Thursday, November 30, 2000 by Dave Winer.
When engineers build a suspension bridge, first they draw a thin cable across a body of water. Then they use that cable to hoist a larger one. Then they use both cables to pull a third, and eventually create a thick cable of intertwined wires that you can drive a truck across (actually hundreds of trucks).
That's a bootstrap. First you take a step you know is on the path, learn from it, and use it to lift up the next level. And unlike the designer of a suspension bridge, software developers must be more flexible, because the pace of innovation in our art is so rapid. We don't know exactly what next year's trucks will look like, how much they weigh, or how many wheels they have.
That's why networking technology comes in layers and why they must be designed with more power than they need to get today's job done.
Over the years in this column I've told the story of a bootstrap.
We've started several technology layers including RSS for content syndication; XML-RPC and SOAP for distributed applications; and our newest format, OPML, that allows hierarchies with attributes to be shared over the Internet. Manila, which we launched a year ago, is managing 15,000 websites around the world. We're adding more sites and more servers all the time.
Now, in the process of bootstrapping UserLand, other small groups of developers have started their own businesses to explore the same technologies. They're swimming in a sea of big companies like IBM, Microsoft, Sun and AOL. It takes courage and a strong belief in ones' ideas and a certain amount of naivete to hop in the market. But they are hopping, and some of them are meeting with the investors of Silicon Valley, and the pitches aren't going so well. I'm going to try to explain what's going on, and why we need the help of the investors, and why it will be very profitable to help them get going.
First, a contrarian point of view about incubators.
In my humble opinion, incubation has gotten an undeserved black eye. Everything about the idea is good, except it's been done wrong, in the wrong environment, with the wrong goals. The incubators have been about brands and markets, product concepts, applications. Instead they should have been about technology.
I was one of the founders of Symantec, a company started in the 1980s, with incubation and bootstrapping in mind. Our plan was to grow via mergers and acquisitions, to create an environment where technologists would be happy, and would create new wonders and utility that would be commercialized by a shared sales and marketing organization. This was a solid idea, and pretty close to the incubation model that became popular later.
However it didn't boot up as planned because when the companies merged, they didn't have compatible interfaces. This was true at all levels. The products didn't work together. And the companies weren't started to be merged. If we had known we were going to merge we would have done things differently. Very differently.
When I left Symantec I was determined to solve the problems at a technologic and human level, so that a "backplane" like Symantec could work. There were detours along the way, the Mac market went kaput, and then the Internet, a blessing, because it simplified the role of the backplane, came along and ripped up the whole road. But we've now put it back together, and networking, the missing piece of the 80s and 90s is in place. That's what the XML-based formats are all about.
Let's find a way to start lots of standalone companies, fund lots of independent-thinking developers with a twinkle in the eye, it's so cheap, but let's give them standards to work with. Let's make it so that when the companies merge, the services plug together, and work. Based on many years in this industry, I think this feature would please the VC mind. The mergers they conceive would be implementable, they'd work.
But this would also be a corner-turn for the VC industry, because, as it's now structured, they funnel tens or hundreds of millions of dollars through a relatively small number of companies. However an opportunistic VC would have an incubator (call it something else if you want) that flows much smaller amounts of money to the bright eyes, without much fuss and not many strings attached, while we figure out how to get a few backplane companies going.
This is a bigger bootstrap, it's the way we can build for the next boom. If we can do this, it will be easy to merge companies, and get the synergy we talk about but rarely deliver.
PS: Bootstrap is an ancient computer science term. When you turn on a computer it bootstraps, or "boots." First it loads the most ancient bit of code, probably written in the 1970s. It runs a program written in the 80s, which in turn launches a program written in the 90s. Each of the levels loads only for the purpose of loading the next bit of history. Doug Engelbart was the first to use the term in the context of this piece, as far as I know. All engineers bootstrap all the time. To understand bootstrapping is to understand software, imho. It's the process that matters, not the bits, or system requirements.
PPS: The computer science term is derived from "Pulling yourself up by the boostraps." In the physical world bootstrapping rarely works. In the virtual world it always works, or so it seems.
PPPS: An incubator is a business whose business is starting other businesses. It was a hot idea in the investment world a couple of years ago. Bill Gross's IdeaLab is the canonical incubator.
PPPPS: Kaput is a Yiddish term for blowing up, disappearing in a decisive way.