This presentation was for the Crisis Communications conference in Brisbane in March 2011 sponsored by the Eidos Institute and Queensland University of Technology. The presentation covers the use of Ushahidi and Crowdmap during the Queensland floods in January 2011 for crowdsourced reports by the Australian Broadcasting Corporation.
Purpose<br />Purely crowdsourcing<br />Real time data layers<br />‘Supplemented’ Crowdsourcing<br />
Editorial Challenges<br />Verified vs Unverified events<br />Trusted sources attribution<br />Publishing personal details<br />Business names and references<br />Requests for help and information<br />
Next steps<br />Broader strategy for emergency digital services<br />Integrated mapping strategy<br />Collaboration and partnerships<br />Sourcing data in suitable ‘map friendly’ formats<br />Building culture of participation<br />
My presentation today is about a crowdsourcing mapping trial which the ABC did during the Queensland flood using a platform called Ushahidi. Ushahidi is an open source crowdsourcing platform that has been used around the world during crises and emergencies during the past few years to gather public reports and display them on a map based interface.While not strictly social media it was quite a new and different way for the ABC to engage with and seek input from our audiences during a crisis.I work for a team called strategic development team in ABC Innovation and part of our brief is to explore new platforms and technologies and look at how they might be of value across the ABC and to our audiences. We became interested in Ushahidi as a platform because it seemed to bring together a number of trends emerging in media including citizen journalism, hyperlocal information and real time mapping. I’d like to start off by giving you a bit of background on the technology and platform and then talk about our experience in using it during the Queensland floods and discuss some of the opportunities and challenges we felt is presented.
I’d just like to show you this short 2 minute video which I think explains the origin and aims of the platform quite wellUshahidi means ‘testimony’ in Swahili and it was orignally developed in Kenya in 2008by citizen journalists and web developers who were looking for a way to share information during the fall out after a contested election there where there were widespread allegations of fraud and corruption. It allowed ordinary citizens to report via SMS, email, twitter or online reports of electoral fraud and the violence that followed the election. From there it developed as a free and open source platform and has been developed around the world to cover a range of events.
Since then the platform has been used to cover a lot of major global events and crises such as the Haiti earthquake, Gulf oil spill, the floods in Pakistan and more recently the earthquake in Christchurch and the Japanese earthquake and tsunami.One of the real strengths of the platform is it’s accessibility as you only need a phone with SMS to submit a report.
We started talk to the team at Ushahidi in early 2010 and also to teams from other media organisations who were trialling the platform for covering events. Al Jazeera were the first large media organisation to use it during the conflict in Gaza in 2009 when it was difficult to get journalists in and out of the region and reports from on the ground were very scarce. The BBC did a few trials during the spate of Tube strikes in London last year where people could report from their Tube station whether trains were running or not.
The Guardian did a trial during the Pope’s visit to the UK where there were a number of miraculous sightings reported including one of Jesus’s face appearing in a slice of yellow pepper in HullAt the end of the last year we were considering whether we could use this platform to crowdsource public reports during a crisis during the Australian summer. There was however a lot of concern about whether this was a role that the ABC could play given were are also the official emergency broadcaster. There were concerns that it would be too risky to publish unverified reports from the audience particularly in a situation like Black Saturday where the event was so fast moving and changing to quickly. However we also felt there was an increasing demand and expectation from our audiences for more near to real time and on the ground reports.
What we ended up deciding to do was to run a non emergency trial of the platform in January and collect reports of feral animal sightings from across Australia for the month of January. We called the trial Feral Month. We actually got quite an amazing response to this project too with over 1600 reports being submitted over the month.Anyway I came back to work distubingly early in January to launch this site on the 3rd of January. On the 4 of January when it became apparent that the floods were going to be of a scale we hadn’t seen for some time Local Radio came to us and asked to launch an implementation for the floods. Luckily we were able to get the site up pretty quickly using a hosted version of the platform called Crowdmap and it took a few days to get it up and running.
I’ll wanted to spend a few minutes now showing you the map and how it works for those who haven’t seen it. The way it works is you set up a series of categories which you can see here on the right hand side and when reports are submitted they are added to one or more of these categories. So you can see here some of the categories we used were Property Damage, Road Affected, Hazards, Evacuations, Electricity Outages. Then coming down we have Help/Services which you can drill down into to see Relief Centres/Emergency Accomodation, Hospitals, Police stations and other Help/Services. The last three categories we added during the recovery which were Volunteer Efforts, Recovery Assistance Required and Schools Recovery.You can see if I select one of these categories I can drill down and view the actual reports in each category in different regions. So for example if I go into Hazards and Contaminated Drinking Water you can see there are 5 reports in this region which I can drill down to view the individual reports. If I go to Stanthorpe report it saysAs a result of flooding issues at water treatment plants, residents connected to the Stanthorpe, Wallangarra and Killarney water supplies have been requested to boil any water used for consumption. Each report has a time stamp and a stamp on it to indicate whether it is verified or unverified report. You can see that this report has been verified and is marked as being from a trusted source. You can also see on the smaller map reports from surrounding areas which are mainly road closure information for Stanthorpe
This is an example of the form used to submit a report if you were submitting it online. You put in a title and description, specify the location and select the relevant categories for your report. You can also add your contact details which aren’t published on the map. Most of our reports came in as online reports with a smaller proportion coming in via SMS, email and twitter. All of the reports go into an admin interface at the back in and need to be moderated and approved before the are published to the map.
During the crisis we had approximately 1500 reports submitted to map. About a third of these so about 500 were from the public and about 1000 were added by ABC staff and moderators.Just to give you an example of some of the typical report coming in from the public this one is during the crisis which is an unverified report from Goranda which says ‘About 48 hours with two feet of water everywhere. Lost a lot of books and furniture and the carpets are a ruin.Power off for about 48 hours.The road out the front is a river up to 2M deep.’
This table just gives you an overview of the types of reports coming in during the crisis to the map from the public including power outages, property damage, road and bridge closures where to get fuel and photos of the floodsBefore we launched ABC staff had added police stations and hospitals and supplemented the public reports by adding locations for evacuation centres, sandbag locations, water contamination reports and Bureau of Meteorology weather advice
The nature reports and information of course changed very quickly during the crisis and recovery and we had to be able to response quickly to this with the map. This report during the recovery is a request for volunteers to help clean up Brassall State School from the principle asking people to bring rakes, shoves, spades, wheel barrows down for the clean up.
During the recovery period reports from the public were being added about emergency accommodation, volunteers wanted and available, clean up supplies available, food and drink for volunteers and lost and found pets.From the ABC side we were adding details of Centrelink recovery centres, Community recovery centres, free vaccinations, bulk bin locations, tipping stations, school clean ups and emergency pet accommodation
Over the 24 days of the crisis and recover we had a lot of traffic to the site with 230,000 unique visitors and 1,500 reports. We felt it reflected the large demand during the crisis for real time data and mapped data from the audience. We found there were a number of audiences for this map people affected in the region but also family and friends watching events from afar and the wider general public.
Before I finish off I just wanted to talk a bit about the purpose and role of the map and some of the editorial issues that we faced during the trial. We talked a lot during the trial about the purpose of the map. There were some people in the team that felt we should be just focusing purely on a crowdsourcing approach where we were only displaying reports from the public. Others felt we should be trying to incorporate more layers to create a more comprehensive data map that showed road closures, emergency services reports, innundation and weather patterns. In the end I think the model we used would have to be called supplemented crowdsourcing where we used the resources of the ABC to call out for reports from the public but also supplemented this heavily with reports being added by ABC staff and moderators. In a way the different information could complement each other by being displayed side by side on the same map. You could see the official reports and well as the public reports alongside each other for your area.
Just to go onto some of the editorial challenges which I think probably relate quite closely to some of the challenges of social media more generally during emergencies One of our main challenges which I mentioned at the beginning was clearly indicating which reports were verified and unverified. Originally we were thinking about displaying these reports on different layers on the map with verified reports from the ABC and emergency services on one layer and publicly contributed reports in a separate layer. We tried to verify audience reports where we could. In general the feedback we had overall from the trial was that people were able to identify pretty easily the status of the different reports and treated them differently accordingly.We did realise the importance of attribution for trusted sources and were interested in looking at ways were reports from trusted sources could be automatically be published to the map.We found particularly during the recovery stage we had to publish some personal details of people wanting and offering help to facilitate communication during this critical time. Generally we would be very cautious about allowing people to publish their personal details on ABC online however in this case it was justified and in fact essential for the reports to be useful. We later removed personal details but left people’s twitter handles. Similarly with business names where we wouldn’t usually under our editorial guidelines be able to publish business names and addresses as it could be seen as an endorsement we did allow reports from businesses offering supplies and support on the map.And lastly we had to be very careful to try to explain the purpose of the map so that people didn’t think we could assist directly during the crisis but that we could quickly pass on the details of where people could get help.
The trial we felt was very successful is showing us the strengths and weaknesses of this platform and this approach to crowdsourcing public reports during an emergency. We did realise though that because it was so resource intensive it is probably something that we could only role out for quite significant crisis events.We wanted to look at it in terms of a broader cross divisional strategy for emergency services coverage and develop a more integrated strategy for mapping crisis events. We will also be looking soon at mobile services for emergency coverage to be integrated into our current mobile offerings.We realised that in order to successfully provide the public with the near to real time mapping information they seem to want we will need to work closely in partnership with other organisations and agencies. During the trial we were adding a lot of the reports from agencies manually but it would be much better if we could use structured data and metadata to aggregate reports in one place for people. These official reports can then be complemented by reports from the public in the one location.Lastly I think it’s important to continue to build on this model of participation and acknowledge the value of it both within the ABC and more widely in the community.