Social contouring: our proposal to fight against COVID-19

Framing the problem from a societal perspective

As the novel coronavirus spreads globally, it has become apparent the contagion moves along the contours of human lifestyles and culture. The virus spreads in alignment with risks inherent with human behaviours, with some activities more hazardous than others. Epidemic spread has been studied through communities. Earlier research has shown that the response within local communities and the traffic between communities is an important factor in the spread and mitigation of the virus. In the case of Ebola, reducing traffic between communities with approximately 50% has been highly effective in suppressing the outbreak[6]. A current study is analysing the spread in a university campus through networks structure according to study plans and disciplines[7]. Considering lifestyle characteristics is still missing. The application of an individual-level model of the SARS-CoV2 transmission that is location-dependent in a study showed that age distribution and family composition affects the virus spread[8]. We hypothesize that the virus spread is contingent to people’s lifestyles and their characteristics. Some behaviours might be loaded with propagation factors that put subjects more or less at risk of contagion without knowing.

The protracted lockdown has us all wishing for more human interaction, and a restart of economic activity without risking the infection that would continue the spread that has bound us to our homes.

How to gain visibility on the propagation?

A heuristic mobile-digital response

Various stakeholders (from public institutions, private companies) have attempted to provide a technological response to the pandemic at the tip of our fingers: mobile apps. These produce native-digital data raising privacy concerns that do not provide a comprehension of the virus spread. These data allow pointing at some people and places (e.g. via contact tracing). They respond to whom, when and where but mobile phones can introduce errors (radio distance is not spatial distance, and much less infection distance). Moreover, public health response might not follow with the current available resources (e.g. tests, hospitals’ capacity). On-going solutions track locations, symptoms, test results and some demographics. Asian governmental solutions also track and measure myriad behaviours at an impressive granular level (see South Korean datasets). Data are partial, reductive, permit surveillance and control. However, 80% of people in some European countries would probably download an app in that would provide them information about the spread. A survey in France shows the acceptance of usage of personal telecommunication data varies according to age, political orientation and educations level. These technological solutions are aligned to a numerical argument: decrease the number of contaminations (see factor R<0). Perhaps useful to some extent. The speed of the propagation could match the design of some of those solutions; we also believe a data-oriented approach can be useful if well framed (see MyDataCommons). However, the solution does not have to be solemnly native-digital: apps physically embedded in mobile devices and producing “social traces at high-frequency”[2]. Data produced from ethnographies and observations (qualitative research) can provide specificities that could be later scaled for comparison in combination with national surveys. To be inclusive, but also to exploit existing resources, other possibilities have to be considered. “We cannot assume that all communities want and need the same approach to managing exposure via mobile devices, nor that they want or need the same universal solutions. How can we critically discuss and continuously assess who this technology is designed to protect, and who is being externalised?” [3]

A possible societal response

When considering the novel coronavirus, we postulate that the pandemic is socially situated among the differences in individual characteristics from country to country.  For example, urbanization, proximity of living arrangements in households, such as the numbers of household members, square meters of living area, age, and intergenerational living can all impact how the virus spreads. What is critical is that the novel coronavirus has an infection period that may be part of the pre-symptomatic and symptomatic phases, thus people are spreading the disease before they know they have it. In the absence of a vaccine, this requires a social response to the disease and an understanding and deconstruction of the socially situated disease.

Without the help of biomedicine, and in order to prevent the overload of healthcare systems, a multisectoral response is required. Virology, epidemiology, data science, engineering sciences, social sciences, and biomedicine must all contribute in a holistic manner to ensure a successful response. Taking a hands on approach, where we respond to the virus as it spreads to break it down and learn how to counter in. This requires challenging mind-set polarities, cooperative thinking to avoid blocking up ourselves in criticism. We could recall pragmatism theorized by John Dewey: the social knowledge must be produced in such a way that it is both useful for the continuity of social experiences and tested, or in probation (mise à l’épreuve) by its social effects.

A societal response requires taking action at two temporalities, a long-term vision inclusive of ethics, critics, reflections, problematizing, theorizing, analysing; and another one that is shorter for testing solutions, making corrections, accepting mistakes, constant dialogue. Then it will be possible a science that is actionable, we will build solidarity and trust through coordination. In this pandemic, we are all involved in a common project, even those now muted.

We should also rethink the dynamics of society, including the different levels that compose it but have been historically analysed separately. First, the social as a whole with social structures, social rules (Durkheim). Second, the traditions and values underlying inheritance where structures are structured and continue to structure (Bourdieu). Finally, individual beliefs that are actionable and therefore desired (Tarde’s social theory of monads). These can be imitated while creating at the same time distinction. This final point of view introduces “networks of attributes”[4], one could think of neighbourhoods made of monads, or particular desires in propagation.

It is particularly fruitful to revive Gabriel Tarde’s theory of contagion not only to understand the social dimension of the virus spread, but also to understand the proliferation of technological solutions. We are experiencing a horizontal contagion of the COVID-19 according to individual similarities, and a vertical contagion, in a suggestive mode where there is a prestige gap between the inventors and their copiers (see Apple, Google and DP-3T solution). Boullier has called it “The tyranny of delay”.

“We have a vision of the pandemic that is distributed, fragmented with a multiplicity of actors and heterogeneous conventions.”[5]

Mapping social contours: our proposal

To better support the adaptations the current situation is demanding, we believe a solution has to be modelled (formally and mathematically) including the differing dimensions of: virology, epidemiology and socieconomy. This could be complementary to a geographical modelling, currently in evaluation in Germany and the Netherlands.

After two hackathons and continuous discussions since 3 weeks now, we have started with exploring a possible social model and built a critically-assessed questionnaire to gather data and enable clustering. To model, from a medical anthropological perspective, we have defined 8 dimensions that describe lifestyles according to propagation factors:

  1. Individual’s view of government measures
  2. Circumstances of socialization
  3. Resources available
  4. Working conditions
  5. Food supply
  6. Communication means
  7. Health state
  8. Demographics

From the questionnaire responses, we seek to understand better people’s lifestyles in relation to the virus spread and contribute to the containment of this epidemic by providing them information to readapt. We address two main questions:

  1. Which lifestyle factors can be identified contributing to the propagation?
  2. Which social contours can be enabled to relax the lock down?

Social contours are safe zones, or safe paths, that people can undertake without the risk of causing a new outbreak. Social contours allow people, communities, to start interacting with others and reactivating the economy.

Methods

By conducting in-depth questionnaires, our aim is to obtain a snapshot of contours of human lifestyles and culture that the novel coronavirus moves along. The data will separate behaviours that contain risk, compared with those that do not, to better define how society can transition from the lock-down. There are two methods we have identified to analyse this data:

  1. Community detection[I-Louvain] to understand what factors characterize a group of people, and make one lifestyle interact with another one across communities. This produces a social model of the epidemic. Communities or networks are considered as social structures and their formation is often explained through homophily. Homophily is the likelihood of being friend, or being in the same group of somebody else with whom one shares similar characteristics like tastes, occupation, social class. Few studies integrate heterophily (i.e. to interact with others from different groups) that we also take into account. Indeed, dissimilarities have social influence in preferences and opinions[9]. One particular study is of our interest, authors disaggregate homophily often constructed from general categories, to study social structures (exogenous) and in parallel the mechanics of tie-generation from more granular attributes of socialization (endogenous)[10].
  2. Computational simulations. Through our social models we can simulate in real time a virtual version of society measuring the impact of a policy on the propagation across communities. One could understand better the differential impact of measures according to specific regions and cultures.

The results will allow conducting risk assessments by introducing a situated characterization of lifestyles.

Achievements:

We have created the questionnaire under a legal and ethical framework. It is available in five languages: FR, EN, ES, DE, IT. Accessible upon request.

We have collected a sample in Switzerland and other countries.

We have analysed a sample of 771 responses of French speakers.

We have conducted an exploratory analysis of communities. We identified 11 types, and described 5 lifestyles with the highest impact on propagation.

Compass of possible solutions

Prof. Dominique Boullier suggested: “A citizen-centred design, a responsible and democratic application development, which does not start from the solution but from the framing of the problem and then builds all the chains of possible solutions, of which the digital part is only one part and whose performance will depend above all on its integration with the strategy and the organizational design adopted.”

So let’s try to sketch here some possible solutions recalling our initial problem: the virus spread is contingent to people’s lifestyles and their characteristics. How to capture the propagation factors according to those lifestyles that put subjects more or less at risk of contagion?

In  the compass below we present  current solutions adopted and created by different actors that are positioned in oppositions at the end of the arrows. For each quadrant, one can find the type of solutions and the data collected, they can include different actors. For instance, State and Experts produce crowdsourced data for mobile apps that inform Citizens about the infections.

 

In the second compass below, we characterize the type of solutions that are actually provided by different actors that might, or not, cooperate with others. Self-generate social response is where communities are creating adaptive practices, mainly individually or per household where we don’t have visibility yet. And collecting intrusive sensitive data will not help us understand those practices.

 

The current situation is demanding adaptations that can come from different actors and that will not be the same for all. We see four dimensions of a social response that could be informed with our results:

  1. Internal: I understand the virus and what can do to me
  2. Individual: given the virus, I need to do this or that to protect myself according to my lifestyle
  3. Collective: how is people around me handling this, can I join them or do I self identify differently
  4. Act together: let’s reach out authorities, advocate according to our lifestyles…

 

These are the “we” (all ideas in this post are ours):

Jessica Pidoux

Judith Herzog

Marie-Pierre Vidonne

Maurizio Arseni

Michele Loi

Miles Fahlman

Paul-Olivier Dehaye

Rémi Petitpierre

Winfried Tilanus

Bogdan

with personaldata.io

and MyData.org

We are currently trying to make this project happen with the support of versus virus. The project has been highlighted in the hackathon and we already received some funding from that award. There is more funding available (20k) from the Swiss Confederation to apply for. We want to conduct a national survey, and we are looking for epidemiologists, experts in public health, developers, data scientists, sociologists and any other person that would like to join this fight with us. The questionnaire will be available soon if you would like to tell us more about your lifestyle.

Write us at socialcontouring@protonmail.com for more information.

Leave a Reply

Your email address will not be published. Required fields are marked *