Open-Sourcing the Creative Process

This article is an adaptation of a short talk that I gave last week, as part of the SDM Systems Thinking series. The recording and slides are available here. With gratitude to Lois Slavin. Much of these ideas are due to Eric von Hippel's peerless book, Democratizing Innovation.

When we think about innovation, we might think about the traditional institutions where innovation tends to happens. For instance, individual inventors or research and development departments and labs. But as Robert C. Allen points out,1 there is in fact another institution, wherein innovations occur as a result of the efforts of a number of people and in an incremental way. He called it collective invention. Nowadays, and for a large part due to the Web, one might posit, collective invention has become ubiquitous and might go by different names like open-source in software. What I think is intriguing is that this idea can actually be traced to at least as far back as the 19th century.

Between the years 1850 and 1875, in Cleveland, England, blast furnaces were seeing a noticeable increase in performance, as a result of increases in height and blast temperature. As it turns out, neither research and development labs nor independent inventors were an important source of invention for the iron industry during that period. Moreover, the British government didn’t really fund research in that field. And yet, the industry was seeing considerable productivity growth due to the actions of many individuals. That pattern of work appeared to have three distinct attributes. Firstly, information was freely exchanged between all the main players in the field. Secondly, the information that they disclosed to each other was put to actual use, to actually improve newly manufactured units. And thirdly, the resulting innovation was a series of small increments.2

The table below shows how the average height of blast furnaces went up between 1851 and 1871. Here, columns are years, rows are furnace heights and the values in the cells are the number of new or rebuilt furnaces. As the author observes, one can’t help but notice some of the more risk-averse “followers,” as it were, who would always produce furnaces of slightly lower height, just to play it safe. You can see those in the top diagonal.

We see another example of this type of collaboration in Cornwall, England, in the year 1811. During that period, Cornwall was one of the most advanced engineering centers of the world. Interestingly enough, a lot of the inventive activities that were undertaken there, were undertaken outside the patent system. And so in the early 1800s, the efficiency of high-pressure steam engines, which were being used to drain mines, among other things of course, was going up. Here as well, as Nuvolari notes in his paper,3 much of the advancements could be attributed to incremental improvements by a number of innovators.

We see here that 11 years after James Watts’ patent expired in 1800, a group of mine captains, as they were called, began publication of a monthly journal, whose goal was to share best practices and create a climate of competition. And very shortly after that development, the thermodynamic efficiency of steam engines began improving steadily up until the mid 1800s. This is a phenomenal shift. Information sharing had a direct impact on technical advancement.

The third and final example is from the ‘70s. Technicon was one of three common brands—the others being DuPont and Abbott—that made these automated clinical chemistry analyzers. These were machines that were used in clinical labs to perform tests on samples of body fluids and tissues. For Technicon, their machines were seeing both an increase in sales as well as an increase in user-generated research and development. This was in contrast with the competition.

This table, from von Hippel and Finkelstein’s paper,4 makes that point a bit more vivid. You can see that Technicon’s innovations are both more numerous and nearly half of them were developed by users.

As the authors point out, this difference between the user-bases of the two companies may be attributed to the ease with which users could modify the product to suit their needs. Put differently, Technicon’s platform afforded modification more so than Dupont’s. What’s interesting in this example is the idea that removing locks, in a sense, can help with innovation, customer satisfaction and sales.

More recently of course, examples are aplenty. Take Web browsers. Firefox has just 30-odd paid engineers working on a product that is used by hundreds of millions of users. It’s been able to do that thanks to the thousands of contributors who submit bug reports, patches and contribute to discussions involving product design and development, among other things. Several features started as user contributions and then made their way into the product, like Session Restore. Similarly, Chrome, the other major competitors in that field, is for the most part, based on an open-source project called Chromium, from which it takes the majority of its code-base. It too has thousands of contributors who provide value to the project.

So, while the James Watts and the Richard Arkwrights of the world have undoubtedly produced work that has been instrumental to the advancement of the human race. And while the Newtons, Einsteins and Teslas are remembered for their creative leaps of imagination, we have this alternative pattern of innovation that occurs as a result of micro-inventions and incremental developments over a period of time and by a group of innovators. The Web has played a major role in allowing influential user-innovations communities like GitHub to emerge, thereby bringing more people to this alternative method of thinking about innovation.

When we talk about innovators, we need to know what they look like. And by studying the literature, it would appear that innovating users have the characteristics of lead users. Lead users is a term that you may have heard in marketing and business circles. For those who haven’t, it’s a term coined by Eric von Hippel to describe users who are both ahead of the market and who have an internal drive to modify products so that those products better suit those users’ own purposes.

And, as it turns out, the modifications that those lead users tend to make, end up being pretty attractive to the market, as you can see in this chart. Attractiveness here is the sum of the novelty of the innovation and its expected commercial value.

So a good way to reach lead users is to remove locks. In November 2010, Microsoft released the Kinect for the Xbox 360. It was a $150 motion sensing add-on that allowed gamers to interface with the Xbox using gestures and speech. And while most of the world saw it for what it was, “roboticists,” to quote this Wired article, “saw something else.” One of many user-innovations to emerge was a project by a graduate student, who coupled the Kinect with an iRobot product to create a device that was able to map its environment and obey commands from human beings. What’s interesting with this example is that Microsoft was initially opposed to people tinkering with its product. But it wasn’t too long before they changed their position, quite possibly because they saw that a slightly more open Kinect led to some favorable results, like financial benefits.

This might seem counterintuitive at first, but there are real financial advantages to free revealing, like the distribution of costs and risks, much like we saw in the first two examples. In 1909, the members of the Institute of Metals gave another reason, when they wrote that the general increase of knowledge gives rise to general improved practice, which might lead to a larger use of the product.5 And of course, larger use means more sales and higher profits. Moreover, that increased use can benefit a manufacturer in the form of network effects, whereby the more people use the product, the more valuable that product becomes and the more likely it is that customers might purchase other products from that manufacturer.

There are also personal benefits. I’d like to paraphrase another quote from the Institute of Metals, who said that each person has some valuable piece of knowledge, and by sharing it, he or she will likely end up learning a dozen pieces of knowledge in return.5 This is a wonderful way of looking at the world.

Another benefit is reputation, which isn’t necessarily about vanity, but also about the desire to increase one’s value in the market. And it’s for that reason that we see so many people dedicating their time to helping others on websites like Stack Overflow or contributing to projects on GitHub.

Users who reveal things for free have an interest in seeing the work reach many people because they want others to scrutinize it, improve it and perhaps even create derivative works based on it. There is this mantra in open-source, which also goes by the name Linus’ Law, that says, “Given enough eyeballs, all bugs are shallow.” You don’t get to pick the people who come along and scrutinize your work, and so they might come from a variety of backgrounds and have a range of expertise. Ultimately, the work ends up benefiting from that wealth of expertise. That drive coupled with the elimination of a big barrier-to-entry for many people, which is parting with their money, makes it easier to achieve fast and rapid diffusion in the market.

And therefore, as a result of achieving rapid and wide diffusion, an innovation might actually end up benefiting from being a first-mover. In his book, Information Rules, Carl Shapiro points out that one of seven key assets for a technology to become a standard is for it to have a first-mover advantage.6 By being the first to freely reveal information, an individual or firm can reap the benefits of being a first-mover, i.e. their innovation can eventually become so widely used that it ends up becoming a de facto standard. It is that line of thinking that was likely behind Tesla’s recent decision to release their patents.

People aren’t always motivated by money, of course. Sometimes, they show a desire to “give back” or to do something for the social good. When Jonas Salk, inventor of the polio vaccine, was asked who owned the patent to his vaccine. His reply was beautiful in its succinctness. He said, “There is no patent. Could you patent the Sun?”

Finally, there is the practical matter of it not always being easy to keep things a secret. So the real choice, for a manufacturer, becomes, “Do I release this now, on my terms, or do I wait for someone else to leak it, in a form and method of their choosing.” For instance, in the blast furnace example from earlier, the author of that paper makes the point that the improved designs could not have been kept secret because those furnaces were constructed by contractors, and the designs were created by engineers who frequently moved between firms. So keeping things a secret isn’t always easy.

I’d like to shift gears now and talk about open-source in a completely different domain. The first time I saw open-source applied to publishing was in 2011 when I came across a project called The Data Journalism Handbook. It was a free reference book that had been put together over the course of a two-day workshop in London called the Mozilla Festival. And it included contributions from an international group of data journalists. The outcome of that effort was a book that was available online for free, in a number of languages, and then also in print.

That project made me think about how the diffusion of the written word in society has changed over time. In the 15th century, Johannes Gutenberg improved the moveable type printing press, and as a result of that, the number of books in Europe increased by two orders of magnitude over the coming few centuries. Then in the ‘90s, the Web emerged, and with it came a new medium called The Blog. So we went from zero such blogs to perhaps hundreds of millions over the course of a couple of decades. And with the Web, came Amazon and Alibaba and print-on-demand, so all of a sudden, people could publish their own books, and they could distribute electronic copies to millions of people. And, if they wished, they could even print them abroad, for a fraction of the cost, thanks to easier access to printers in Asia.

That kind of led to the Book of Bad Arguments project. Back in 2012, bad arguments were being used to cause a lot of harm to a lot of people, as they always have been in history, I suppose. And so I saw it as an opportunity to turn the scribbles that I had made in a notebook, a decade or so prior, about how to argue, into something more accessible for a broad audience. In terms of content, it’s a book on fallacies. But in terms of implementation and management, it’s an open-source project, making it no different in spirit than the other examples that we’ve seen so far. The book is available for free online. It’s shared under a Creative Commons license, and it was initially supported by donations.

As with any project, there were a number of challenges. For instance, how exactly could we reach a wide audience without a distribution platform and with limited resources? How could we encourage others to help make it better? How could we turn curmudgeons—if you will—into contributors? And so on. The answer to all of those challenges, as I saw it, was to give away the work for free. In order to encourage others to contribute, I made the project a website, so that I could have full creative control over it, rather than having it be, say, an e-book on Amazon where the author gets to control very little of the screen real-estate. I then added copy that essentially said, “Be the change, file a ticket!”

The process, shown above, is nothing too novel, though I am using fancy names to give the illusion of sophistication. The only difference from the traditional model of doing this sort of thing, is that the work goes out to the world in a slightly underdeveloped form. With this project, I shared the work on release with three active communities, whose focuses are in different areas. Namely, design, writing and philosophy. I ran a Google Ads campaign, targeting keywords related to fallacies and logic, and then spent the coming weeks and months incorporating all the feedback that came in, analyzing usage data and refining the work in response to user sentiment. Once the feedback slowed down, and the project was being shared faster and wider than before, I invested a significant amount of money in the project to take it to the next level, which involved producing a print edition.

This is a timeline of visitors, or sessions as Google Analytics now calls them. On July 15, 2013, the website went live. We got about 4,600 visitors on the first day and then about 1,500 visitors per day thereafter. On September 8, 2013, an influential website called io9 wrote about the project. Visitors for that day reached 21,000, and that’s what I’d consider to be the tipping point for the project. Each one of those spikes that you see is when an influential community or website linked to the project.

Almost immediately after that tipping point, people started to write in. And the feedback covered everything from layout to design to content to editing. For instance, I had no idea that in left-to-right languages, pages on the right are always odd-numbered. Another reader's suggestion to add variety to the backgrounds was a non-negligible design change that vastly improved the book's appearance. It was Linus’ Law in action, albeit it was being applied to a book here rather than to code.

Another group of readers were interested in expanding the work and they volunteered to translate it. Four such translations have been completed to date—by Maria, Diogo, Sadiq and Anton—and about half a dozen others are in the works. Another volunteer—Steve Hunn—was kind enough to create an app of the book for iOS and Android. Others are currently working on adaptations of the book for different domains and age groups.

Each new language, of course, is like a whole new project. It opens the door to a whole new subset of humanity. And, as the data shows, the work of these contributors is actually being used. The Spanish edition, for instance, has been read by almost 110,000 people and the Brazilian Portuguese one has been read by over 11,000.

The reasons that people give for wanting to contribute are varied. One contributor said that he wanted to see the work in his own language. Others said that it would be the best way for them to give back to the community. Another said, quite candidly, that he wanted to gain experience translating text into his mother tongue, and this was a good opportunity to do that.

A question that several people had asked early on was, “Who would pay for this, if it’s available for free?” Well, now we have some data to answer that question. The website has a slider that allows people to donate anything from $1 to $20, with the suggested donation being $2. Over 720 people have donated to date, and what’s extraordinary is that 52% of those who donated, donated more than the suggested amount of $2. These donations help pay for hosting, bandwidth and other costs.

The project's first pivot was shortly after its release, when someone posted the title on Goodreads. All of a sudden, I realized that some people viewed this as an actual book, even though it wasn’t an e-book and there wasn’t a print edition of it available. So I created a mailing list and asked people who might be interested in a print edition to sign up. About 2,000 people signed up within the first few weeks. That number eventually went up to around 6,000. With that shift, I realized that the user-base was slowly changing. All of a sudden, users had different expectations. People tend to have personal relationships with books, unlike with websites.

So we got 4,000 books printed in China. That’s two and half months of anxiety and sleepless nights captured in a single photo. On December 8, 2013, the three pallets finally ended their journey from the Port of Oakland to three Amazon warehouses. I then became an effective supplier for Foyles in England, and also for bookshops in Australia and San Francisco. Amazon wouldn’t ship books to Canada, it turned out, so I sold directly to readers in Canada, with the help of my daughter’s stroller.

This is the book at Foyles’ Royal Festival Hall branch. Foyles were phenomenal in their support for the project from the very beginning. If you’re in England, you should definitely check them out. During that same time, three publishers bought the rights to the Russian, Korean and Italian editions.

This is the Korean edition that came out just a few weeks ago. The Russian and Italian editions are also currently available in the market.

We recouped the original investment in the first five months. I was transparent with readers about where their money went. The first edition’s price of $16.99 gave us a margin of just 9%, which was enough to break-even if we sold 75% of the inventory. We ended up selling more than that.

We’ve sold over 21,000 copies to date, thanks primarily to a small publisher in New York called The Experiment, who took over printing and distributing the hardcover in May 2014. And they’ve done a wonderful job with it. The project remains a not-for-profit, so any royalties that we get from book sales, will go back into the project. In terms of reach, as of today, we’ve had over 1.5 million people read the online edition and over 116,000 people engage with it on Facebook and Twitter.

To conclude, I’d like to share a few lessons, and tie this recent example to some of the things that I mentioned at the beginning. Locking something open can afford innovation, particularly when there’s a need to experiment due to there not being an established theory for an optimal design, as one of the authors from earlier put it.

An open product can still be financially sustainable. The author Neil Gaiman realized that he was selling more books in Russia where people were pirating his books, left, right and center. So he simply stopped worrying about piracy.

Free revealing can sometimes be the poor man’s clinical trial. You put something out and then look for a rash. Or to borrow a metaphor from Joi Ito, you can think of it like the immune system, where you subject something to all kinds of forces and then it gets better as it reacts to those forces.

Not every dark horse is spotted before the finish line. For many people, getting out of obscurity requires a lot of digging. Gatekeepers can sometimes be obstacles in that effort. And so leveraging the Web as a platform, and seeing users as potential collaborators, can be of tremendous value.

Reaching lead users and benefiting from their contributions can be of great value in the long-term, since it allows the project to evolve and remain relevant. It's a rephrasing of what Clay Christensen called the Innovator's Dilemma.

Communities like GitHub are a wonderful platform for innovation and for reaching others. I think GitHub works for all kinds of projects, not only coding ones. The whole platform is designed to afford highly-effective, hassle-free collaboration. And then, most new products end up failing due to misunderstanding users’ needs. So having users be a part of the upstream and downstream processes for a product can help mitigate the risk of failure.

Trust, transparency, pragmatism and the abandonment of ego are all things that I think are crucial for running a successful project of this sort. For those of you who might have seen The Oatmeal’s recent Kickstarter, I’m sure you valued the wonderful way in which the creators of that campaign engaged with backers and updated them with progress and developments. It creates an atmosphere of bonhomie that benefits everyone.

And then finally, the more people who see your project, the more likely it is that it will be successful. Or as I believe Cory Doctorow put it, the more lottery tickets you buy, the more chance you have of winning the lottery. And with extra exposure, comes the opportunity of benefiting from network effects.

Over the past year, I’ve been working on another project, which will follow in the same footsteps as that first one. It’s a story about a tailor and his accomplice who struggle to escape a land where things aren’t what they seem. In the process they, and by extension the reader, learn about computer algorithms. This project is also illustrated by Alejandro Giraldo. The project is looking more like an embryo nowadays, which means that I’ll probably subject the world to it soon. And, of course, I’ll be on the look-out for rashes!

1 Robert C. Allen, Collective Invention, 1983.

2 Ibid.

3 Alessandro Nuvolari, Collective invention during the British Industrial Revolution: the case of the Cornish pumping engine, 2004.

4 Eric von Hippel and Stan N. Finkelstein, Product Designs Which Encourage—or Discourage—Related Innovation by Users: An Analysis of Innovation in Automated Clinical Chemistry Analyzers, 1978.

5 Gerard, A. Muntz, The relation between science and practice and its bearing on the utility of the institute of metals, Journal of the Institute of Metals I, 1909.

6 Carl Shapiro, Information Rules: A Strategic Guide to the Network Economy, 1998.