At the inaugural Open Source Business Conference in 2004, the discussion centered on how to fund open source's survival. Just six years later, the OSBC conversation has taken a 180-degree shift to focus on whether proprietary software's shelf life is nearing its end as open-source software economics increasingly drive technology innovation.
In a nutshell, the cost benefits of high-quality, free software came to outweigh the industry's former concerns about risks associated with "rebel code."
This trend, not visible in 2004, started with early adopters like Google. As Red Hat CEO Jim Whitehurst highlighted in his opening OSBC keynote speech Wednesday, Google couldn't exist, and certainly couldn't have adopted its current business model, but for open-source software.
Whitehurst backed this up by pointing to Google's planned 10 million servers. Adding up the roughly $2,000 in software costs per server (for the operating system, database, etc.) means that it would take roughly $20 billion for Google to run those 10 million servers with proprietary software.
Today, perhaps Google can afford that. When it started? Not a chance. In a world of traditional software economics, Google would not exist.
That's how open source enables the early adopters. It's why open source, not proprietary software, is driving cloud computing through companies like Northscale and Cloudera, but also through projects like Linux, which IBM sees at the heart of cloud computing.
But it's not necessarily about technology. To quote Whitehurst, "It is the economics of open source that are driving the development of the cloud." Microsoft can build great technology. It hasn't yet learned how to give it away, at least, not outside a narrow niche of its product line.
How will these economics apply to old-school industries like automotive?
Whitehurst pointed out that the typical car today has 100 million lines of source code, and Frost and Sullivan forecast that number to climb to 200 million to 300 million lines of code in the next few years.
If we're not careful, Whitehurst warns, electronics will become the bottleneck to innovation in things like cars:
We wouldn't feel comfortable servicing a car with the hood welded shut, and we shouldn't feel comfortable driving around 100 million lines of closed-source code. We are using 20th century development models for developing software that go into these products....
We are likely to see innovation slow significantly over the next decade if we don't adopt open-source development or open-source principles.
Given how fast the industry has moved toward open-source infrastructure as standard operating procedure, I suspect we'll see the open-source adoption that Whitehurst urges is critical to future innovation.
What we haven't yet seen, at least not consistently throughout the open-source world, is how to monetize open source. That question remains constant since 2004. While the survival of open source is no longer in doubt, the survival of some open-source business models very much is.
Here's a hint: the best "open-source companies" don't sell open source at all. From Google to Red Hat to Facebook to Oracle, such companies sell value around open source. That's the winning model. The trick is to figure out how to apply that principle to a particular business.