Joi Ito follows CÃ©sar Hidalgo’s talk on knowledge networks to offer thoughts on how networked knowledge is transforming the lab. When he accepted the job as director of the MIT Media Lab, Joi tells us, Nicholas Negroponte warned him, “Don’t ever assume that you run the lab – don’t try to give orders.” This wasn’t too unfamiliar to Joi – it’s like running an open source project, which is something Joi has done for years. In a case like that, you don’t give orders – you show your biases.
One of Joi’s main biases is as an internet guy. “I think of my life in terms of what I did before and after the internet”. In the early days of the internet, organizations like the ITU held massive, long meetings about standards for networks. Spending lots of time agreeing on standards may make sense in building infrastructures that are hard to change, like railway systems. But in the networking space, the big standard developed by the ITU – X.25 – got trounced by a less planned, but more flexible open standard – TCP/IP.
It’s hard to innovate when you have to ask for permission, when you need government permission to connect a modem to your phone. Moving to a disruptive model of innovation brings the costs of communication and collaboration down, and adopting models like “rough consensus and running code”. David Weinberger’s idea of “Small Pieces Loosely Joined” suggests that we shouldn’t attempt to know and understand the whole of a problem – instead, we benefit by creating things that are small, modular and connectable.
Joi notes that in his work with large Japanese companies he’s often faced situations where it costs more to do a feasibility study than to carry out a project. In large companies, there’s way too much discussion of potential downside, and not enough discussion of upside. The venture capital economy reverses this equation – the downside tends to be fixed and the potential upside is exponentially massive. It used to cost $10 million to launch a startup – it might take $10,000 now. The majority of Joi’s investments are $100,000 in size. At that scale of investment, you’re planning on failing, and you’re trying to make failure cheap – spending $300,000 of time to save a $100,000 investment is a losing bet. You want to amplify the winners and let go of the losers.
He reminds us that 99.9% of open source projects fail. You’d never fund a project like Wikipedia, even though it costs very little to try it. Innovation is simply different when the cost of failure is low – it can be easier to adapt a project to a new purpose than to start over. Paypal started as a mobile ap, and YouTube as a dating site with video. Joi urges us to embrace serendipity – “If you plan everything, you can’t get lucky, and you really need to get lucky.” The Media Lab may look random, he tells us, but it’s a serendipity engine.
As an example of how networked knowledge might work, Joi talks about his work with Safecast, a networked response to the Japanese earthquake and Tsunami and the Fukushima crisis. As Joi was interviewing at the Media Lab, the disaster unfolded, and Joi found himself at the center of an international network that involved academics, nuclear scientists, hardware designers, data visualization experts and others. Collectively, they’ve developed low-cost geiger counters which are mounted on cars and driven around Japan collecting massive sets of data. They can now demonstrate that there’s more radiation in areas outside the exclusion zone than within it, raising complicated questions about the political decisions around moving people from the areas near the reactor.
There are some important lessons learned from the project. Open data matters – Joi’s team publishes all their data under CC-0, meaning it can be very widely shared. He declares a bias for interoperable data. The work on geiger counters is evidence of the importance of open hardware and open design. )It’s also a lesson in the creative power of crises – there was a wave of innovation around geiger counters right after Three Mile Island and around Chernobyl.)
Joi’s vision for the Lab is rooted, in part, in projects like Safecast. The problems they took on were unsolvable without building a vast network filled with various types of expertise. If the Media Lab is going to open itself to learning from networks, Joi believes we need to move from being a “container” to being a “platform”. We need to be suspicious of our tendencies to look within our own walls for solutions and to look for better ways to work with people outside our ordinary orbit. We’re taking some steps – a Media Lab blog, the use of creative commons licenses to publish that blog, work on open data policies, and a new commission to consider our IP policies, looking for ways to be more open and cooperative.