Lessons learned from Volunteers Interactions with Geographic Citizen Science – Morning session

On the 27th April, UCL hosted a workshop on the “Lessons learned from Volunteers Interactions with Geographic Citizen Science“. The workshop description was as follows:

“A decade ago, in 2007, Michael Goodchild defined volunteered geographic information (VGI) as ‘the widespread engagement of large numbers of private citizens, often with little in the way of formal qualifications, in the creation of geo­graphic information, a function that for centuries has been reserved to official agencies.’ (p.2). The collection and use of this type of crowdsourced geographic data have grown rapidly with amateurs mapping the earth’s surface for all kind of purposes (e.g. collecting and disseminating information about accessibility in urban centres, for crisis and emergency response purposes, mapping illegal logging in remote areas and so on). A subset of these activities has been described as ‘geographic citizen science’ and includes scientific activities in which amateur scientists (volunteers) participate in geographic data collection, analysis and dissemination within the context of a scientific project (Haklay, 2013) or simply by using scientific methods and equipment. Although, there is an extensive discussion in the VGI and geographic citizen science literature about opportunities as well as implications (e.g. data coverage, data quality and trust issues, motivation and retainment of volunteers and so on), examples from the actual interaction are not so widely discussed, neither has evidence been collected from a broad spectrum of case studies to demonstrate how volunteers interact with those technologies and applications, what they are looking for and what it is that they need/try to accomplish (at a scientific, project and personal level) and what are the common design mistakes that influence interaction.” The following is a summary of the talk and presentations:

Welcome & Instructions – Artemis Skarlatidou the workshop is linked to our ERC funded project Intelligent Maps (ECSAnVis) and  EU funded Doing It Together science (DITOs) and the COST action – our work deal with geographical applications of citizen science and data collection. There is the COST Action CA15212 which got 243 members in 39 countries – all exploring aspects of citizen science – Work Group 1 (WG1) for scientific quality, WG2 education, WG3 society-science policy, WG4 the role of volunteers in citizen science, WG5 data and interoperability, and the synergies in WG6. In WG4, which Artemis lead. we’re looking at stakeholder mapping, motivation, needs and interaction issues, and mapping citizen science across Europe. Another relevant group is the ICA Commission on use user and usability issues, the International Society for Photogrammetry & remote sensing that have a WG V/3 that look at citizen science and crowdsourced information. Sultan Kocaman explained the ISPRS link – WG V/3 focus on the promotion of regional collaboration in citizen science and geospatial technologies within the focus of ISPRS area of education and outreach.

Louis Liebenberg presents Smartphone Icon User Interface design for Oralate Trackers – Louis Liebenberg who for 3 decades have been developing software to allow hunter gatherer to protect their knowledge of tracking. One of the challenges that Louis address is the understanding how our scientific thinking evolved. Louis suggests that tracking is an example for hypothesis testing and rational thinking that evolved in in tracking by hunter gatherers. He worked with !Nate from the San people since 1985 – the context of technology use by San for a long time. Already 100 years ago, hunters discovered that arrow points can be made from fence wire and started using them. This is an example of how hunter-gatherers adopt to technologies around them. Hunter-gatherers are not isolated: they always interacted and traded. Developing a software for a smartphone (you can get an Android phone for $10 in South Africa today), is similar to adopting the fence wire for the arrows 100 years ago. He learned from master trackers – the level of sophistication of trackers is astonished him since the mid 1980s. In the Kalahari, dogs were introduced in the 60s, and therefore the knowledge of tracking and the practices of hunting change. He used tracking and certification in it in order to secure employment. Master trackers are expected in an egalitarian society to show humility, so it is possible to miss them if you go and ask “who’s the best tracker here?” – the certification is a way to provide recognition and work. The tracking provided employment in the 1990s in surveying the movement of animals in the Kalahari. The persistent hunt – when you do it without any equipment, running animals down until they die from exhaustion which is an adaptation that humans have to be able to do that. Karoha was one of the persistence hunters but also able to use CyberTracker and use the system. Parallel to the software, Louis develop the tracker certification, to know if the data is reliable. As Master Trackers die, the knowledge is lost, so the certification provides an opportunity to encourage the younger generation to develop the knowledge and benefit from it. The level of details in animal tracks is very high. There is a high level of ambiguity in tracking and requirement to learn about claw marks and knowing what are the possibilities then it is possible with high certainty to understand which animal it was. Trackers also develop hypotheses on why the shape of hoofs is the way it is, and interpret activities of animals from the track – for example, identifying new ways of interpreting the behaviour of an animal that was not observer before. For example, the ability to guess that caracals are jumping upright in an attempt to catch a bird. CyberTracker started with the early Apple Newton with a GPS module, and then evolved into the Palm Pilot and continue to evolve. The interface was very limited in drawing icons – icons are either phonetic symbols (e.g. using a wheelbarrow to describe an item that sounds similar to the word in Africans). The details can be very extensive – species, age, number, male/female and so home. The data can provide information on abundance and potential of work are the communities. In a project in the Congo, they follow the trackers of different animals and they could show they Ebola impact Chimpanzees, Gorilla, but also other animals and then this was important to understand that you can identify Ebola in wildlife before it spreads into the human population. There is also a wide use of CyberTracker in citizen science on monitoring endangered species, and different projects by indigenous communities  Australia. They can also show that there are different results from what ecologists identify. A paper from 1999 about Rhino was co-authored by a tracker, demonstrating different models of publishing with citizen scientists. The first high impact that was co-authored by trackers was published recently in biological conservation. Questions: how to communicate from hypothesis by hunter-gatherers to the scientific sphere? The need is collaboration: data collected and organised by the trackers, and then the scientists write the report, but providing a report is challenging. The reality is co-authoring as there is always need for mentoring, reciprocal approach between scientists. Louis also circulates papers with experienced scientists to improve the paper. We all need peer review support. In terms of consent and engagement: there is a need to develop the relationship of trust and understanding – the first people who were involved in CyberTracker worked with Louis for 5 years, and Louis engaged as a tracker before they were willing to work with him. Some of the early papers in the Kalahari used trackers without mentioning their name even though the trackers carried out the research. Scientific institutions are one of the last authoritarians institutions – citizen science. Scientific elitism is intransigent and this makes citizen science exciting.

Lessons from supporting non-literate forest communities in the Congo-Basin to record their Traditional Ecological Knowledge – Michalis Vitos & Julia Altenbuchner the context of the Congo-basin is the second largest rainforest. This is a forest with 29 million people, with at least 500,000 nomadic communities that rely on resources. The forest is divided into concessions and then they are sued for resource extraction – how to make local groups heard? Local communities are excluded from protected areas. In the last few years, some legislation is changing – e.g. the FLEGT of the EU to control timber import and request for social payback and responsibility. ExCiteS collaborated with communities to support such process with technology. The challenges are dealing with non-literate groups who are also non-technologically literate. We use pictures as a way to communicate: the application working in a simple fashion – showing categories of things that people want to map, each category is leading to more specific options – the information can be captured and deciding if we want to save information and we can collect video and audio that are geotagged. In 3 simple steps, information can be captured. The process starts with a dialogue of what important for the communities, and then with this agreement on what will be collected. We do explore the usability of the application. About 70% can use the application, but 30% have a problem with categories – you follow a path of mapping banana, avocado and cacao – this requires categories, e.g. one of the set. Some participants found that confusing. Adding more icon to the category is becoming more complex. One approach was to test audio feedback in a local language – explaining the icons and what they mean. The experiments with the audio feedback help a bit, but not a lot. The next step was to go directly to the final icons and go directly to the final card – adding an NFC chip and adding the control to it. Participant finds the specific icon and then touch the card with the phone. With Tap&Map the success rate gets close to 100%.

Julia – the next issue is making sure that communities can manage their data- the vision is of intelligent maps  – having data collection, then local data repository and management, and then visualisation. But there is a challenge of the mapping and this was done by using UAVs and creating within a short time a high-resolution imagery. However, people don’t need maps as they know their area, but the maps are for communication. The maps are being used to check how the map is used – people felt under a lot of pressure when using the map. and the next experiment was not to put under pressure, and instead of doing a treasure hunt: going and looking for data by trying to find German Christmas decorations. The tracks of the people who participated in the study we can see how they looked for information. What we know is that people can use maps and understand them – the reference map. Now we want the thematic information – so when people take ownership and correct issues: this was done using the icons that were used as a resource and then to correct information. People were doing well in correcting information using a Tap&Map approach. We get feature corrections over 90%. This an ad-hoc approach: even without much exposure – we need to allow people to be sensors and the brains behind it.

Forest hunter-gatherers and Extreme Citizen Science: Reporting wildlife crime in collaboration with local and indigenous communities in Cameroon through community-led co-design – Simon Hoyte work in Cameroon for the last year and a half with Baka hunter-gatherers. Working in Cameroon in the south-east corner.Working with Dja reserve, working ZSL and 5 communities. In Cameroon, there are many issues with conservation – gorillas, chimpanzees, parrots, pangolins and elephants. Indigenous communities are lots of time are forgotten – those groups are familiar with the forest, with knowledge of 50,000 years and colonial approaches exclude. The technologies that are being used are Sapelli data collection tool, then there is the data management tool GeoKey and the CommunityMaps from Mapping for Change. The process starts with the community free prior informed consent – first starting with the concerns of the community and also building trust by staying overnight in the village and connect on a personal level. That is an important recommendation. Icons are being drawn from the sand, to a paper and then into the app. Functional actions changed from tick to thumbs app, or recording changed. XML layout of the project allow changes in the field. The second recommendation is the co-design that increases motivation. Audio and video are allowing information to be shared, including tracks – it allows a verification. Audio provides more information. Describing what people found. Indicators on the device are important – when recording is active a red icon allows you to see that something is working. The phone is checking for connection every 4 minutes. Using ID screen to recognise reported – can be used elsewhere. The community protocol also addresses who manage who will manage the phone and look after it. The report is upload and shared with the authorities – we need the diverse outcome. So in summary: trust building, co-design, media, feedback, simple tools, anonymous ID, community-led, and diverse outcomes. The map providing further more information.

Community based monitoring of tropical forests using information and communication technology (ICT) – Søren Brofeldt an example for a study that rely on Sapelli and expand the software to create the Prey Lang App: working in Cambodia, in the Prey Lang – 200,00 people who rely on the forest, and huge pressure of deforestation and a lot of the logging is illegal and it is supposed to be protected. The Prey Lang Community Network (PLCN) created around 2005-2007 and it is now a group of 600 people who are doing work over that last 10 years, and patrolling the area, confiscating chainsaw and catch wood and logs. Trying to address logging in the area. 2013 they try to communicate the problem to international society – to do what they wanted to set a forest monitoring programme and create a system to document illegal logging and provide evidence-based advocacy. The issue is to compile information and document breaches. The data is captured by Sapelli, and the information is validated by PLCN and scientists, which then helped in compiling report locally and globally, which then led to the positive platform. The platform was tweaked a bit and include information through a decision tree, they have different aspects. The things that they developed: unique functions – choosing icons or doing activities – they had basic activities in the first version: they have seen it as too simple. They started with 9 basic functions with 614 end-points of activities. By the third version, they had 9 functions, and 1663 options: types of trees, types of information, species and so on. They now have 10 functions (e.g. dropdown, word complete). Complexity does not lead to incorrect use (if training is adequate and added functionality is done in co-designed way). When people are experienced – people who use the app for 2 years can get into more complex functionality over time. Some of the issues with data – poor documents, double counting. over time, human errors are decreasing, and also technical issues. Poor connectivity and technical issue are a major issue – more than local ability to use. High quality is possible with active data management is needed.

Designing Human-Computer Interaction for Citizen Science Initiatives in Rural Developing Regions – Veljko Pejovic & Artemis Skarlatidou we need to understand how we move initiative from developed to developing regions in citizen science application. ICT4D point to environmental constraints: roads, electricity, There are also that this area lack skills in the workforce and cultural constraints. Clashes with assumptions. in the Extreme Citizen Science context: we need to identify solution adaptation in participatory design, there is a need for holistic implementation, and we need to make sure that we think about the whole process – from data collection to policy and this challenging. Finally, we also to consider the champions and engaging then (the book “Geek Heresy” by Toyama talks about it). The aim is to identify guidelines – this was done through participatory studies that are similar in the rural developing world and carried out 9 interviews with researchers with extensive experience in the field. An hour-long interviews x 2. The questions explored different aspects including interactions. The finding – need to mobilise the community by taking into account societal organisation (e.g. egalitarian aspects). Need to find local champions. We need to identify the ecosystem of the technology: chargers, cables. Also need to consider how the technology that was built to a different context work: rough fingertips, reflection in the screens and so on. There is also the issue of using hierarchical icon organisation which is pretty intuitive for educated people but it is challenging for participants (users) and also navigation buttons. This matches evidence from Medhi et al. Chi 2013. Juxtaposing this with illiterate users in urban Brazil, they managed to deal with hierarchical organisation and navigation – might be that the exposure to smartphones helped in developing these hierarchies. Icon design is different, but we can see that realistic icons with context are more suitable to use, not just an object. There are issues of actions and how to represent them. Getting honest feedback on the spot is a challenge – users don’t criticise before (Dell et al CHI 2012 – “yours is better”). Long trust relationship help in getting honest feedback. The participants lack the vocabulary to discuss HCI issues. To maintain motivation, there is a need to make data collection visible and ensure the real-world impact of data collection. Recommendation: develop context-specific apps – not genetic, and consider application interface that matches user’s skills and geographical information is a key.

Introducing user issues of the Global Forest Watch application – Jamie Gibson – developing with Vizzuality better maps and visualisation. Trying to think of citizen-focused GIS, interacting with the citizen in the design. Global Forests Watch (GFW) was developed in the last 3 years, and it is allowing to see the world’s forest and how they change. They wanted to tell a simple story: where forest is gained and lost. With few clicks, you can see the impact of conservation. GFW allow seeing how deforestation is implemented and how it is stopped. There is a need for global engagement – opening it to a whole crowd of people. Forest don’t have a connection to the web, and try to take data online to the field, walk to the area, investigate recent forest loss and report new areas – 4000-5000 users. They aim to integrate citizens into the design process. Forest Watcher is being used in important areas of the world and not where the most connected people area. They analyse where people use the app – when there are forest fires in Spain, people are updating GFW and explore. Use the analytics to find the places where we want more people to look and explore. This is integrated with interviews and usability testing. Working with experts who been working for a long time – including Jane Goodall Institute, Amazon Conservation Team, CAGDF, and BirdLife. As people use the application they build ownership and they provide a better feedback and richer information. In terms of what they learn, including the use of persona to think about monitors: need to have lots of other things that try to sync after the 14 days offline – the internet is slow and changed the app and the back end to make it faster. Use it to understand frustrations and find ways to wow moment. Face, name and story improve the quality of the thinking and understand their frustrations.

Lessons learned from Missing Maps – Jorieke Vyncke Her personal background is in interest in work that links to humanitarian purposes, and since 2017 is the missing maps coordinator. She is looking at the humanitarian organisation focus -more than 34,000 staffers in MSF and about 470 locations around the world. In many parts of the world there are empty maps and not geographical data. They discover OpenStreetMap and working with the American and British Red Cross, HOT and over 40 partners. They have principles from the Ostrom on working with groups. They compare rural and urban parts. In Idjiwi in DRC, the east of Congo – working with a multitude of problems: violence, refugees and more. Due to a measles outbreak, they needed population and mapping data. Included 250 remote volunteers who mapped 28,000 building in about a week. This helped in creating population estimation – critical for the logistical planning. They managed to identify 94% of the population. An example from Bangladesh in Hazaribagh informal settlement. The area was mapped with both local and remote mapping – including factories and tanneries – locating the workers that they wanted to reach – combining students from the university with workers that were reached through the union. The experience of mapping is done by the technical local students to make things happen. Using smartphones and field papers process. Paper is still effective, and then also the edit data in pairs on how to do the mapping – the end result provided an occupational health survey. The process motivated the community and they continue to use it. In different areas, they use remote mapping but the most important thing is to create a local mapping community and that makes a decision between empowerment and remote mapping with the importance of saving life.

Keynote: Approximated Reality: the use of digital tools by traditional communities in the Amazon – Vasco van Roosmalen working in Ecam – Equipe Conservacao Amazonia in Brazil since 1999. The big challenge is how to reconcile different visions of what the world is. In the Xingu area in Brazil, there was a need to create an ethno-map of the region. The community discusses what they want to map and how they want to represent them, but it also needed to be cartographically accurate as this is how you communicate with external bodies. The whole map is created for the community: to use resources, to remember the dead and to defend their land (using patterns of body paint). We can see that protected areas in the Xingu. Another area that he was involved in mapping is near Surinam – in an area the size of Holland with 2000 people, the community recorded information about their region. This helped in justifying the resources and the protection of the area. An area that is very rough to access, and the local survey by the community managed to map the area done that in 6 maps. The community collected much more data than what the map can show – over the coming years, they mapped with different groups millions of hectares and they developed a process of creating the maps. The collaboration with Google Earth Outreach led to the interaction with Chief Amir of the Surui. The link with commitment with Rebecca Moore helped in filling up areas that are missing and attaching video and audio to the map. They then wanted to record illegal logging using mapping tools and this was done with OpenDataKit – the data collection challenges are accuracy, ease of use, speed, etc. In 2008 started to understand REDD and developed the Surui Carbon Project – need a tremendous amount of data from the air and from the ground. The use of information such as the circumference of trees was done with ODK. They use Garmin devices: they weren’t scratch resistance. Now they use a Samsung smartphones that are cheap and can be replaced easily. For the GPS in the rainforest, it is challenging and they use barcode on the trees. They used the ODK build but discovered that it is not an easy interface: using a programmer in the staff and that is a limitation in terms of allowing to build forms easily. The project managed to demonstrate that indigenous people can collect data but the REDD credits were more challenging and they got them in 2013. Cultural maps where created in other indigenous lands in Brazil. The importance not just to demarcate the land but to collect data and help them to manage the area. Today there are many challenges – 13% of the Brazilian territory. In the Brazilian Amazon, there are many communities – 25 mil people of which only 350,00 indigenous for example, Quilombola groups and many other groups. There was no information on other groups and some of them are disadvantaged – e.g. Quilombola required mapping 7000 communities, they are descendent of West African slaves – they were persecuted, faced a lot of violence, and when slavery was abolished they were forgotten, but from the 1980s they are recognised in the constitution, but not enough recognised officially. His team was involved in creating a new map of the 7000 communities for which only on a team of 40 is looking after in the government level in Brasilia. They used approaches that are similar to the Indigenous mapping in order to record information and manage the land. They had people who became experts in mapping and then demonstrating how to map the land using google earth and demonstrating data collection. The communities also collect socio-economic data – using ODK and understanding their community and developing a life plan for the area (plan for the next 10-30 years). The question is who is listening to the information but by whom. A social network analysis of Facebook (which is 83% of users in Brazil use) Looking at interactions show that local association are not linked to environment, human right and there is missing links to health, to a specific campaign on the Belo Monte Power Plant but it is not linked to the community. They care about health, education, income, and only fifth is the environment – need to talk about what matters to communities. How to make conversations about them in the centre of the discussion and move beyond putting them in the corner of the environment. We need to engage with people with their communities in a way that makes sense to them.