Recently we got a chance to interview Erik Hersman. He is the co-founder of Ushahidi, a web application created to map the reported incidents of violence happening during the post-election crisis in Kenya.
For those who don’t know you. Could you please introduce yourself?
Certainly. I grew up in Africa, Sudan and Kenya to be exact, and I live here in Nairobi with my family. I’m intrigued by the way technology helps us overcome inefficiencies in the system, of which we have our fair share on the continent. This led me to start blogging at WhiteAfrican and AfriGadget many years ago, and it was the driver for me co-founding Ushahidi and building the iHub this year.
During the Kenya elections in 2008 you decided to set up a crowdsourcing project called Ushahidi. What is this exactly and what made you create it?
Ushahidi was a reaction by an ad hoc group of African bloggers and technologists to gather information from ordinary people during a very tense and crazy time. There were media blackouts and intimidation, inability to get to places where activities were happening, and many things being swept under the rug. We built Ushahidi over a couple days as a simple way for anyone to send in messages of what was going on around them via SMS, email or web form.
We didn’t think much of it at the time. In fact, it was so plain and simple that we were a little embarrassed by the attention it garnered. Finally, we realized that what we had done was not about the technology, but about the way it allowed information to flow up from ordinary people in an emergency – at odds with the top-down information from government, media and international NGOs that we usually see. With the crowd included, we had a more holistic view of the events.
There are still many different views and opinions on what crowdsourcing is all about. Could you give us your thoughts on crowdsourcing?
To me crowdsourcing is simple: it’s gathering information from the crowd or using the crowd to solve a problem. Nothing magical about it, except for the way you do it and what you do with the information. Providing simple visual tools for quick understanding of what is happening is a big part of it, especially when a lot of data is collected and the noise starts to overcome the signal.
When companies or organizations would start to think about doing something with crowdsourcing. What are the do’s and don’ts you’d tell them?
Our perspective on crowdsourcing is that only about 10% of your equation should be the technology. The remaining 90% is split between managing the crowd, messaging to them and providing an incentive for them to participate. or us, that comes in many different forms, as the Ushahidi platform is used all over the world for a wide variety of crowdsourcing needs.
We’ve had hundreds of deployments of the platform, here are some examples:
- Haiti – disaster relief (effectected population + diaspora + volunteers + relief organizations);
- Washington DC – blizzard coordination (Washington Post + DC community);
- Italy – Open forest fire (public + volunteer fire fighters);
- Tanzania – Election monitoring (public + election monitors);
- England – Mapping the London tube strike (BBC + public).
When working with crowdsourced data you’d probably face some challenging problems. What are those and how should we deal with them?
There are a couple big challenges.
First, getting people involved and finding the right incentives for them to take part in the crowdsourcing.
Second, verifying information, it takes a while and there are not any great solutions for it yet. We’re building SwiftRiver to help with this second issue, a way for us to use people and algorithms to filter and curate the stream of incoming information.
Third, analyzing real-time information is hard, yet very important. Our timeline and map are good for snapshots, but we need better analytical tools that are easy for everyone to use.
Could you tell us a bit more about the way you analyze the information?
This is a very subjective question, as the way that we analyze the data varies from the way other groups using the Ushahidi platform might analyze their data. We tend to look at the incoming stream of data and bucketing it into urgent/non-urgent queues that we separate and escalate depending on the needs. It differs from deployment to deployment, as the groups who we partner with change. In Haiti it was very much tied in with local humanitarian responders, and in the Kenyan Referendum and Tanzanian elections this year, it was tied in with the local security and electoral commission.
Inbuilt into the system, on the front-end is the map and timeline. Since the very beginning we’ve thought that pairing space with time gives some great insights into real-time information. What we need to come up with is some additional analytical tools, visual most likely, for everyday front-end users to have access to.
If I’d like to start collecting crowdsourced data. Where would I start? What sources and systems should I use?
This largely depends on what data you’re looking for and your goals for crowdsourcing it.
Generally speaking, if it’s around what we term a “hot flash” emergency, then your best bets are to go with the channels that everyone is already used to using in your country/community. Usually that’s Facebook and Twitter in the US, and both have great APIs with which to support gathering that data quickly.
If you’re setting up to crowdsource something over time, we suggest you set up a specific site and use multiple input channels for it. In our parlance, this is a “slow burn” initiative, gathering crowd information over the long-term. Unlike the “hot flash” data, which people are giving out regardless, here you have to provide clear incentives to them if you want serious engagement. Again, in the US, Twitter and Facebook are still great input channel options, but I’d also look at SMS, email and webforms to spread it wider.
If you could make any crowdsource dream come true… what would you love to create using crowdsourcing?
Definitely a better Zombie map! We must get our act together, more Zombie reports and sightings are critical if we’re to survive the coming onslaught.
I think we’re just on the edge of the way technology makes crowdsourcing so compelling. Watch Carlo Ratti’s talk/Assaf Biderman’s – if there is a way for the ‘internet of things’ to feed into crowdsourcing platforms like ours, then we could have a real-time view of key stats in a city. Take the example of Copenhagen. There will be more things coming down the pipe that do upload real-time data in a usable and open way, which is a big deal.
What does the future look like when you have crowdsourced possibilities in mind? And what role do interaction designers play in this?
It looks like a lot more data. A lot more noise if we’re not careful.
Interaction designers are key to making it easier for the crowd to play ball, and more importantly, helping us all make better sense of what’s being collected. Meaning trumps quantity here, especially in a real-time world where time to understanding and acting are critical.
In February you’ll be giving a talk at Interaction 11. Could you give us a bit of a sneak peek? What will the talk be about?
I won’t be talking to much about Ushahidi. Instead I want to touch on some of the constraints and cultural difference in Africa that make designing and launching new services here so different than their counterparts in the rest of the world. It’ll be about how we see the world differently, making it hard for people with Western assumptions of what the internet and phones are to design for Africa.
If you want to meet Erik Hersman in real life: he is one of the keynote speakers at Interaction 11. It is the fourth annual conference hosted by the Interaction Design Association (IxDA). Each year, IxDA aims to gather the interaction design community to connect, educate, and inspire each other. This year it is held in Boulder, Colorado (USA).