Through GSM mobile telephone technology and its ability to local people geographically, we can now attempt to measure collective emotions in urban environments. This opens the door to a host of new social and commercial applications for this generation and the next.
For sometime now there have been tools to measure individual emotions. So why are there no tools to assess collective emotions? Afterall, new technologies have a ringside seat where collective emotions well up. During large-scale popular events, for example, the modern tools of communication become the prefered means to share one’s feelings. And every event, country and city has its own rhythm. Be it a music festival or New Year’s Eve in France, Saint-John’s Night celebrations in Poland, a football match in Spain and Romania, the Olympic Games in Beijing or even the tragedy of 11th September in New York. Could we assess this geographically specific rhythm and evaluate these emotions?
Urban Mobs, the results of our research, went on display in December 2008 at the Grand Palais in Paris. Called Dans la Nuit, des Images (Images in the Night), the installation consisted of eight flat screens facing each other, on which spots of light would appear, swelling into luminous fountains, disappearing, jumping from one angle to another. It was easy to recognize the shaded contours of a well-known cityscapes in the background: Paris, Madrid, Barcelona, Warsaw, Cracow and Bucharest.
The visual effect of the eight screens was clear, but what was their true significance? The catalogue describes the work as a Cartography of Popular Emotions and explains how it was developed using data from mobile telephone networks in different countries.
Video 1 – SMS messaging during the Music Festival– Paris, 21/06/2008
Video 2 – Representation of crowd movement (handover) during the Music Festival – Paris, 21/06/2008
We created the images using the Orange Labs Laboratory and data from the Orange mobile network. In Global System for Mobile Communications (GSM) there are multiple information flows. It transmits calls and SMSs, but also all sorts of indicators on the quality of reception and emission from telephones and their geographic locations. It is this last point that interested us. By definition, a mobile telephone is likely to change location and it is important that the operator knows where the terminal or handset is, at least at the time of a call, so it can be handled by the the closest base station antenna. If the mobile moves several hundred metres, it shifts to another base station, and the operator will automatically be informed of this. This made us think: if the movement of a mobile phone from base station to base station can be tracked during a conversation, we could reconstitute most of its trajectory – ie. where the person making the call went. It this can be done for a large number of mobiles simultaneously then we can capture the movements of a crowd. And, as Orange is present in several European countries, we can use our whole network to follow the crowds simultaneously across the continent.
To test this theory we decided to monitor a handful of popular events where big crowds were likely to be on the move and making telephone calls, and then attempt to retrace, through the activity of the crowd, the outline of the event set against the backdrop of the city.
This is what we monitored:
• The Music Festival, 21st June in Paris, France and on the same day, the Midsummer’s Fires festival in Poland, in Warsaw and Cracow; and• two games during the Euro 2008 football championship, which took place in Austria and Switzerland. We decided to monitor the reactions in Bucharest, Barcelona and Madrid during matches between Romania and Italy, and the final, between Spain and Germany.
Orange Labs collected and prepared the data, and faberNovel was in charge of the graphics.
Lets start with the precision of the geographic tracking. We can track phones in an area that we call a cell, the zone covered by one base antenna. Fortunately, this area is not very large in a dense urban milieu like Paris, not much more than a few hundred metres. Clearly, it is not possible to position a single telephone handset in a street, but for the purposes of the exhibition, and given the scale of cities like Paris and Bucharest, the level of precision we achieved was enough to understand the dynamics of the events. We should also point out that only phones that are in use, actually making a call, can be picked up. However, though we can only track telephones in use, it is possible to make a statistical extrapolation, assuming that in a large gathering, statistically there are always some people calling, and they are excellent markers to track the the crowd as a whole.
In addition, it isn’t absolutely accurate to say that one looses track of mobiles that are not in call. In fact, the network divides the territory into large zones, called location zones, and each of these groups hundreds of cells. The idea behind these zones is to predict where phones might be: when one tries to connect one mobile user to another, if the network had no idea where the mobile is located, it would need a considerable amount of time to scan all the cells in France, for example, to find the intended handset. This could take a long time if the network had to work its way from Dunkerque, to Marseille, to Strasbourg…
In fact this doesn’t happen. The network constantly memorizes in which cell each mobile phone is located, whether it is in use or not. It would take too much computing resources to monitor each phone, so the operator makes a compromise: it memorizes the cell of the mobiles actually in use; but for the others, it only memorizes the "location zone" in which they are located. Thanks to this information, we can generally locate the position of everyone on our network, even if they are not using their phones.
Another key consideration is the size of the crowd. Just how many people can you deal with? In reality, if you want to systematically track the movements of millions of people, as was the case in Paris during the Music Festival, you have to plan accordingly. What we showed at the Grand Palais was actually taken from events that were six months old, so we had time to process the data. But it is also possible to track movements in real time, or at least the movements of about a thousand people. The constraint is synchronising the flow of data from a million people. And that is what we are working on at the moment.
The questions that come up about our efforts are not only technical. Locating people via their mobile phones raises the issue of confidentiality and privacy. The sceptre of Big Brother looms. Clearly everyone has the right to privacy as regards his presence in a specific location. It goes without saying that, in our exercise, no personal information was exploited as we kept within clear European Union laws. All data were rendered anonymous, that is we did not know who owned the phones and no one on the team had access to that information. We also never isolated a group of less 2,000 people (to comply with another legal ruling) And finally, the paths visualized on the screens at the Grand Palais did not belong to a single telephone, followed on a long term basis: we never captured the trajectory of an individual handset over the whole period of the tracking experiment as if we were trying to trace its history. We processed the data from the GSM network in the same way you might monitor road traffic by counting cars passing by on a highway. Our experiment was global, aggregate, anonymous and without memory.
There is a brave new world of mapping opening up. If one uses the typology of “live maps”, which has been done elsewhere, then this particular visualisation would be in the category of “republic maps”. A republic map shows dynamic mapping where the points represent people, without distinguishing one from another, along the lines of the republican principle: one man – one voice. All the people made visible by the system at a given moment are represented in exactly the same way. This should not be confused with other types of mapping under development. For example, some investigations, Citysense from Sense Networks for example, analyse places by distinquishing and visualising different groups of people who live there (the young, rich, fans of …). This type of study is very close to marketing analysis; its aim is to segment potential consumers in time and space. Another type of personal mapping is associated with the community services of the Social Web (and Web 2.0) In this case, it is the mutual locations of a group of friends that are shown on the map.
Each type of dynamic map raises different privacy concerns that require different solutions. In our case -- global visualisation of telephonic activity in space – raises the issue of crowd control by an operator or supervisor with bad intentions, similar to problems raised by the visualisations done by the City Lab of the Massachussetts Institute of Technology MIT. Some people even talk about modern “geo-slavery” influenced by the data of geo-localisation. What we do is completely different. We find that this method of global usage of localized data is easy to monitor both from both a legal and public policy point of view. This type of data – what we call republic data – whether it is produced and stored by the public or private sector – be it mobile telephony, road traffic, public transport – is subject to to strict controls by both the authorities and civil society. Regulations are evoling as part of a transparent public debate. This data, considered globally, could be said to be a “public good”, something that is useful to the community as it contributes to a better understanding of web surfing, urbanistion, planing or simply understanding the phenomenum of urban life. It is different for the other forms of dynamic mapping where small, fly-by-night companies, even the users themselves, use the potential of GPS (now available in telephones and computers) to develop all sorts of applications without giving any serious thought to the long term consequences of making this type of information available online.
Apart from the artistic effect of the videos of the Grand Palais, they clearly have a real sociological interest. They show centres of attraction that vary with time. The map comes “alive”; the temporal dynamics of the presence of people in the city opens a new dimension of mapping. Just take a look at the dynamic images from the Grand Palais exhibition. The videos Paris as seen by SMS, for example, shows a light geyser rising from the Pace des Prince on the night of the Music Festival during the concert of the German rock group Tokio Hotel, followed by another at the Auteuil racecourse where France 2 TV channel organised another big concert. These were both events during which a young and technophile public could and did send thousands of messages. In another example, the video showing calls made by foreigners in Paris, you see how most of them met at the Eiffel Tower or in front of Notre-Dame, the classical tourist meeting points, and they were quite indifferent to the festival going on around them.
In the videos that captured people in Romania or Spain during the 2008 European Football Championship (Euro 2008), you can see a sudden increase in the calls during the half-time periods and telephonic silence at crucial moments in the games. You can also see the the explosion of joy that accompanied victory, and follow minute-by-minute, for example, the victory parade for the Spanish players, when they returned to Madrid the day after the final.
Video 3 – During the match Spain – Germany in Madrid, 29/06/2008, 19h55, then Final whistle – Spain is the champion - Madrid, 29/06/2008, 20h24
We are continuing to make improvements to our system including incorporating new possibilities for traffic follow-up, afforded by the 3G mobile phones and the possibility of aggregating data from several different operators. which would give better geographic precision to our system. Now that we have demonstrated what can be done artistically with this technology in Urban Mobs, we can turn our attention to other potential applications. The movement of pedastrians, for example, could be of great interest to many different organisations including communities and companies that organize events. Think, for example, of potential security issues or how to manage crowds at amusement parks. Urban planners could benefit from integrating data from public transport flows. And road network and transport companies would have an interest in tracking people to help their planning, even perhaps better detection of traffic jams in real time. Data from technical networks can become a public good, to use the economic term, useful to to help communities understand the dynamics of the physical spaces.
This technology can also benefit targeted advertisements and geo marketing. Researchers are already hypothesising about the potential of linking mapped information on goods and services to the real flows of potential customers for both merchants and customers. This becomes especially interesting if we imagine that in the future there could be a network of mobile terminals connected to the Internet. Sense Networks, a US company, developed Citysense, which uses geo-localized data to construct dynamic segmentations of the population to target commercial options in the urban landscape. Citysense, which is in beta test in San Francisco, lets Blackberry and iPhone users see where the most numbers of people are congregating in a geographic location and then connects to search engines that identify bars and clubs in the vicinity.
Social networks and Web 2.0 developers are also keen to use this type of technology to help people who share the same interests and passions link up with soulmates in specific geographic locations. “Flash Mobs”, “Aka-Aki” and “Latitude”, for example are all services that want to scan networks of friends or people in a specific location in real time to help people organise and get to meetings on time. This type of application will clearly develop in the months to come, all over the world. We are convinced that the aggregation of data coming from different sources will take these kinds of applications to a higher level. For this reason, it is important that application developers remain as autonomous as they can from the sources of the data, something that is well understood in both the US and Europe.