While the notion of the wisdom of the crowd has popular resonance, government should take care to think about what it wants to achieve when engaging the crowd.
This is the second in a series of three posts based on a talk I gave at the Institute for Government (IfG) at their Crowdsourcing Policy event, 23rd January 2012. The first drew on our work on public engagement generally to make a simple point: the starting point for government interested in engaging the public in policy making should be to let the purpose of the public engagement dictate the process used. Government should not let the excitement about using something funky like crowdsourcing drive everything else – just as it would be a mistake to let the fact that you’ve seen citizens’ juries, for example, work before lead you to running lots more in different circumstances.
The third post explores some wider principles government should bear in mind before even thinking about engaging the crowd.
In this second post, I want to explore four uses government could put crowdsourcing to. This post draws on the ways that crowdsourcing has been used in general, and explores their relevance to the policy world.
The first problem amendable to crowdsourcing taps into popular notions of the wisdom of the crowd. This is based on the premise that, while no-one knows the exact answer, the crowd will get the answer right together.
In my talk at the Institute for Government I illustrated this by inviting 100 or so people at the event to guess how many beans there were in the bag of beans pictured at the top of this post. Each person was given a bean to help them. You’d expect the distribution of answers to fall on a basic bell curve (the bench pictured below roughly follows this shape). If crowds really are wise, you’d predict that the correct answer will be the mean of the total number of guesses.
The bigger the crowd, the more likely it is that the mean will represent the true number of beans in the bag. The rise of the internet and social networking has made it a much easier to increase the size of the crowd. It would have been relatively simple for me to open up my call for guesses to my twitter followers in order to increase the size of my crowd.
This is, in the form of the jelly bean example, a fairly trivial use for crowdsourcing. However, I’m finding it hard to think of use for government that it couldn’t do much better with scientific, economic or other technical advice rather than resorting to a glorified guessing game with the public.
My second use is one that I think is of more relevance, however.
It would be more time consuming to sort the beans in order to work out the distribution of colours in the bag. Given that everyone in the audience had a bean, it would be fairly easy to quickly collect information about the distribution of different colours. Whether it was worth doing or not would depend on whether the system for collecting the information from each individual was quicker than having one person separating the whole bag. Again, the larger the crowd the more accurate the information.
This is something that has been used in the policy world. One example that caught my eye a couple of years ago was initiated after typhoon Ondoy devastated the Philippine capital Manila. A geology professor, Alfredo Mahar Lagmay, put out a call on Facebook for data on the maximum flood height at different locations; the crowdsourced flood prediction map was born.
Any bean manufacturer worth their salt will want to know which beans are most popular so that that this colour can be increased in proportion to the least popular colours. It would be a very simple task to ask the crowd what their favourite colour is and then working out if the proportions of bean colour match the results of the research. Again, the crowd would obviously have to be big enough to ensure that the results are statistically robust – and care would need to be taken that there isn’t some reason why green lovers are more likely to respond than blue lovers, but these are simple enough problems to overcome.
You can see how this could be useful for understanding the best times to open a doctors’ surgery, for example. Put a call out to the population living in the catchment area of the surgery and you’d expect to get a good idea of how best to arrange the opening times of the surgery. Of course, you could do this through a simple survey. Developing a crowdsourcing platform online would allow you to automate the process of data collection, but more importantly, it would allow participants to interact with, and learn from each other thus potentially improving the quality of the final answers.
During my IfG talk, I proposed one final use for crowdsourcing policy.
In Use #2, I noted that crowdsourcing can be effective at uncovering preferences. However, it’s possible that there was someone at the back of the room with their eyes closed who was bored of the conversation, it was clear to them that what we need is a new colour of bean, that those of us operating on a less spiritual plane haven’t thought of which will really differentiate my pack of beans in the market. With this new colour we’ll clean-up.
If I set the process up right and ask enough people, that killer idea that has occurred to the 1 in a million people will be surfaced.
This notion of policy innovation was one of the key motivations behind the government’s Spending Challenge in 2010. On the basis that Whitehall is unable to spot government waste on the ‘frontline’, George Osborne invited civil servants and members of the public to identify where public money could be saved. The Challenge clearly highlighted that there is a real appetite within the public to get more involved, as over 100,000 people responded.
However, getting the crowd to turn-up is only part of the problem. As I highlight in my final post in this series (link to come tomorrow), there are a number of principles that government needs to bear in mind before it asks citizens to give up their time and energy engaging on policy issues.
Bell curve bench: c r i s
Philippines Crowdsourced Floodmap: Screenshot from 13 January