About ten years ago I started writing Free for All , a book that set out to explain the open source software movement to the world. Within days, I realized that the topic was so big and sprawling that I started telling people that it reminded me of what Charles de Gaulle said: it is practically impossible to rule a country that made 246+ types of cheese.
After a few months, I managed to muddle through and ship a book that was just a small glimpse at a burgeoning corner of the software world. That's changed dramatically. If there were 246 types of cheese then, there may be 2 million or more today. Open source development is the dominant framework in the software world and it remains one of the most fertile and ambitious areas of human exploration.
I thought it was a good time to pause and take note of what I got right and what I got wrong in the book. It's not that ten years is an important anniversary, it's just I wanted to experiment with putting out an iPhone application filled with text and this seemed to have the most universal appeal. My other books are filled with too many equations for the small screen.
The onset of the second great depression also set me thinking about the nature of an economy free of cash. The open source world is a good example of the rise of what might be called the "post monetary world". The computer industry makes it easier and easier to count money with maddening precision. When things can be counted with precision, they will be taxed and when things are taxed, the governments will inevitably squeeze most of the life out of them. The open source world is one example of people escaping the clutches of the taxable economy. But I'm getting ahead of myself. I going to save most of these points for the end.
The most valuable part of this essay may be the parts analyzing what I missed when I wrote the book because I hope that they'll help lay out the groundwork for the next generation of licenses, projects and distros. The movement has been very successful and I have every expectation that it will continue to flourish. But it has not found, as Torvalds half-joked, "total world domination" and I'm less and less sure it will ever find it. It may become the dominant philosophy in the software development world; it may even become the dominant operating system and license; but I think that the need to pay the rent will eventually create a hollow shell that's more open source in name than in the spirit.
Again I'm getting ahead of myself. There have been many victories over the last ten years and it would be a shame not to spend some time celebrating the gifts they've brought to our world. They may not be perfect, but they are still worth noting and enjoying. So here's a list of ways that I think the book succeeded followed immediately by a discussion of how it failed.
When I wrote the book words like "Linux" and concepts like "open source" were just on the edge of joining the general consciousness of programmers and the business community. Most still saw software as an item that was purchased for money like a car or a refrigerator. It came with some kind of assurances from the manufacturer and your job as the consumer was to use it like a black box.
The open source movement stood that idea on its head in much the same way that Thomas Jefferson and the other democratic radicals inverted the ideals of government. The power, they said, came when people participated in the process which, in the case of software, happened by reading the code, tweaking the data structures, revising the if-then statements and melding your brain with the code.
Open source software has one crucial advantage never enjoyed by other utopian dreams: super low costs for duplication. The communal farms could never make infinitely many copies of the crops to feed the slackers, the blowhards and the troublemakers. But the open source projects can easily give away tens of thousands of copies to non-contributing leeches as long as the people with the ability to work are part of the mix.
Open source software is also unique because the complainers, the whiners, and the blathering trolls contribute in their own special way by flagging problems. Bug testing continues to be a real challenge for every software project and the incessant yakking jerks who can bring down a community theater project are often pretty helpful if the organizers are smart enough to see through the tone.
The strength of this one advantage has pushed proprietary software companies to emulate many of the features. There are now "developer's licenses" that encourage programmers to tinker with packages. There are free introductory programs and endless give aways. For many intents and purposes, there's are few differences between using a proprietary package and working with an open source project. Aside from the price, many proprietary packages have adopted the most useful parts of the open source philosophy. It is the dominant mechanism for creating code in the world today.
The open source community found a great deal of success well before I started writing Free for All . The most famous example may be the Netcraft survey of websites that proclaimed in 1996 that Apache was the dominant tool running a webserver. This success came out of nowhere and clearly rattled Microsoft and the other proprietary vendors.
It was easy to see that quick success and start guessing that the desktop computers could fall just as quickly. The user interfaces built on top of Linux were already virtually equivalent to the front-ends from Apple and Microsoft. It wasn't hard to pick it up and conclude that the rest of the world would see the advantages and switch, just like the guys who ran all of those websites.
This has not proven true. Even today, Microsoft and Apple dominate the desktop. While Apple's latest OS is built on top of BSD, the other great open source operating system, it is filled with increasingly proprietary corners. Both of the companies continue to be very popular and very few of the desktops sold come with a Linux OS. (This is, I think, a bit of a distortion. I'm writing this on an Acer Aspire One, a machine that came with Windows. I considered buying the Linux version, but the price was the same. So I got the Windows license and then installed a version of Linux that lets me boot the machine in either Linux or Windows. I like each of them for different tasks.)
While the open source operating systems are just as good as the commercial versions-- and in some cases they're markedly better-- that doesn't mean that the average human is going to pay much attention. The average human can't read the source code much less improve it and so they really see no benefit from the openness. They just want something that works.
This is probably the biggest reason why the open source OSs can't get real traction in the larger world. People like going to the Apple Store to talk with a Genius, even if the Genius turns around and tells them that the only solution is to spend $200 on some new package. The hand holding is very valuable.
There's even been some erosion in the web server business. Microsoft has drawn closer to Apache, although it has not overtaken it as some predicted might occur back in 2007. These numbers are far from perfect because they're skewed by the millions of small sites that are hosted in big server farms.
The open source community has certainly tried to sell support but this has been a tough sell even to the other programmers. Somehow many of the firms have found it hard to charge a reasonable amount for something that's assumed to be free. There are a number of firms offering support, but they are only moderately successful. Somehow there's not much call to pay for the cow's time once the zeitgeist converges on the idea that milk should be free.
There has been a good deal of success in hidden open source operating systems that are buried inside of appliance-like devices. The core of the iPhone is BSD, the core of Tivo is Linux. Even some parts of Microsoft's operating system relied heavily on open source code. While all of them comply with the licenses by redistributing their code when it's required, all of these gadgets withold some proprietary code, some secret sauce that prevent others from building quick clones.
A measured amount of local success may be a far cry from total domination, but it's still something to enjoy and savor.
A few open source projects are also financially successful; they bring in millions of dollars in revenue and distribute the code to everyone who might be interested in reading or revising it. These are unique because they pay their programmers real salaries and generate real returns.
Two of the most prominent hybrid organizations are Mozilla, the creator of the Firefox browser, and MySQL. Mozilla makes its money on an advertising deal with Google. In return for setting Google to be the preferred search engine, Mozilla reaps tens of millions of dollars a year. It has even paid its leader a salary of a half million dollars. That's not too shabby for a tool that's given away for free.
MySQL is the favorite database for many web developers because it's ubiquitous and free, at least at first glance. The company makes its money with a mixture of licensing, support and training. In the past, I've given one of their courses on storing sensitive information with MySQL using examples from my book, Translucent Databases .
Sun Microsystems bought MySQL in early 2008 paying about $1 billion for the rights to the company. That's a big price tag for a company and a check that any entepreneur would be happy to cash. If anyone who doubts that open source can make money for people, they should shut up because there are hundreds of smaller versions that are cashing smaller checks each month.
The company is much more like a traditional software company than a grassroots project. Most of the code is written by company employees and the very little that comes from outside can't be included until the copyright is assigned to the corporation. The company owns the database code, every single line of it. They just happen to release large parts of it with the GPL.
The company is also cagy about the rights associated with the license. In general, the GPL has one major requirement: if you redistribute the software in any form, you're required to make the source code available too. There's no requirement to pay anything and there's no way for MySQL to stop you from redistributing it if you include the source code. So there's no requirement for anyone to pay MySQL for a copy covered by the GPL license.
I've watched MySQL salesmen hide this fact in a murky fog of legalese. Perhaps, they suggest, you might need a license. Why don't you just buy one just to cover yourself . They also point out quite correctly that it's good for the users to support the product. It's a clever mechanism and it's generated enough revenues to keep the company moving forward successfully but it's a far cry from some neo-commune.
Is this something that validates open source development or something that confirms that the software business as another business where the titans of industry battle over control of the profits?
It's worth noting that a long-time competitor of MySQL called Postgres has never found the same business success as MySQL despite having what some felt was a superior product with better features and faster performance. I'm not sure if I'm willing to guess whether Postgres is actually better, but the fact that there are enough fans to make these claims means that Postgres is a real competitor for MySQL.
The companies that tried to wrap themselves around Postgres, though, never achieved any major or even minor success. Great Bridge failed to attract enough attention and died after the 2001 crash. There is a long list of companies that provide support for the code base, but most do this in the context of creating web applications and helping with database design. They don't serve the same function as MySQL.
These companies aren't going to be acquired for $1 billion. They may have a very sustainable model that will carry them forward for a long time to come, but they're not going to do much but bill by the hour for their time.
While the failure of others does not mean that MySQL's success doesn't count, it does suggest that the open source model is very fragile and hard to pull off. If the guys who like Postgres can't get a working firm off the ground with what some feel is a superior product, then perhaps the model isn't that good.
Where are the copycats for MySQL? Where are the other open source databases? One of the other good examples, Sleepycat, is dramatically smaller than MySQL and it targets a much smaller niche: embedded products. Many people are using Sleepcat's product without seeing the logo or knowing that it's hidden inside their software because Sleepycat sells only to other programmers looking for a library that will save them time. Sleepycat sold out to Oracle and is now one of Oracle's products -- one that can be used with an open source license.
Mozilla's deal is also a stunning validation of the open source philosophy that comes with some worrisome caveats. Google doesn't write a check to Mozilla because the browser is open, it writes a check because many people use the browser. Does the openness help? Perhaps. It's given the browser a cool sheen and that alone might attract a number of users. Firefox's plugin environmenet is very attractive, but Microsoft's browser is also extensible. In the end, I think most people are interested in it because it's free and it works well. Access to the source code means little to almost all users.
I could imagine another company having the same level of success as Mozilla with a closed source solution: give away a quality product and make money on the ads that flow through it. There's even a good example, the Opera browser. The small Norwegian company one that has never dressed itself in the same open source dress, but it pulls in hefty revenue from the same deal as Mozilla. It routinely claims to have the fastest brower around and this technical success draws in users and hardware manufacturers who license the technology.
Neither of these examples disprove the idea that developing open source software can pay the rent but they do show that the link is not as obvious as it may be. Mozilla and MySQL may have succeeded despite releasing their source code not because they did it. Their openness was a cagy business strategy to gain publicity and lure early adopters, but I don't think it was the fuel for their fire.
About ten years ago, I asked a big programmer at MySQL why he thought they succeeded over Postgres, even though Postgres had some important features that MySQL couldn't duplicate at the time. He told me that MySQL was simple to install. They spent hours testing the installation process and bent over backwards to make sure that any could get it running in five minutes. That, he told me, was the secret of their success. Once they wormed their way into all of these machines, they could start ramping up the license fees to support the enterprise. It was and continues to be a great business model that helps many people but it's not a validation of open source principles.
When I wrote Free for All , most open source software wasn't very easy to use or very polished. Most Linux distros would start up with a scrolling page of messages that helped the programmers but confounded everyone else. Some packages had decent user interfaces but most of the tools had glitchy user interfaces that required a fair bit of learning. One car magazine once explained that a car had ``character" if it took you more than 15 minutes to explain how to wiggle the lever when you loaned it to a friend. Most of the open source software had plenty of character then.
Much of this has changed. The leading edge distros are highly polished and many of the programs are both easy to install and relatively easy to figure out. Many packages still have a some character but the rough edges are disappearing.
I think this development is helped by two forces: competition and the openness. The first is a fact of life for all open source packages. There are no lock ins or contractual relationships that keep users in line. Furthermore anyone can take a good piece of work and make it a bit better. The marketplace is constantly shifting as each person adds a small bit to the entire commonweal. It's both a communist and a free marketeer's dream come true.
The openness also lubricates this competition. Any author knows that the packages must ship with a good build file if anyone is going to use the software. The open source authors can't rely on any mandate from above to enforce some rules so the individual authors have an incentive to play nicely with each other and work out as many kinks in advance.
The polish on the top of the open source software is often skin deep. In many cases, the documentation is thin or even non-existent. If I had a dollar for every place-holding, blank comment I've seen in open source documentation, I would have enough to hire a writer to fill in the blanks. This is one of the unsolved problems for open source development.
Many programmers often talk about the need to recruit more writers but this rarely happens. The best documented open source projects are either the ones that sell a book separately or have some corporate sponsor paying for the writers. As someone who writes both software and text, I'm not surprised. It reminds me of many of the community theater productions where there's usually plenty of actors who want the limelight but a big shortage of orchestra members, costume designers and lighting techs. There's little ego gratification in the backstage work and so the community groups either hire professional orchestras or draft the actors. Documentation authors also get little ego gratification. They don't get to decide much about the direction of a project, but they're just supposed to dutifully track it and explain the decisions of other egos.
There's also a perverse incentive to ignore documentation. Many open source projects gave up the chance to sell the software itself when they chose the open source license and so selling support and documentation is the only route available to them. If they want to make money, they need to leave enough complexity and ambiguity in the code to make it difficult for users. This leads to a strange effect: programmers want to make it easy enough to lure others into using the software but hard to get beyond a few simple examples without contributing resources to the project.
In the end, I think programmers contribute just enough documentation and polish to make it easier to collaborate with others but they usually stop when they're actually doing some work for someone else. They're willing to share they work they do for themselves but others need to be prepared to roll up their sleeves with an open source project if they want the software to do anything new or slightly different.
There are a number of functioning companies that do a good job helping people use open source software. Red Hat, one of the early darlings of the stock market, lives on and continues to sell subscriptions to the people who run the big server farms. They concentrate on the hard core tech guys and succeed at selling them some support.
There are others doing well. Ubuntu is now one of the favorite distributions for desktop machines thanks to the high level of polish and support. According to one article by Ashlee Vance in the NY Times, Canonical, the company shepherding Ubuntu, is reaping close to $30 million in revenue per year, about enough to support the 200 or so official employees. The money comes from companies like Dell that ship machines with Ubuntu on it and other companies that need support. That's the beginning of a sustainable mechanism.
As I write this, I'm haunted by the fact that proprietary software has some serious advantages for supporting a development team. While all of the secrecy and limits to sharing are onerous and terribly limiting, they also enforce a fundamental level of fairness. All of the users of Microsoft's Windows contribute something to the development. All of them pay into the ecosystem, helping keep it sustainable. Open source projects can only absorb time and that's something very precious.
There's also a strange kind of socialism built into the one-size-fits-every-computer license that comes from Microsoft. No one can escape being a partner because no one can escape paying something. The manufacturers are the gatekeepers and they keep the users in line.
Open source projects have no leverage over their users. There's a intoxicating level of freedom that's wonderful if you happen to be adept enough with computers to compile the software yourself. But most of the world can't rebuild the binaries and so most of the world has to rely on the kindness of the programmers who can do these kind of things. Most of the world is beholden.
The real danger with the open source information ecosystem is that the strong will opt out of working with the rest of the world. There's nothing the non-programmers have to offer that's of any value to the coders and so the coders move off on their own, developing very cool projects that do very cool things that are only of interest to themselves. They have the freedom to drift off into their own world and there's little that the rest of the populace can do to keep them in orbit.
Well, that's not true. Eventually the programmers will need to eat and pay the rent. They won't have any of those messy dollars to accomplish that task, but maybe they'll find a way to do it.
The neat fact is that proprietary software closes this feedback loop. It may encourage weird distortions in the marketplace by creating too many weird features and virtual chrome tail fins that speak to the buyer in the showroom, but there is a feedback loop in place.
A few weeks before I started this section, Forbes magazine listed the "biggest losers of 2008", a companion article for their list of the wealthiest people in the world. Bill Gates made the list because Microsoft stock lost 45% of its value during 2008, "wiping $12.3 billion off his personal balance sheet in 11 months." Of course he still has a few other billions to play with. The list is a bit deceptive because you need to have a large fortune first before you have a chance to lose it.
When the onset of the second great depression began, I dug out one of my favorite sections of the book from the chapter "Wealth":
But money is the answer only if you want piles of paper with pictures of famous Americans on them. Several countries in Latin America generate huge piles of money from drugs, oil, and other natural resources, but the countries remain quite poor. The leaders who end up with most of the money might like the huge disparity, but it has very distinct limitations. When it comes time for college or medical care, the very rich start flying up to the United States. Johns Hopkins, a hospital in Baltimore near where I live, provides wonderful medical service to the poor who live in the surrounding neighborhood. It also has a special wing with plush suites for rich people who fly in for medical treatment. Many are potentates and high government officials from poor countries around the world.Money is a poor storehouse of wealth. It's fragile because the value of it changes day by day but the numerical nature won't let it adapt. Interest bearing accounts and inflation protected securities are fragile mechanisms. If you borrow $1 from a friend, you've got to pay back a $1 to remove the debt. It doesn't matter if your fortunes have changed. It doesn't matter if inflation or deflation have changed the relative value of the piece of paper. When debts are drawn up there, there's no way to know what the future might bring.
People in the United States can enjoy the synergies of living near other well-educated, creative, empowered, and engaged citizens. People in poor societies can't assume that someone else will design great roads, build airlines, create cool coffee shops, invent new drugs, or do anything except get by on the few scraps that slip through the cracks to the great unwashed poor. The ultrarich in Latin America may think they're getting a great deal by grabbing all the pie, until they get sick. Then they turn around and fly to hospitals like Johns Hopkins, a place where the poor of Baltimore also enjoy quite similar treatment. Wealth is something very different from cash.
Most folks in the free source world may not have big bank accounts. Those are just numbers in a computer anyway, and everyone who can program knows how easy it is to fill a computer with numbers. But the free source world has good software and the source code that goes along with it. How many times a day must Bill Gates look at the blue screen of death that splashes across a Windows computer monitor when the Windows software crashes? How many times does Torvalds watch Linux crash? Who's better off? Who's wealthier?
If only there were something better than money to bind our economy and make it possible for people to trade with one and other. When I was writing my book Digital Cash , I found a number of people who experimented with alternative ideas of currency. Some purposely avoid a word like "dollars" and use the currency to count "hours" of work. In the past, the United States currency was backed by tobacco leaves in a warehouse before the country shifted to gold and silver coins. Today, it's backed by the full faith and credit of the U.S. government, a rather unsteady foundation.
Open source software is relatively unique because it's a functioning ecosystem without money. Somehow a fertile world filled with functioning software continues to grow without a currency to mediate the trades. People give freely because they usually receive more back than they give. It's a good deal for everyone and so it thrives.
When I wrote Free for All , I spoke with Sameer Parekh, an entepreneur who built a semi-proprietary tool upon the foundation of the Apache server and sold the result. He explained to me that he often contributed some but not all of his team's proprietarty work back into the Apache code base because it would make it easier for him to work with future versions of Apache. His donations would shape the direction and keep them working in rough synchrony with his team. Even though he was working on maximizing his own profits, it made sense for him to contribute to the commonweal.
The point was that even people who are making money off of the common code base have an incentive to improve it. All of these small contributions stick together in ways that they never worked with utopian farms. The cost of leeches is much lower than when real goods get in the way.
Open source projects also succeed because they don't have the weight surrounding for-profits and even not-for-profits. When there's no money involved, there's nothing to tax. That means there's no need for all of the accounting and legal infrastructure that bind together corporations. When there's money, it must be counted. When there's no money, well, there's no reason to worry about it.
I've often wondered if computers would help contribute to the downfall of our traditional form of capitalism because computers can count so well. It's now possible for a corporation to account for every single penny flowing through them. They can shave royalties into tiny little fractions and then add them back together again. Everything can add up in some big spreadsheet and the law insists that it must.
The problem is that the old philosophy we have for accounting and taxes came from the days when computers couldn't keep track of everything. There was, to paraphrase Milan Kundera, a fair amount of laughter and forgetfulness built into the economy and this led to the forgiveness that let the world function and move forward. Computers don't forget anything and so they are doomed to tabulate everything, creating mountains of data filled with a strange internal logic that weighs us down with heavy responsibility. Everything that can be counted, must be counted and the taxes must be paid.
Tom Daschele is just the latest victim of a system that taxes a person's "income". If something has a value and you receive it, you're supposed to compute the value and pay a tax on this income. In his case, he received the use of a limousine for free. As limousine rides have a real cost, then Daschle was supposed to pay a tax on this income, something he eventually did.
This reminds me of the delicious debate over the taxes that accrued to the person who caught some of the balls hit by Mark McGwire and Barry Bonds in the process of tying and then breaking Hank Aaron's record for career home runs. Some troublemaking reporter started asking the question and the poor Internal Revenue Service was hard-pressed to give a straight-forward answer. Was it income for the lucky or unlucky fan? Was it a capital gain? Are there any deductions available?
There were so many possible solutions and no one knew what was correct. Here are just some of the interpretations floated on the discussion board of the Wall Street Journal:
There are more and more examples of similar problems creeping into real businesses that try to deal with the increasingly precise accounting provided by computers. When every action is tabulated, every action can be taxed in some strange way. It must drive the accountants and comptrollers of the world mad.
The open source world escapes all of these troubles because money isn't in the picture. Free is such a powerful force on the Internet because people can spend their time sharing something with a real value to one and other but one that can't be computed in terms of money. The small gifts of information are free but they add up to something real for everyone. Just as the power to tax is the power to destroy, the power to live free of taxes is the power to build a functioning ecosystem with the textual equivalents of a wink or a smile.
While this ideal is great for life on the Internet, it poses real problems for the programmers and info-farmers on the Internet who want to interact with farmers of real food and builders of real homes. The small exchanges of information are a valid ecosystem on the Internet, but they don't work well setting up a mechanism of exchange. There's no easy way for the programmers to swap cool code with a farmer because the farmers can't reciprocate and won't get much from the deal. The farmer isn't going to look at a nice recursively descending parser, smile with admiration, and then install it on his cow feeding computer.
The biggest challenge, though, is that atoms can't be duplicated like bits. Sure, a programmer can share Linux with the farmer but the farmer can't feed the programmer with the same openness or simplicity. Distributing food doesn't scale. As electronic engineers like to say, there's a real impediance mismatch between the amplifying circuitry of both systems.
Will open source programmers find a way to trade with the rest of the world? I can think of two mechanisms and neither of them is very powerful. First, the open source programmers can sell their services to non-programmers and use this income to pay for the real things from outside cyberspace. They can work together to share their work, but they can sell the rest of the world the chance to use the software.
The second is to work with hardware manufacturers who are selling something that can't be easily duplicated. These guys can charge a premium for convenience of buying a box that does the right thing and then pass this along to the open source programmers. Already, most equipment manufacturers support open source teams either directly by hiring them or indirectly by giving grants to their favorite projects. Some of the highest profile projects list the hardware manufacturers as their biggest sponsors.
Neither of these is a very perfect mechanism. Selling support is a difficult game because of its perverse incentives and the fact that it's so brutally competitive. There's no customer lock-in, something that favors the consumer but hurts the supplier. There are a good number of successes in this field but it's not clear that support will generate enough of a surplus to fund the next generation of software. Again, there are perverse incentives at work. If the programmers do a good enough job and produce software that's stable enough to give them the leisure time to attack the next big idea, there's no reason for many of the users to pay for support. The successful open source project can't support the creation of a new idea.
The solution may be to be happy with the mechanisms that open source nourishes. It may not create surplus capital that can be spent on something completely new, but a good project will spawn plenty of spare time. Open source projects are very good at mixing together the spare time of programmers and turning it into working code. The surplus capital may not be granted directly to the creators but it will still exist.
When I wrote Free for All , I was struck by the parallels between the way that George Gilder described capitalism in his book Wealth and Poverty and the way that the open source community described their projects. Both saw the work as pure gifts made with little guarantee that anything would be returned. Gilder's view might not resonate with someone staring at the monthly bill from the credit card company summarizing the pure gifts of credit granted over the month before, but the characterization is more and more apt as people openly talk about walking away from their mortgages.
Gilder saw the rewards that flowed back to the entepreneur as a smart feedback loop for society. The best inventors, the ones that gave the most to society, would be rewarded with more capital to play the game again. The mechanism put the power to build the next round of goods in the hands of the people who did the best job with the previous round. It's a smart plan for society as a whole.
Open source software breaks this feedback loop. Any capital thrown off by a successful open source project doens't flow directly back to the control of the original creators. Linus Torvalds doesn't get a few dollars from every version of Linux like Bill Gates gets a few dollars from every version of Windows. So Torvalds doesn't have the money to push the next generation of software in any direction his whims might suggest. He may inherit some social capital and the deep respect of his peers, but that's not the same thing as a war chest of several billion dollars.
The capital is left in the hands of the users who can, if they want, invest their time in helping with the next generation of tools. There's no big feedback loop that piles up the capital in the control of the industry titans. It leaves the capital in many tiny piles where the users may want to invest it as they choose.
This isn't necessarily a bad thing. There are few entepreneurs who have a second act. Many of the wild successful inventors end up squandering their huge piles of capital on ideas that never find the same level of demand. If lightning rarely strikes twice, why bother giving the great entepreneur a pile of capital to waste on building a failure. A number of people love to point out that Microsoft's secondary businesses have always generated losses. The shareholder probably would have been better off if Microsoft never invested in the Zune, the keyboards, the XBox and a number of other perfectly respectable but money losing businesses.
Alas, the tax code throws a monkey wrench into the world. Successful companies are usually better off investing the money in new products instead of returning the excess capital to the investors. The IRS nabs its share of dividends but it doesn't have a mechanism to take a bite from silly or quixotic projects. So society discourages the capital from being reallocated smoothly.
Open source is dodging this problem by living in the post-monetary world where no one has figured out how to tax the gains. Capital flows immediately back to the users where it is shared equally by all. Alas, while this avoids the problem of quixotic capitalism-- let's call it the Zune or the Bob problem-- it has troubles all of its own. The users may not spend the capital with any better insight than the rich entepreneurs. The sysadmin with a smoothly running stack of servers running Apache doesn't always take the spare time to write a new set of tools. He may just fire up the browser and waste time reading gossip blogs, downloading porn, or buying Pez dispensers from eBay. These may be perfectly nice ways to spend the afternoon, but they don't move society forward any more than a vanity project run by a rich entepreneur.
There are also structural problems. Big challenges require big piles of capital. Open source projects don't generate big piles of capital and so that's why there are so few quantum leaps generated by the open source world. Some scientists, for instance, like to argue that the US grant makers at the NSF, the NIH and DARPA should give a few big grants to big organizations instead of spreading it out around to hundreds of universities. Building the atomic bomb, they point out, required a big team like the Manhattan project.
Before closing, I thought it would be nice to issue a few challenges for the open source world. It's not like I'm in charge of it. No one can be. Nor is it even clear that there are any solutions to these challenges that are out there. It's not like a parent telling a child to clean a room, a problem with a concrete solution. These are conceptual problems and they need new ideas and those are rare gems buried deep in the ground.
The simplest solution may be some form of cash.
There have been some experiments with creating bounties that all people to put up some cash to the first person to implement a particular feature. These aren't bad but they haven't succeeded very often. I think it's largely because they're reactive not proactive. In many cases, people don't even know the nature of the feature they want. They couldn't write the spec for the bounty if they tried because they don't even know what they want. They just want the software to be more better and more cooler and more intuitive. But they don't know how to even say how or why.
Wealth appears when people can work easily together and share in the profits. Poverty emerges when the feedback loop is broken or attenuated.
Most of the support for the open source community comes from the big hardware makers. They know that their box of chips is worthless without good software and they know that open source is a fertile ecosystem. So they sprinkle some cash on the mechanism as fertilizer.
But hardware manufacturers can only be interested in selling more hardware. There are large areas of software development that need new sources of revenue that will flow into the open source community.
One of the hidden factors in the success of open source is the way that it encourages people to contribute when the source code is readable and copyable. Many suggest that HTML became a dominant language when the first browsers included the "View Source" feature that allowed everyone to learn from others. Imitation is a powerful tool for information transfer.
Tim O'Reilly once told me that many people focus on the legal issues of ownership when they talk about open source, but he always felt that much of the power came from making simpler, more open interfaces. The open source languages have gotten much easier to read and understand. Ruby on Rails is a powerful and simple tool for building web sites. Javascript frameworks like jQuery make programming much easier. These are all good steps along the path There are still challenges to hiding the right amount of complexity while exposing the right amount of flexibility. We will need to go through three or four more generations of languages before we end up with something simple enough for the average human. Then we will be closer to finding a way for everyone to contribute to the open source revolution.
--Peter Wayner
Baltimore, MD
March 2009