I read about this happening in an American university some years ago. They built a new building and planted grass around it. The grass was trampled by users and the planners used the grass marks to instal the footpaths int he areas that users needed.

I read about this happening in an American university some years ago. They built a new building and planted grass around it. The grass was trampled by users and the planners used the grass marks to instal the footpaths int he areas that users needed.

An interesting idea. If desire lines are visible already then does this not suggest a certain amount of democracy (crowd sourcing) has already occurred. As people have already voted with their feet. Usually when I have encountered desire lines they provide a logical short cut to save distance/time across a piece of grass. I have often wondered why planners, landscape architects haven't used a bit of common sense when designing some paths or if desire lines are clear why not just turn then into managed paths.

There might be some merit in using technology to explore the nature of desire lines as the occur to inform the debate or use crowds to effectively design their own paths in newly developed areas. There should not be much technical, ethical or social issues as mobile technology is becoming ubiquitous and if data is collected on a voluntary basis with proper safeguards to the data (e.g. anonymous), however as stated above would you need to go to that trouble if desire lines are clearly visible in the first instance.

If you review the OSM dataset, there are a lot of tracks and paths in the wild:

http://wiki.openstreetmap.org/wiki/Maps#Biking.2C_Geocaching.2C_Hiking.2C_Sport

Of course you can setup a nw portal, dedicated for mapping only this kind of ways, but as installing the OSM Rails port (the website) is tricky, I wouldn't recommend this approach.

Another aproach is to use GPS raw tracks (maybe from the OSM GPS tracks dump, gpsies.com , ....) to filter this tracks. But as the GPS signal quality is especially in the woods very poor, mapping this tracks needs IMHO a real on the ground survey.

Desire lines a clearly visible only in parks and green areas, where the grass is not there anymore. This is usually not the case in urban areas, where bicycle and pedestrians go on each other lines, or on the roadway, without leaving any sign.

Maybe a new type of feature "desire line" can be added to OSM directly.

Sure, but usually we model it using highway=path or highway=track

http://wiki.openstreetmap.org/wiki/Key:highway

Interesting idea. I don't mind collaborating and pitch in services from ideaken to make it happen.

The idea is interesting. I thing its comes presenting the subject in a way directly involves people who feel part responsable

This idea could provide an alternative to the notoriously unreliable method of forecasting future transit trips using traditional four-stage transportation models. I have long argued that it would be better for planning transit if we used the private trip matrix for determining where people really want to go, rather than the transit trip matrix which is overly influenced by existing transit availability.

But this new idea may be even better. Why not simply ask people to indicate their preferences on a map? Of course, the problem with stated preference surveys (which crowd-sourcing seems to be a version of) are also unreliable for planning and investing because they allow a high level of whimsy. How could we counter this?

In Galway City (west of Ireland), we have done some work on the mapping of desire lines using aerial photography and site visits. This was done in the context of permeability and connectivity with recommendations for knocking holes in walls, paving desire lines etc. I really think a crowd-sourcing GPS-based approach would be best for mass mapping of desire lines. The public could be encouraged to map their shortest and safest routes using a mobile app. If anyone has ideas on how to start such a study, or if you would like to discuss the work we have done in Galway, just leave a message here or email me at [email protected]

First of all, I would like to break the problem into three distinct [and yet related] questions:

1) How and to what extent VGI should be collected for this purpose: VGI could be collected by asking people to log their routes (using mobile devices), choose their desired routes (e.g., on a web interface), or a combination of both. While the first method would yield the routes that people tend to use frequently in their daily lives, it cannot reflect the actual routes they "wish" to take. On the other hand, the latter method shifts towards optimistic and perhaps unrealistic routes, due to the fact that people are not aware of or cannot simultaneously consider all of the contributing factors that make a route superior for them or other citizens (e.g., traffic load, distance, and what they enjoy during their rides). They may also be asked to propose a number of routes and prioritize them. Or for example be asked to propose routes for specific groups of people instead of themselves, in order to minimize their level of selfishness in choosing routes and yet enjoy the collective intelligence thing (e.g., ask them to suggest routes for people who go to shopping, kids, someone going to work, or somebody who wants to enjoy a ride).

2) How to convert the high amount of unreliable and uncertain VGI into useful information? We need to minimize the noise and let the patterns emerge. Just an idea: we could consider the routes in VGI database as virtual movement of people. Hence, we could evaluate the level of virtual traffics for every given piece of street or area, and eventually evaluate those desire lines (those segments of streets with highest VGI-driven traffics).

3) What Mikael Colville-Anderse says is true that cyclist citizens know or experience so many things that experts don't. However, the route(s) demanded by citizens is not always the best route for all people, or at least simultaneously the most secure,feasible, and efficient route to be maintained in near future. Citizens often analyze things at a very limited scale and do not have a systematic insight to the complexity of urban environment. Hence, we should answer to the questions that: To what extent VGI should be used as a basis for decision-making? How could it be mixed with other aspects that the expert is aware of? How could it be augmented using other types of data? We could refine and abstract VGI and give an alternative to the decision-maker with the following conclusion: "Based on my data, this is the route that our citizens like to take". Hence, other parts of analyses would be systematic and qualitative. However, I believe that there could be more to it. We could find where the VGI-driven routes coincide with suitable/applicable/feasible streets. Where are the routes that have high traffics, which routes lead to dense areas, which routes cut the traffic. Such occurrences have to be scored and quantified. Eventually, the VGI-driven routes that are precisely analyzed in their context of occurrence would be of higher value and weight to be merged with behavioral observations (or other types of observations or analyses).

You should be able to predict the desire lines using an Ant Colony Optimization if you have an origin / destination matrix for the locality