More selected projects

Walking with Data

What do you need from a companion when you go for a walk? How far can your mobile phone provide this? Just log on to to explore the possibilities!

produced by: David Upton


This piece links your mobile phone to a remote website, and downloads code which enables your phone to talk to you and to take an intelligent interest in you and your surroundings. Based on insights from psychogeography, it tries to expand your perceptions and help you to see your surroundings n a new light. All you have to do is log on using your smart phone: the interactions between you and your phone are the core of this art work, and you can access them whenever you want and take them wherever you go..

What do you need from a companion when you go for a walk? How far can your mobile phone provide useful conversation?

Concept and background research

The psychogeographic practices originating with the Situationists have been a major influence, together with the cybernetic view of art as process rather than object – 'Art, clearly established in our present culture as a form of behaviour rather than a simple array of images or objects, now takes on the quality of a transaction” (Ascott, 1975)


I think we've forgotten something of the freshness of the early digital art pioneers: for instance the interactions of Pask's 'Colloquy of Mobiles'.


I've always seen psychogeography as one of Foucault's 'techniques of the self” - a means by which we can work on our own thoughts, conduct, etc., in order to transform ourselves. It works by trying to force us to see things differently. (Foucault, M, 1982). My insight is a kind of 'computational psychogeography', where some of this transformative impetus is supplied by a machine.


I first developed a Raspberry Pi system using sensors to extend our perceptions , but then realised that the mobile phone offered far more affordances, not least easy internet connectivity.


I researched data available over the internet, usually via APIs. There is an almost unlimited amount of this. It is easily possible to build a Merzbau of data, of varying quality and usability. Kurt Schwitters, the inventor of the Merzbau, wrote; ‘Merz means establishing relationships, best of all between all the things in the world’, and as ‘a standpoint that anyone can employ’: so many things are there to be used. (Webster, 2007, p 132.)


COVID19 and the 'lockdown' made it even more important that the system should seem like a human friend rather than just a random generator of ideas. (There are several of these: a computational example is the Randonautica system ; a similar system uses a pack of cards.


To make the system relate better to the users, I use the 'Big Five' psychological personality scales. (Gosling et al, 2003). These offer simple and quick method of 'scoring' a user by their actions and correlating these scores with personality traits. (Though psychologists differ on how useful they are, Cambridge Analytica apparently used them to help get Donald Trump elected!)

Socring is not done to judge users, but to improve the site's responses to them. For example, a user who regularly uses the 'help' pages will be perceived as nervous, and the site will try to sound more reassuring. Calibrating this system - deducing scores from user actions - is however difficult, and will take time.

I trialled the system at the 4th World Congress of Psychogeography in September 2020 as a first step to this calibration.


The work is based on a mobile phone, since everyone has one. It does not require anything special by way of equipment or software, although of course a mobile phone is a miracle in itself, with complex affordances and processing capability. Yet we all have one, so nobody is scared of them, and everyone uses them naturally.

P5JS gave me an ideal medium to write in, saving a lot of hard work creating an 'app' for Android and IOS. It allows the central computer and the smartphones to communicate. P5JS runs on virtually any browser, and is largely platform agnostic. Phones do differ a little in their affordances, but most have a common set, including GPS sensitivity, the ability to hear and to speak, and the ability to browse the internet, all of which can be accessible through P5JS. Typically, the phone originates a request, including latitude and longitude coordinates, which goes to the web site: the site looks up data on one or more APIs, builds an HTML response, and returns it to the phone.

The web site at the heart of the system is my own artist's site. It runs on a Python server in the cloud, which should make it highly robust and capable of handling bursts of activity. It builds a mySQL database, based on users' responses and interactions. This should allow it to scale easily to any size that I am likely to need, at least at the start of the project.

Quite a lot of the data the site redistributes is derived from APIs to other sites. This gives a very wide range, though there are surprising omissions. It also makes the system partly dependent on external sites, which may change, and network connections, which may fail. For these reasons it has been built as far as possible to 'fail gracefully', and I have also started to build in an analysis system which should help me to spot when an API is no longer working, perhaps becasue it has changed its rules. To minimise the number of calls, APIs which only update once a day are only looked up once and then stored as text until the next day.

Future development

This is always going to be a work on progress. Given the fluidity of its sources and of technology, I do not expect it ever to be 'finished'. New affordances will appear on phones. New systems will arise to prevent mis-use of phone technology, for instance by requiring the user to confirm when cameras or GPS may be used. APIs will come and go, and change. New ways of securing data transfer across the internet will appear. (For example, certain data can no longer be exchanged to different sites, or by sites which do not have an SSL connection. Five years ago, this site would have been easier to write, but that early version would no longer work today since the Cross Site Origin (CORS) rules came in.)

I want to build in more interaction, so that the system can 'understand' when users talk to it; and I want to give users the opportunity to download a record of their walk, perhaps in the form of a diagram or of a poem.

I would like this project to be seen as a 'machine of loving grace', as in Richard Brautigan's poem. (Brautigan, 1967). Such is the academic and media suspicion of data collection and transfer that this is an uphill struggle. It may be seen as 'big brother' even by those people who put an Alexa in their own house to listen to every word that is said there!

Self evaluation

I could have planned the site better, instead of building it in stages as inspirations came along. If I had, I would have started with the W3 design framework, rather than putting it in half way through. Owing to my inexperience of Javascript, I did some things the hard way. Also, some p5js add-ons didn't work as I expected, and I did not have time to research why. COVID19 did not help by causing the internet to slow down!

What I did do was a lot of active research into psychogeography and the ways a system like this could be used. I exhibited it, and its Raspberry Pi-based predecessor, at successive World Congresses of Psychogeography, which gave me a lot of feedback. I did some useful research into how other systems can facilitate derives. I developed a better idea of the ergonomics of walking with a smart phone. This research helped me to 'ground' the project.


Ascott, Roy, 1975: 'Table' (quoted in 'Telematic Embrace', ed Shanken E, Univ of California, Berkeley, 2003, p 172)


Brautigan, Richard, 1967; for the publishing history see, for the text of this poem see


Foucault, M, 1982: 'Technologies of the Self'', lecture, accessed via


Gosling, S, Rentfrow, P and Swann, W, 2003, 'A very brief measure of the Big-Five personality domains' , Department of Psychology, University of Texas, Journal of Research in Personality 37 (2003) 504–528


Webster, G, 2007: Unpublished PhD thesis on Kurt Schwitters' Merzbau, Webster, g: Open University 2007, accessed via (DOI: