Paul's Internet Landfill/ lj/ This post wants to be free

This post wants to be free

I am currently reading Digerati, by John Brockman. It's about a bunch of technological elites of 1995 (Stewart Brand! Denise Caruso! Bill Gates! David Gelernter!) all talking about The Net(TM), child pornography laws, and each other. A lot of it is pretty bad, and since it was published by Wired the design is bad, too: they use this weird font that "wires" together serifs on letters, and instead of providing real photos of the elites they publish big ugly blocky pixellated pictures. But it has some interesting predictions and insights, and between that book and thinking about LaTeX yesterday I have stumbled on an articulation that is pretty important to my understanding of my values and what I want to be when I grow up.

It goes like this: my computer blows my mind. The idea that I can have such a powerful machine sitting on my desk -- and that I got it for free! -- is pretty magical in itself, but at least I can understand that. People paid real money for this computer, and as newer technology came along the value of my hardware depreciated to the point where it was less valuable for somebody (thanks T and B!) to give it to me rather than sell it on the market.

The software and data on my computer is a totally different matter. I have a complete suite of operating system and applications that I obtained essentially FOR FREE. Furthermore, this software is not a hand-me-down -- it is modern (well, reasonably modern -- I do use Debian stable, after all). I also possess a fair amount of data that I downloaded from the Internet FOR FREE -- from textbooks to research papers to comic strips.

How is this possible? It's possible because people decided to release this software and data to the world over the Internet without charging for it. The reason this can work is because software and data are funny goods: they are what economists call non-rival. Short of slashdotting, if I decide to read perditathecat's blog, then I am not prohibiting you from doing so. In fact, it is the opposite: the value of her blog grows as more people read it. There are some real- world costs to copying software or reading Internet sites, but the copying costs are very small -- small to the point where Internet companies can charge a few dollars a month for Internet access and still make healthy profits.

Note that I am talking about copying (or marginal) costs here, not production costs. The cost of writing a software application is often very high in terms of human effort, but once it is written the cost for distributing it is very cheap.

Most goods and services in the economy are rival goods, which means that they are scarce. Donuts are rival goods: if I eat a donut you cannot eat the same donut (I could regurgitate the donut for you to eat, but then you are consuming a value-added product, not the original good). If I want a donut and you want a donut, then somebody has to make two donuts, and the cost of making each donut is expensive. Our current solution to this is to monetize the economy: we charge money for each donut: the price for a donut must at least cover its production costs. Because the economy is monetized, the donut- makers need to purchase stuff as well, so some compensation for making the donuts need to be factored into the price.

One key observation is that the world of non-rival software and data is also an economy -- one that is different and largely distinct from the monetized economy. The software on my computer is complex: far more complex than any one person could have produced. It is made up of a lot of components that a lot of individual humans put effort into. This is exactly what an economy does: it pools the efforts of lots of individuals to do more than any one person can. Each person who contributed to the software on my computer enjoys the results of that combined effort -- but since software is a non-rival good, I get to enjoy the results of that effort too. Moreover, since copying costs are basically free I get this software FOR FREE.

Make no mistake: the non-rival economy is just as real as the monetized one, in the sense that it generates goods that tangibly improve human lives. Thanks to this software I can write resumes, typeset my thesis, organize my data, scan in newspaper clippings, record my finances, and find all kinds of entertainment. Maybe more importantly, it allows me to communicate FOR FREE with other people all over the world, thanks to e-mail. I can now feasibly apply for jobs in another country almost FOR FREE. The existence of the non- rival economy has probably raised my productivity and improved my life overall -- and this is true for lots of other people who choose to participate in this economy.

Here's another weird thing: when you artificially restrict access to non-rival goods, everybody suffers. If Vim cost money, then fewer people would use it. That would hurt the software quality (because fewer people would report bugs and contribute patches), the user base, the technical support resources (because users would have a harder time supporting each other), and the number of useful plugins other people have written. When you increase the user base of your software or data, pretty much everybody benefits.

What about the original software authors, who put in all the time and effort into writing the software? How are they paid? I see two answers for that, neither of which has to do with fame or branding: in the non-rival economy, the authors don't need to be paid. Alternatively, they have already been paid, because they enjoy the exact same access to software and data that everybody else does.

Think of it this way: the reason we need to monetize the economy is because goods are scarce. Supposedly, you use money to figure out who has put more effort into the economy, so those people can get more scarce goods than lazy people who contribute little. (No, that's not really how it works, but that's another blog entry.) But goods are not scarce in the non-rival economy. Within that economy money does you little good because the goods are available for free anyways.

So what is the incentive to produce software and data? To have better software and data. Every contribution people make into the non-rival economy improves the quality of goods for everybody. There need not be any altruism going on at all.

Don't get me wrong. Not everybody can be a free rider. The non-rival economy is still an economy, and it needs inputs of human energy in order to survive. If nobody contributes anything, then the entire shebang falls apart. Fortunately, people do not have to contribute much in order to make real impacts, because the value of a contribution is amplified so much.

In my eyes, the key to keeping the non-rival economy strong is for people to realize how much they benefit from it, and to maintain this culture where people feel they can make contributions that are valued. For those who don't like to think, we can break this down into two simple rules:

0. When something aggravates you in the non-rival economy, allocate some time to addressing the problem.

1. Look for ways in which you can contribute to the non-rival economy and thus repay it for the value it has contributed to your life.

That's it. That might mean reporting a bug, programming a helper script or plugin, supporting a fellow user, sharing your observation, writing a translation, writing documentation, composing music or logos, proposing software you need... whatever. There are lots of ways to contribute.

If we think of the non-rival economy as being distinct from the monetized one, the question of "How do you make money from open source?" becomes kind of silly. Within the non-rival economy, there is no need to make money -- and in fact trying to map the monetized economy to the non-rival one often does more harm than good.

Yes. There's an elephant in the room. But before I address it, let's at least acknowledge that the non-rival economy is an economy that contributes to human good, and that compensations for contributions within the economy itself are self-rewarding.

The elephant is not that software/data authors are not compensated for their work. They are. The elephant is that the non-rival economy only provides non-rival goods, and people need some rival goods to survive. Even software programmers need to eat, and they usually have to wear clothes, sleep under roofs, and get computers and Internet access. Now opportunity costs come into play: every minute a programmer spends working on software they are going to give away is potentially a minute taken away from a job that pays the bills. How do you address this problem?

I don't have great answers for this, but I find that thinking of the non-rival economy as distinct from the monetized one helps me understand the problem better. It suggests two solutions: keeping activities in the economies separate, or attempting currency conversions.

My personal solution has been to keep my activities in the non-rival economy distinct from those in the monetized one. There's work I do for pay, and there's work I do for fun (which occur both in the volunteer sector and the non-rival economy), and as much as I would like to I combine them I don't think it will be able to happen. This solution has more-or-less worked for me, but it has some implications. It means that I don't make a lot of cash, and that people mock me because I only "work" part-time. It also means that I cannot contribute as much effort as I would like to open source software, writing documentation and essays, or other non-rival goods production. I don't think this solution works for everybody, and depending on my next job it may no longer work for me. But it's feasible for some people.

The other idea is to attempt currency conversions: mapping effort in the non-rival economy into hard currency somehow. Some people and organizations (Hi mynatt! Hi da_lj!) manage to do this with varying degrees of success, but I am worried that we are depending on a lot of bad/unsustainable ideas to try and get open source software written.

The natural temptation is to amortize the heavy production costs of software to the copying costs by charging for distribution. For reasons mentioned above, this hurts the non-rival worth of the software. There may be tradeoffs people can exploit here, but simply charging for one's software does not work well in the long term; sooner or later somebody will release a free alternative of comparable quality.

I think the Trolltech and MySQL people have models that work better. By releasing software under the GPL, you force all derivative works and embeddings of the software to be GPL as well. For companies that are uncomfortable with this (i.e. want to sell their proprietary software in the monetized economy) there is an option to purchase contracts for the software under some other licence. That maintains the non-rival benefits of the software while permitting some income -- but when some competitor comes along with some BSD-licenced alternative, this method of making money fails.

I am worried about the degree to which we rely on our new best friends Sun and IBM to develop open source software for us. These companies are firmly rooted in the monetized economy: they want to make lots of cold, hard cash. But they are spending enormous sums of money to develop software -- Eclipse, OpenOffice, OpenSolaris -- which live in the non-rival economy. How are Sun and IBM making money by doing this? If they aren't making money, then we should be deeply suspicious, because then the funding is not sustainable.

I suppose the efforts of Mandriva and SuSE to package their software distributions and sell them in the software market works okay, but there is a definite cost to the non-rival economy. Unless these companies take special effort to include their consumers in the development and improvement of their software (like Mandrake's old Cooker campaign, which may exist for Mandriva) then we run the risk that people use their software passively. That's deeply dangerous to the success of the non-rival economy: there aren't enough SuSE/Mandriva employees to keep the economy going by themselves -- the efforts of SuSE/Mandriva users can make a huge amount of difference. Probably I am misrepresenting SuSE and Mandriva users; but the same sentiment applies to those who use Linspire, MacOS X and other "end- user" distros. There's nothing wrong with using user-friendly distros; the danger is missing the connection between one's software and one's ability to contribute to the non-rival economy.

Red Hat falls into this category too. It makes money by selling services and support to enterprise customers, while keeping its software open source. By doing so, it takes a hit from repackagers like CentOS and White Box Linux, but the value of the Red Hat name (and associated initiatives like its certification programs) appear to be keeping the company profitable. Notice, however, that although it is a big name in the Linux world, it really is not that big a company -- it's no IBM or Novell or Sun.

Red Hat makes a lot of contributions to the open source world, but it is another company that doesn't seem to be translating its non-rival economic activities into dollars directly.

Then there's Ubuntu and Canonical. I can't figure Canonical out at all. The company is hiring all kinds of expensive developers (Hi sfllaw!) and from what I can tell doesn't have any way to recoup those costs in the monetized economy, so there would appear to be a serious balance of payments issue between the two economies. What's going on? If somebody could point me to an explanation I would be very grateful.

Moving away from Linux, I think Craigslist is an example of a company that balances the non-rival and monetized economies well. They make money by selling job ads in three cities. All other Craigslist activities are free and usable by everybody. In a non-rival sense, charging for job ads hurts the number of people who post ads -- but the market is such that in a monetized sense Craigslist is undercutting the competition. Selling ads finances all of Craigslist's other operations, which allows it to support its activities in the non-rival economy. Meanwhile, it gets eyeballs and brand identity by providing free classified ads all over North America, which makes it a natural destination when people in San Francisco, New York or Los Angeles look for jobs. In essence, Craiglist's non-rival economic activities serve as its marketing campaign in the monetized economies, and it appears to keep them afloat. Furthermore I bet it is not in Craigslist's interests to sell ads in more than three cities; I bet that most of the effort and headaches in the company go towards supporting their paid advertising, and that moving to paid ads in more cities would increase the number of staff they need to employ.

There's a weak spot, of course: if somebody with as many eyeballs comes along and gives away all advertisements for free, then Craigslist could be sunk. Google might want to do this, but I can't think of another company that would want to -- the Yahoobot runs HotJobs already, and charges lots of money per ad.

Craigslist also demonstrates why considering the monetized and non- rival economies separately is helpful. In the non-rival economy Craigslist is huge: it's a well-known website that lots of people trust and use. It generates a lot of value for a lot of people by bringing folks together through classified ads. But in the monetized economy Craigslist is tiny -- their website says that they employ all of 19 people. It turns out that selling job ads in three cities can support the salaries of these 19 people in addition to the server and bandwidth costs of providing a continent-wide message board, but I bet that nobody who works for Craigslist is that rich in the monetized economy. So is Craigslist tiny or huge? In one economy it's tiny, and in the other it's huge. There's no contradiction.

None of these thoughts are new. Eric S. Raymond and Lawrence Lessig and many others have been making these points for years. But today something clicked for me: I am beginning to understand why open source and the Internet might be sustainable, and maybe I am starting to believe that my activities in the open source world are not all hubris and procrastination. Writing this long essay out helped me articulate this realization. My (undoubtedly naive) hope is that by releasing this to the free Internet (and thus as a non-rival good), maybe other people will clarify their understandings as well.

Livejournal URL:

Mood: open