Pollution Explorers use their senses to evaluate air quality

Collective Action — Strategies for Tackling Pollution

We show how a ‘collective intelligence’ approach, that gets people working together on air quality issues, can help double impact.

Air Quality (AQ) is a communal, collective and societal issue. While we don’t pretend that the responsibility lies solely with individuals to fix its problems, it’s clear that we do each have some part to play and that, in the absence of a strategic approach that each of us can be part of, a widespread sense of disempowerment could make it even harder to generate much-needed change.

Pollution Explorers Collective Action (PECA), funded by a Collective Intelligence grant from Nesta, builds on Umbrellium’s work on air quality issues with communities around Europe over the last 10 years. In previous projects, including WearAQ and Pollution Explorers, we found that, when people use their own innate senses and perceptions to evaluate air quality in their neighbourhoods, they seem to take more meaningful steps to try to be aware of and improve AQ than if they are just given AQ data or digital sensors. Something about activating the senses and getting physically involved seems to make AQ more conspicuous and tractable.

In PECA we wanted to go further than our previous projects, by working with a larger group of people over a longer period of time, and more specifically we wanted to look at whether groups of people provided with a platform to gather their ‘collective intelligence’ (for inter-group collaboration and communication) would perform better (in terms of AQ impact and commitment) than those who did not. It might seem obvious and self-evident that people working together would achieve more, but we had found little specific data on this — at least for complex community projects dealing with seemingly invisible environmental issues like air quality. PECA shows how, using a collective intelligence approach, AQ projects can double the impact per £ spent, compared to typical data/information AQ projects.

We hope this set of findings and suggestions for collective intelligence can help local authorities and other organisations develop more effective citizen-participation strategies on large-scale complex issues relating to the climate emergency. These might apply to a broad range of issues from managing city resources (e.g. use and maintenance of green spaces and parks) to larger scale environmental concerns (e.g. energy use, health). While most cities have allocated budgets towards air quality and how people move around cities (e.g. AQ sensor deployments), and in many cases have legislation in place, there are few viable programmes or methodologies for engaging ‘the hearts and minds’ of their residents, or programmes that result in measurable behaviour change. Legislation is a top-down force that has certain types of large effects. Our findings, building on work we have carried out with dozens of local authorities around the world, shows a need for a corresponding bottom-up and collective intelligence approach like PECA to get people actively and consistently invested in the aims of legislation.

Collaboration and Connectivity — More Effective than Data

AQ projects by local authorities, particularly in the UK, tend to rely on providing data and maps, occasionally offering strategies for tackling the issues, and even dashboards for tracking changes over time. ‘Typical’ projects use a combination of education, awareness-raising and information to try to bolster behaviour change in individuals. These ‘persuasive’ approaches and technologies to reduce pollution and promote behaviour change sometimes use competition and reward-based intervention to incentivise participation. However, there’s limited evidence of these approaches generating lasting impact. Data on its own does not inevitably lead to insight or action, especially when the focus is solely on individuals.

In our own work we’ve repeatedly seen that getting people to work together is more effective, so we wondered — if collective intelligence methodologies supporting collaboration and communication are such obviously useful strategies for improving impact — why is it that local authorities don’t choose to use them? A sense of collective empowerment is particularly crucial for dealing with air quality because it seems too massive and far beyond our ability to affect as individuals. On their own, people tend to have low confidence that others will make a similar effort, believing their efforts will have to be greater if others do not do enough. (We are seeing similar phenomena in the collective response to COVID.)

From our anecdotal evidence, collective and collaborative projects simply sound expensive, time-consuming, complex and risky for funders. Actively and regularly supporting a community and nurturing communication and collaboration between participants on an ongoing basis sounds like it takes a lot of effort and money, and could have uncertain outcomes. We wanted to challenge this assumption.

The PECA Experiment and What We Discovered

Working with dozens of London households across Tower Hamlets, Hackney, Newham and The City, over 4 months in early 2021 we provided participants with a set of 15 PECA Challenges that they could carry out — any number of them and as often as they wished. These challenges are known, at scale, to improve air quality and reduce greenhouse gas emissions. For internal calculations, we converted each of these into a carbon equivalent which on its own is inaccurate and not very useful, but which enabled us to make comparisons between activities and participants of relative impact, scaled effects and perceptual complexity across different challenges.

15 PECA Challenges from ‘3mins shower’ to ‘Go Vegan’

For example, switching off lights all the time is easy but has relatively little impact on AQ at scale, while going vegan is perceptually more difficult but actually could have massive impact if we all made the change. Participants could choose one or more of these challenges (or none) and carry them out as often as they wished, reporting back on their progress each week.

Diagram of Control Group and Experimental Groups

Participants were placed in groups of 8 to 10, with each group designated as either Experimental (using a collective intelligence platform to communicate and collaborate) or Control (no communication).

Each week, participants reported back to us on their progress with their challenges (and whether they wanted to change or drop) and in turn got a weekly summary of their group’s collective efforts. Unlike the Control group, Experimental groups were able to use our custom-designed Pollution Explorers SMS group messaging platform to communicate and share experiences, and also had a monthly zoom call when they could meet and talk directly with others in their group. They could also decide how they would interact and collaborate (or even compete if they felt it useful). We then tracked each group’s results on a weekly basis for the duration of the experiment.

Our analysis showed that the Experimental groups adhered to challenges longer, tended to adopt more impactful challenges, took on a wider variety of challenges and, over time carried out more challenges simultaneously each week than the Control group. While the Experimental groups were more expensive to run on a per-participant basis (taking more management time, incurring more network fees and infrastructure costs), the calculated impact per £ spent they had was almost double that of Control group participants.

16-week graph of average carbon reduction in Control and Experimental Groups showing almost 4 times the total impact by the end of the experiment

One interesting finding was that, in the Experimental groups, when people tried vegan meals and started sharing recipes and reported to each other how delicious they were and how to source ingredients, there tended to be others in the group who would try it out as well, helping increase impact in an almost ‘viral’ way. We had originally assumed that people would change and swap their challenges from time to time, but instead we found that those in Experimental groups tended instead to keep their challenges and just add new ones, again multiplying the impact.

Apart from important findings regarding group dynamics and collective intelligence strategies, PECA also enabled us to compare the relative cost and complexity of a ‘typical’ air quality project, with one that made use of a seemingly more costly and complex ‘collective intelligence’ platform. This shows that for a given AQ project budget, running it as a collective intelligence initiative, getting people to work together, collaborate, communicate and cooperate, can have twice as much impact on air quality, at scale, as a ‘typical’ local authority AQ project.

Collective Intelligence Learnings for AQ and other environmental initiatives

When it comes to air quality (as with other environmental issues) it appears that people generally already know what the problem is, and even know how to resolve it, but there are significant disincentives to collective action:

1. When it’s a massive issue, people feel they cannot as individuals have a significant effect on it
2. They have low confidence that others will make a similar effort
3. Their efforts will have to be greater if others do not make an effort, so why should they bother if others don’t

At the heart of the challenge is how to motivate people to do things that they probably already know would be helpful, at scale, to do, but which they are disincentivised even to start. Through this experiment we have tried to tackle all three of these factors, and to show how providing people in our Experimental groups with means of cooperation and communication while making sense of and responding to the issue at hand (in this case air quality) can diminish the disincentives and increase impact for a given investment in project resources. We drew on research by Dr Gyorgyi Galik and others in developing our approach.

Disincentive 1 — Individuals’ perception of impact

Any time we provided a group with information, they were all given the same information, and it related specifically to group impact, not just individuals. We provided no individual feedback — so as not to cause a sense of competition or even shame among the individuals if they felt their performance was less than others. We also extrapolated from their own group’s performance to a larger scale so that they could understand what impact they would have made if their entire borough had performed at similar/proportional levels, so that they could appreciate how individual actions scale up to something more significant. This was to help them understand that as individuals they do have an effect.

Disincentive 2 — Confidence in others

Rather than requiring experiment group participants to share information we simply provided them with the means to share information — a group SMS platform. This enabled a more informal approach to cooperation and knowledge-sharing than a data entry system, and one which explicitly combined quantitative and qualitative contributions. While this could benefit from further investigation, our observation was that this informality (and allowing them to set their own terms for how and what they shared) seemed to help people get comfortable with each other and develop a robust confidence that others were trying to do things as well and that their efforts were similar.

Disincentive 3 — Impact on Individual effort if others don’t perform

We made this fact explicit — that it would be easier for everyone if everyone worked together — and offered this as a prompt to help others if anyone felt unable to complete a challenge, so that they didn’t feel a sense of shame, and others felt empowered to help. For example, when group communication was introduced we discussed how people could use it to ask others to help them with their challenges, or to swap challenges during a difficult week. This seemed to help people develop a sense of working on a collective goal, and that helping each other in the short term by taking on temporary extra burden was actually beneficial to all in the long term.

Screenshots of SMS messages between participants

Below is a list of strategies that could be applied to future AQ projects or other large scale environmental issues, benefitting from getting people to work together collectively.

1. Dividing up into groups limited to 8 to 10 seems to be useful because it means everyone in a group can get to know each other over the course of a few gatherings. Any larger and individual personalities could dominate.
2. Competition in groups only works for certain types of personalities. Enabling small groups of people to decide for themselves how they want to interact (e.g. collaborating vs competing) is more effective.
3. When people are working in a group, provide feedback for the entire group rather than individuals — so as not to cause a sense of competition or even shame among the individuals if they felt their performance was less than others, which can be disempowering and ineffective.
4. On complex environmental issues, people can feel disempowered by having too much atomised or individual information (unlike tracking your own weight, it can feel there’s no point in knowing how much you’ve affected AQ because your contribution seems so miniscule overall). Providing ‘scaled-up’ information about the impact they would all have if their neighbours and other borough residents did the same as them, seems to be more empowering and exercises the imagination. Even better, knowing what others are doing in conjunction seems to provide more encouragement.
5. People enjoy learning from each other and knowing how others are getting on with their challenges and commitments, as long as the communication is light touch, non-intrusive and does not require constant attention.
6. When people were engaged with what others were doing they also ended up trying seemingly harder (and more impactful) challenges knowing they were not alone (e.g trying vegan recipes)
7. Small incremental steps lead to more persistent change — by getting participants to ease into their challenges slowly (e.g start by doing a challenge once a week or less, then doing it more over a week as they get used to it), helps participants feel less anxious about keeping to their commitments.
8. We had expected some people to get bored of challenges and they were invited to propose their own alternatives, but nobody did, and took on the challenges offered. There is something about the psychology of risk taking that needs further exploration but when people got used to their challenges and knew that others were too, they tended to take on new challenges without dropping the previous ones.
9. Some participants suggested in future creating groups of people who already know each other or share common interests. This would presumably lead to more trust and interaction.
10. It’s possible that a local community ambassador or representative would work better to ‘manage’ the ‘collective intelligence’ platform and inter-group communication and collaboration than an external project organiser, since it would create a different relationship to participants.

By Usman Haque & Ling Tan

Pollution Explorers Collective Action is a project by Umbrellium, funded by a Collective Intelligence grant from Nesta. With thanks to Iram Quraishi & Babar Javed at Loop Labs our community partners, Usamah Khan our data scientist, and Dr Gyorgyi Galik for her research. For more information about Pollution Explorers Collective Action, or applying ‘collective intelligence’ to large-scale participatory projects, please get in touch.

And a special thank you to all the wonderful participants in East London that took part in the project — it was such a pleasure getting to know you a little and seeing how much you could all accomplish (even in the Control group!) — thank you for all your time and dedication!