Wired Magazine's Jeff Howe talks about Crowdsourcing & TopCoder
TopCoder had a chance recently to talk with Wired contributing editor Jeff Howe, author of the best-selling book “Crowdsourcing: Why the Power of the Crowd is Driving the Future of Business” and credited with first coining the term ‘crowdsourcing’.
Q: What motivates the crowd to participate in a project or piece of work?
The first thing I’d say is ‘motivations’ is an area in dire need of more research, but what we know is that while it varies from project to project, it’s a complex tangle of all sorts of motivations – money being just one of them and often not the most important. In fact, some research has shown that money can act as a dis-incentive in that if you offered your mother-in-law money for cooking Thanksgiving dinner, you probably wouldn’t be invited back. Offering money to people who are using Mechanical Turk to search for Steve Fossett’s plane wreck wouldn’t work. There are things that people do online because they love to do it and they feel that they are doing the world good. And then there are things that they do because they want to make a little extra cash, but they are also doing it for non-monetary reasons.
Karim Lakhani from Harvard Business School did a study and he asked contributors to an open source software project if there were 25 hours to a day, what would they do with their extra hour and 88 percent said they would code. So, some people just want to do what they do. I certainly relate to this as a writer, in part it is laborious and can be a chore, but it’s something I truly love doing, and if I had an extra hour to work on say, fiction or short stories, I would probably take it. And I’m not unusual, I’m not an aberration. I think I’m the norm. We all have things that we enjoy and it could be making music, it could be coding. People are passionate about the products that they have in their lives – I know as parents whether it be a stroller or even a baby bottle we’re quite passionate about what’s wrong and what’s right with it and if we were given an opportunity to redesign it we would have a lot to say.
So, there’s a lot of different reasons that people participate and if there’s one way I have of summing it up I have a saying – that passion is the currency of the 21st Century because the one common theme I see is passion and it could be a passion to practice something you’re good at, it could be a passion to teach people what you already know, it could be a passion to make the world a better place but it’s not that simple ‘carrot and stick’ that we’re used to in a command/control managerial hierarchy.
Q: We’ve seen the substantial success of crowdsourcing when applied to more straight-forward tasks. How does crowdsourcing fare when applied to more difficult problems like building technologies?
Well, I think it’s a mixed bag. I don’t think that there’s any doubt that crowdsourcing is a model that, at least so far, has worked best on tasks that cannot be easily confused. I think that there’s a lot of communication breakdown when you’re asking people to do things – it’s not so much ‘complexity’ as it is multiple steps. I feel we can use TopCoder as an example of how to use a community to create a complex piece of software but a) there’s a lot of built-in modularity going on in that the software itself has already been broken down into modules, and b) the process by which the community codes new work is also broken down into modules so that at any given time, the crowd is working on a relatively simple task, and there’s no question that they need TopCoder to be acting as what in an open source project would be a ‘benevolent dictator’. Someone needs to be calling the shots saying ‘OK you guys work on this, OK now there’s a new contest and it’s to test this winning code to see who can break this winning code’, – someone needs to be setting these sorts of tasks. It’s not so much that the crowd can’t perform complex tasks, but it is the case that someone needs to be running the show; someone needs to be breaking down the problem.
You know, Yochai Benkler writes a lot about modularity in “The Wealth of Networks” which is far more a Bible on stuff like this than what I’ve written, (it’s longer and more theoretical by far), but it’s hard to understate the significance of modularity in crowdsourcing – simply stated, the tasks need to be broken down. And there are a couple of reasons for this and one is that that allows you to communicate them simply. It’s not that people are stupid, as I say in my book, it’s that they’re busy, and their lives are already complex. So by breaking it down into something simple it leaves little room for miscommunication.
And then the second thing, and this is even more important, crowdsourcing is leveraging what the writer Clay Shirky calls ‘cognitive surplus’, and what I call in my geeky Wired way ‘spare cycles’. The same way SETI@home taps into people’s spare cycles on their computer to search for extraterrestrial life, crowdsourcing projects are leveraging people’s spare cycles, their cognitive spare cycles – that hour before bedtime, that 15 minutes of a coffee break. So if a task can’t be completed in five to ten minutes, it’s not that it won’t work in crowdsourcing, it’s just much more of an onus on contributors – so a lot fewer will participate. I think that we’re learning though, that that’s not always a bad thing. One model of crowdsourcing I’m very interested in, and TopCoder exemplifies this, is what the system theorist Scott Page calls a crowd of models, which is really just a crowd of experts.
Q: Competition is at the core of TopCoder’s model – is that unique in the world of crowdsourcing? Does crowdsourcing lend itself to a competitive model?
I absolutely think it does. I think that TopCoder is fairly unique in that it was set up from the get go, with the idea of what Ned Gulley at MATLAB called ‘competitive collaboration’, in that people are competing – Jack’s (Hughes) initial vision, as I understand it, was like these baseball cards for coders, which I love. Being a geek, I think it’s totally great. Look at something even as simple as Google’s Image Labeler, where people go in and are essentially competing to score points on tagging photos the same way that other people would tag a photo, so you get points for building consensus and what do these points translate into? Absolutely nothing. I mean the points don’t do anything but people like getting points – this is what crowd-sourcing has revealed about humanity – is that people like points.
Here’s an off-the-cuff prediction: In the next five to ten years, we will see in the same way that people have wanted to capture the kinetic energy produced when people exercise in gyms, on treadmills or rowing machines for example, they’re going to create online video games that somehow are accomplishing some simple task. And people playing the games won’t really notice that they’re actually doing work – the thing that they’re doing for that video game, whatever the ultimate aim for the game is, will wind up putting dollars in someone’s pocket.
Q: What indicators have you discovered during the writing of your book of what the future might hold?
One thing working on this book did not reveal was that there would be a credit crisis of unprecedented proportion – and that’s just to say that the more I know about a subject the less willing I am to play prognosticator, because I think it’s a chump’s game. But I will say there’s no question in my mind that crowdsourcing, or whatever you want to call it, (I never once felt that that term is somehow tied to the success of the phenomena itself – it would have happened without calling it ‘crowdsourcing’) – will only increase. And that’s because the developments that made it possible, indeed inevitable, have only begun to make themselves felt. We can call it the network effect; it’s a network effect on mankind, on humanity. And all that network effect is just increasing. Broadband penetration is increasing. Education levels, however slowly, are increasing. The costs of the tools of production are decreasing; they’re getting better, faster and cheaper.
So, all the contributing factors to crowdsourcing are increasing in strength, robustness and pervasiveness and more and more labor is going to shift into this network environment where communities are competing with corporations or even supplanting them. So that I do know, but how that shakes out, I’m not so sure. Here’s another thing that we can happily predict – it’s that a lot of, if not probably most, initial crowdsourcing ventures will fail. One thing that amazes me is that people will email me and they say this company is essentially going under so does this mean the crowdsourcing model doesn’t work? I’m like what, are you crazy? All we need to know about crowdsourcing is present in Wikipedia, more or less. Can crowdsourcing work? Wikipedia is your answer. Now, will it work under every condition? Absolutely not. There are lots of things people won’t want to do, or people won’t get the incentive structure right. One thing about crowdsourcing is that it’s really difficult to make it work right.
Look at TopCoder. I mean, TopCoder expended an enormous amount of energy in the set up process. There was a great deal of patience, years of patience expended in slowly building a community. TopCoder, frankly, is one of my models I use when I talk to people – you want to know how to build a community? Look at what these guys did. They spent years and they served the community first – what will the community think is fun? What do our people like to do? Our people like to compete. Our people like to see who’s the best coder. They like to play games with each other. There’s a reason that TopCoder has succeeded and you can look at the inverse of that and say well there’s a reason that a lot of companies have failed – because they’re not going to put in that kind of time.