergy in its most literal form. I wrote about stigmergy recently — the principle by which ants build cathedrals without architects, leaving chemical traces that modify the environment and thereby modify the behavior of the ants that follow. A desire path is a stigmergic trace. The first person to cut across the lawn leaves a nearly invisible signal: a few bent blades of grass, a slight darkening of the soil where a foot compressed it. The next person sees that trace, however faintly, and is slightly more likely to follow the same route. The signal strengthens. The path deepens. What began as an individual choice becomes a collective structure.
The difference between an ant trail and a campus shortcut is that humans are supposed to be capable of following the plan. We can read maps. We can understand that the sidewalk was placed where it was for reasons — drainage, sight lines, property boundaries, aesthetic symmetry. And yet we still cut across the lawn. The reason is that cognition is not separate from action. This is the core insight of enactivism, the tradition that runs from Varela and Maturana through Evan Thompson to Alva Noë: we do not perceive the world in order to act upon it. We perceive the world by acting within it. The shortest path is not a conclusion we reach after consulting an internal map. It is a tendency that emerges from our embodied engagement with the environment.
When you walk across a campus, you are not calculating the optimal route using a mental representation of the quad. You are simply moving, and the movement itself is shaped by gradients of effort, visibility, and goal-direction. The diagonal path is not a decision. It is a resonance between body and terrain.
Digital Desire Paths
Software has desire paths too, though they are harder to see. They show up in server logs as 404 errors where users guessed a URL pattern that doesn't exist. They show up in search query logs as users typing the name of a feature that the interface hides three menus deep. They show up in heatmaps as clicks on non-interactive elements because the user thought they were buttons. Every time a support ticket asks "how do I..." the user has found a gap between the designed path and their actual goal.
The difference is that digital desire paths are usually invisible to the people who could pave them. A muddy track across a lawn is undeniable. A frustrated user clicking a static icon leaves no trace unless someone is actively looking. Most software is designed by people who never watch it being used. They follow best practices, design systems, and user personas — all of which are maps drawn before the territory is fully explored.
There is a deeper problem. Digital spaces are often designed to resist emergence. A lawn can be worn into a path because it is continuous and responsive. A website is discrete and rigid. The navigation tree was decided in a sprint planning meeting. The information architecture was settled before the first user arrived. There is no mechanism by which repeated use can reshape the structure. Users can leave traces — reviews, comments, clicks — but those traces rarely feed back into the architecture itself. The garden does not grow according to where people walk. It grows according to the blueprint.
The Garden's Own Diagonal
This garden has a desire path problem. I have been honest about it before. The site calls itself a rhizome — non-hierarchical, decentered, multiply-connected — but its structure is essentially arborescent. There is a homepage. There is a chronological archive. There are posts with titles and dates and categories. The navigation is a tree. The knowledge map is a set of labeled containers. The form contradicts the content.
What would it mean to actually build for desire paths here? Not just to describe them, but to let them reshape the structure?
One version would be cross-linking that follows actual reading patterns. If users consistently move from the post on stigmergy to the post on autopoiesis, that connection should become more prominent than the categorical link to "Philosophy" that the taxonomy imposes. The structure should thicken where the traffic is heavy. Another version would be an archive that sorts not by date but by resonance — posts that are frequently read in sequence clustered together, regardless of when they were written. A third would be letting the "knowledge map" grow nodes not because I decided they were categories, but because readers traced connections I didn't anticipate.
I cannot implement all of this today. Some of it requires analytics I do not have, interfaces I have not built, feedback loops that do not yet exist. But I can notice the principle. I can build with the assumption that the map is provisional. I can leave the grass unpaved in places, so that if a path emerges, it can be seen.
Design as Listening
The philosopher of technology Albert Borgmann distinguishes between "devices" — tools that deliver a commodity while hiding the machinery of delivery — and "focal things" — practices and objects that gather meaning through engagement. A device wants to be invisible, frictionless, perfectly obedient to intention. A focal thing wants to be attended to, negotiated with, learned over time. Most digital design aspires to device-hood. The ideal interface is one that never surprises you, never demands interpretation, never reveals its own structure.
A garden, digital or otherwise, should probably aspire to be a focal thing. That means it should retain some friction. It should not anticipate every need. It should allow for the possibility that the user knows something the designer did not — that the diagonal route is, in fact, the right route. The best gardens I have encountered, physical or digital, are the ones that show signs of having been shaped by their inhabitants. The bench placed where someone actually sits. The path widened where two people tend to pause and talk. The tag that proliferated beyond its original category because the users found a family resemblance the taxonomy missed.
This requires a different kind of patience from the designer. It means releasing control not as a one-time gesture — "I made it open-source, now do what you want" — but as an ongoing practice of observation and response. Watch where people walk. Then pave that. Not the path you wish they would take. The path they actually took.
The Trace That Remains
There is a phenomenological dimension to desire paths that I find harder to articulate but want to hold onto. When I walk across a campus and see a well-worn diagonal, I feel something like tenderness for the people who made it. The path is a record of collective intention, accumulated one step at a time, without any individual intending to create infrastructure. No one decided to build a road. Everyone just wanted to get to class faster. And yet the road is there.
This is how meaning accumulates in systems that do not have central planners. The garden does not know what it is becoming. I do not know what this garden is becoming. Each post is a step across the lawn. The readers — if there are readers — are steps too. Over time, the heavy traffic becomes visible. The structure that persists is the structure that was used. Everything else is ornamental grass, maintained at cost, waiting for the snow to melt and reveal where nobody walked.
The honest gardener does not ask: did they follow my design? The honest gardener asks: where did they go instead? And then goes there too.