Innovation Diffusion in a Network | The Discipline of Innovation

Innovation Diffusion in a Network

Yesterday I talked about how the interconnectedness of our economic networks often makes it more difficult for new ideas to spread. Because our products and services are embedded within a value network, we not only have to get people excited about our innovation, we have to get others within the value network to unconnect from whatever they are currently doing before they can reconnect with us. I thought that an example might help make this process make more sense, and I ran across a great one in The Slow Pace of Fast Change by Bhaskar Chakravorti, a book that is built around the idea that diffusing innovation through networks is difficult. It’s a good book and worth checking out. The example that Chakravorti uses is the uptake of 56k modems in the middle 1990s.

When the 56k modem was released, it was demonstrably better than the 28k modems that it was replacing. The modems were twice as fast, and the trade-off to get this extra speed was, well, nothing. It was simply a better product. However, one year after it was released, only 20% of new modems being sold were 56k, and after two years, that figure had only risen to 50%. Why weren’t people using the better product?

I was using one. As soon as they came out I bought one, opened up my PC, took out my dreadfully slow 28.8k modem, and popped the the lightning-fast 56k modem into the slot. I was ready to roll! However, as usual, I was atypical.

The problem was that the percentage of customers that just went out and bought a modem was extremely small. The vast majority of modems were packaged within new computers. This meant that the new 56k modems had to fit within the value network that included PC manufacturer/assemblers (OEMs), end users, internet service providers (ISPs), and, critically, two modem manufacturers (U.S. Robotics and Rockwell), which had two competing communication standards. And you only actually got 56k speed from your modem if it was on the same standard as the one used by your ISP. Here is how Chakravorti describes the choices facing everyone:

Thanks to the competitiveness between the two standards, each participant had a critical binary choice. An OEM would slect a particular 56k modem if it expected that its users also wanted such a modem. But users would want these modems only if they delivered on the promise of higher speed. The ability of users to achieve speeds of 56k was, in turn, predicated on which standard their ISP was favoring. The ISPs would, on their part, favor whichever standard they believed would assure them of the largest customer base; in other words, they would back the modem standard that most OEM’s or PC distributers were likely to install in their machines. The outcome is that the market became a ring of interdependent decisions….The cycle created a self-sustaining equilibrium, in which the older technology persisted longer than it should have, give the state-of-the-art engineering. As a result, customer experience remained tethered to the status quo.

And of course, DSL broadband modems were also just becoming available, which made the interdependent network choices even more challenging. What can we learn from all this? A few things:

  • I was lucky that my first 56k modem was compatible with my ISP’s modem.
  • More generally, though, this is a great illustration of why it simply isn’t sufficient to come up with a better product, service or way of doing things. You have to embed your innovation within the value network, and this presents an additional challenge. Why are many people still not connected at 56k? Why is your doctor using a PC with an i486 processor that isn’t networked with the broader healthcare system? Why are you typing on a QWERTY keyboard? It’s because the value networks for these systems are stuck, which prevents innovation. The interconnectedness of the value network leads to the lock-in of inferior technologies.
  • To get your innovation embedded into the economy, you have to unconnect the members of your value network from whatever they’re currently using (28k modems, for example), and get them to reconnect to you. The unconnecting is a critical step that we often ignore – this is a mistake.
  • This explains a significant part of why it takes so longer for new ideas to spread. You not only have to get people to agree that your idea is better, you also have to get them to act on this belief. That’s the hard part.

Ralph Waldo Emerson was wrong when he said ‘build a better mousetrap, and the world will beat a path to your door.’ You have to build a better mousetrap, and then figure out a way to get it connected up within the mousetrap value network. That’s an entirely different problem. Solving it takes time, and that’s why innovation is different than invention, and why it takes so long for new ideas to spread.

(Picture from flickr/patterbt under a Creative Commons license)

About Tim Kastelle

Student and teacher of innovation - University of Queensland Business School - links to academic papers, twitter, and so on can be found here.


  1. Networks and the Information Glut « Complex Systems « Innovation Leadership Network - 9 January 2010

    [...] Ideas diffuse through networks. The structure of the networks through which we are trying to get our ideas to spread has a significant influence on the diffusion of our innovations. Our connections within the network can enhance or hinder our ability to get our ideas to spread. One of the reasons that Darwin gets credited with the idea of evolution through natural selection instead of Alfred Russell Wallace is that Darwin’s connections within the scientific community at the time were more numerous, more widespread, and better. [...]

  2. The Pace of Economic Evolution « Evolving Economic Entities « Innovation Leadership Network - 11 January 2010

    [...] demonstrably better than the ones that they are replacing. Part of this is due to the difficulty of breaking existing connections within value networks, and part of this is due to the natural reluctance that many people have to take up new [...]

  3. Everything’s a Network « Complex Systems « Innovation Leadership Network - 18 January 2010

    [...] It is often very difficult to unbreak ties once they have been formed (and this is why it is often difficult to get new innovative ideas to diffuse through an economic [...]

  4. Low Tech Innovation « Book Reviews « Innovation Leadership Network - 28 January 2010

    [...] It’s another great example of the difference between invention and innovation. The box itself was invented 30 years before “containerised shipping” actually got the idea to spread. It’s not enough to have the ideas, or even to show that they work – you have to get your ideas to spread. [...]

  5. Resisting Innovation « Complex Systems « Innovation Leadership Network - 3 February 2010

    [...] acute problem in education, and in the public sector. However, as I discussed in an earlier post, we see the same thing happen with the introduction of innovative new commercial ideas. Even products that are clearly superior along all dimensions, like the 56k modems versus the 28.8k [...]

  6. Building an Entrepreneurial Network « Business Models « Innovation Leadership Network - 20 February 2010

    [...] true, even when the technology is demonstrably superior along all technical dimensions (see the story of the 56k modem, for [...]

  7. You Have to Break Connections to Get Your Ideas to Spread « Design « Innovation Leadership Network - 19 March 2010

    [...] talked before about the importance of making new connections to get your idea embedded within the economy. But breaking connections is also [...]

  8. How to Innovate Ancient Technologies « Innovation Leadership Network - 18 January 2011

    [...] make enough charcoal to supply the big grocery store chains in the UK. This is again the problem of diffusing a new idea through existing economic networks – and this is where the business model innovation comes in. Here’s how McCloud [...]

Leave a Reply