October 9th, 2005


I'm getting kind of sick of people saying "microsoft will open up a bit, and open source will commercialize as it matures, just you wait and see". Uh-huh. Ok, on the commercialization front: The most lucrative IPO in history (VA Linux) happened five years ago, it's been a few years since IBM cashed its first billion dollars worth of Linux contracts, Red Hat is still publicly traded and profitable. I think they missed it.

Look: Open source is simply the commoditization of software. Interchangeable parts available from multiple sources and able to be serviced by the guy down the block in the corner strip-mall. As markets mature, they tend to commoditize. This isn't anti-commercial, this is capitalism in its purest form.

As for Microsoft relaxing and doing open source: it can't. The conflict between open source and Microsoft isn't with Microsoft's technology, or even Microsoft's culture. It's with Microsoft's monopoly. You can't monopolize a commodity market, more or less by definition. Very few commodity markets stabilize with a single player at much over 50% market share, not even the dominant long-term survivors like Coca-Cola. Microsoft could very easily participate in the open source world, as can anybody else. But they would not be a monopoly, nor could they _become_ a monopoly in the new space by leveraging their existing assets. And Microsoft's problem is, in the world it grew up in (and the way it still thinks) only a monopoly can survive.

The tricky thing about the software market is it's built around a positive feedback loop. People use whatever platform has the most software, and people develop software for whatever has the most users. The classic example of this is the telephone network, where a disconnected phone is useless but as the network grows each individual phone in it becomes more valuable becuase of all the other phones it can call and be called by. The value of such a network increases _exponentially_ with size, and such "network effects" tend to create natural monopolies. The largest faction is best because it's largest, and once it reaches 50% market share it continues up the S-curve to monopoly. In the early stages, product quality comes a distant second to first mover advantage and a motivated sales force. Past the early stages, backwards compatibility with the installed base and denying such compatibility to competing products (by any means) become the twin guardians of market share.

I could go into a lot of details about the peak and stagnation of such a monopoly, and maybe I will in future (upgrade treadmill vs monthly bills, hardware platform transitions, data formats, and so on), but those are simply vulnerabilities to exploit, not something to exploit them. The way monopolies are normally broken is via disruptive technologies, which don't directly confront them but instead render them irrelevant.

For the moment I'll assume everyone everywhere has read the 1997 new York Times Bestseller "The Innovator's Dilemma", by Clayton Christensen. If you haven't go do so. (Back when I wrote investing columns I reviewed this book, and later applied it to Microsoft.) One main advantage of a disruptive technology is the established players aren't competition for the small startups, and they often manifest as a "cheap plastic solution" to a previously expensive problem. Disruptive technologies always start by establishing a new userbase that could not be served by the previous technology, thus they always have a core funding expansion into the existing network, a core that the existing technology cannot co-opt. And any technology that can survive in the market can also improve, and thus meet the needs of more users.

When a disruptive technology expanding into a monopoly market created and (at least in part) sustained by the network effects of a positive feedback loop, the ability to interact with or replace nodes in the existing network is one category of improvement along which the disruptive technology can progress. Initially, in its "roaches under the floorboards" original niche, it needs little or no compatibility. As it expands, it can divide and conquer by finding interfaces to live on the other side of: the ability to talk through the network to a monopoly machine, the ability to read and write the same data file formats as the monopoly machine, web applications based on network storage and javascript, etc.

Open source is a separate network, but not an exclusive one. Any company that wants to participate in this network can do so, nobody can be locked out. This means that multiple small entities can join together into a single network, and the positive feedback nature of this network will also exponentially increase the value they receive and the value of their contributions, just as the dominant monopoly network does.

The monopoly network is sustained by its larger size, but is unable to take market share away from the commodity network due to its disruptive technology nature: the users who chose the disruptive technology did so because it has fundamental advantages the existing technology did not posess, advantages which are inherent and which cannot be added to the existing technology by incremental improvements. The disruptive technology follows the normal pattern of attacking the bottom 10% of the existing market, and the bottom 10% of what's left, until there is no longer anything to attack.

Note that as the existing monopoly loses market share, the positive feedback loop sustaining it weakens, and its advantage decreases. If it ever drops below 50%, the network effects can _reverse_, as the disruptive/commodity technology becomes the new standard that everybody wants to join and support because it's where all the action is. This means that the end game can be fairly sudden.

The real question on people's minds these days is "where is the tipping point"? Everybody knows the game is over if any other platform gains more market share than Windows: Microsoft's sales will simply implode. But could microsoft hold out in a world where Windows had 45%, MacOS X had 35%, and Linux had 20%? All else being equal, the first platform to acheive 50% market share becomes unstoppable, but a disruptive technology vs a sustaining technology is not equal. The market tends to be divided into areas in which the disruptive technology cannot serve at all yet and areas in which it is clearly superior. If Linux can grow market share _at_all_ against an opponent with over 80% of the pie, and increasing market share (and decreasing its opponents market share) improves its position, at what point does it start up an S-curve? (On the other side, how much of the early growth was low hanging fruit?)

The interaction of positive feedback loop network effects and disruptive technologies isn't really one I've seen studied much. We previously saw this happen with the phone network (where the monopoly yielded to a federal breakup pushed for by a cell phone company, years before VOIP came on the scene), and computer hardware (where the commodity S-100 machines fought the dominant market share of the Apple II for a few years, before the matter by a huge disruption of the market that left commodity hardware entrenched with dominant market share. But that's another story...)