Bottom-up

Reader Rhayader asks a good question:

I love free software, but as someone who’s a novice (at best) when it comes to programming, most of the direct benefits I see have to do with the “free as in beer” side of things. I understand the distinction between the two types of freedom, but have a little trouble seeing how an open source project could generate profitable sales.

It’s true that free-as-in-speech software is almost always free-as-in-beer too. But it’s a mistake to conclude from this that free software is non-commercial software.

I think the way to think about this is to compare it to activities that are more familiar to non-programmers. Take playing tennis. If you’re a really good tennis player, there is a variety of ways you can make money. The most visible one, and the one that’s most closely analogous to what Microsoft and Adobe do, is to play games and charge admission. A related business model is to sell video to a television network which broadcasts the match for free (as in beer) and sells advertising.

But there are lots of other ways good tennis players make money. Manufacturers might pay you to endorse their products. You might go on the lecture circuit or write a book about playing tennis. And probably the most common way to make money as a tennis player is to sell lessons.

The key thing to notice here is that in all of these business models, with the exception of the first one, no one is directly paying you for the act of playing tennis. Rather, the act of playing tennis enhances your skills and reputation, and then you leverage those enhanced skills and reputation to sell other attibutes, like advertisements, speeches, or lessons. Indeed, only a tiny fraction of the most elite tennis players can attract large enough audiences to make money charging admission, while the market for lessons is much larger.

Free software business models work the same way. You give away your software—which is cheap because copies of software have a zero marginal cost—and once your software is popular, you use that popularity to sell them other stuff, like a support contract (Red Hat), consulting work (IBM), hardware devices (Google, Apple), web services (Google, Facebook) and so forth.

This isn’t as weird as it seems. It’s how many information businesses have always worked. Newspapers sell their physical product at cost and make their money selling ads. Television stations give away programming and sell ads. HP sells cheap printers and makes a killing on ink.

I recommend this podcast from the Software Freedom Law Center (previously discussed here), which discusses the history of Cygnus (an abbreviation for “Cygnus: Your GNU Support”–aren’t nerds funny?), a company that was founded to make money providing support for Stallman’s GNU project. One of the basic pieces of infrastructure that every computer system needs is called a compiler. GNU included a compiler called gcc that was popular within the computer industry. When a hardware company created a new hardware platform, one of the first things they need to create is a compiler, and it was cheaper to modify gcc than to build a compiler from scratch. As Bradley Kuhn tells the story, Cygnus literally could not keep up with the demand from firms wanting Cygnus to help them customize gcc for their own hardware platforms. At one point in the late 1980s, they had a waiting list several months long to become a customer of Cygnus’s compiler group, and Cygnus would hire pretty much any programmer with experience hacking on gcc.

Anyway, I think the fundamental point to make here is that computer programming is not an inherently commercial activity (any more than playing tennis is), and there’s nothing magical about the Microsoft/Adobe business model of selling copies. If your goal is to sell copies of your software to people, then a free software license probably isn’t for you. But there are lots of circumstances where people create software for reasons other than wanting to sell copies: for internal company use, for academic research, because they need it to make a hardware product work, as an ad-supported web service etc. In many of those circumstances, the limitations of proprietary software just get in the way.

In the early 1980s a dispute over Xerox printer source code transformed Richard Stallman from a shy hacker to into a quixotic activist. Stallman had cut his teeth in a 1970s programming culture in which it was conventional for programmers to freely share source code with one another. When Xerox donated one of its early laser printers to the MIT lab where Stallman worked, he assumed he’d be able to get the source code in order to improve its functionality. But Xerox had not distributed the source code along with the printer, and it had required those programmers who had been given access to the source code to sign non-disclosure agreements.

In Stallman’s telling, his inability to get his hands on the source code—and his consequent inability to tinker with the printer—radicalized him. He began thinking deeply about the implications of proprietary software culture, and he decided he didn’t like where things were headed. And so in classic hacker fashion, he decided to do something about it: he started writing an operating system, from scratch, that would be free of legal restrictions. Today, the software he wrote, called GNU (which stands for “GNU’s Not Unix”), is an important component in virtually all non-Microsoft operating systems.

At the same time he was building his operating system, Stallman was founding the Free Software Foundation and developing the philosophical foundations of the free software movement. He articulated four software freedoms: the freedom of users to use, examine, modify, and share the software they used. Stallman recognized before almost anyone else the profound difference between the culture of freedom he had come to take for granted and the culture of control that was then taking root in some sectors of the software industry.

It’s worth pausing here to underscore the importance of what’s commonly described as the “free as in speech vs. free as in beer” distinction. One of the defects of the English language is that we use the same word, “free,” to describe both things that have zero price and things that are free of external constraints. “Free speech,” “free trade,” “free markets,” and “free will” are all examples of this second sense of free. “Free beer” is an example of the first sense.

Non-programmers have a tendency to mis-interpret “free software” in the “free as in beer” sense. Because software is an inanimate object for most users, like a chair or a pencil, the idea that freedom could be an attribute of software seems non-sensical. Non-programmers largely lack the ability to exercise Stallman’s four freedoms, and so I think it’s hard for them to imagine that anyone could really find them so important. This leads to a suspicion that Stallman’s talk about freedom is really a smoke screen, and that the real agenda is a hostility toward paying for things.

But this isn’t the way things look to programmers. For us, software isn’t just a magical icon in our start menu that we click when we want to accomplish some task. If you understand how software works, it’s natural to want to crack it open and see what’s going on under the hood. And for programmers of a certain bent, being denied the opportunity to do this is felt as a tangible loss of freedom. It means that they’re forced to passively accept that their computers will behave the way some programmer in a distant company wanted them to behave, rather than being able to customize them to suit their own needs. It means, in other words, surrendering some measure of autonomy to Redmond or Cupertino.

Stallman’s work has been tremendously influential. Not only is GNU a part of almost all modern operating systems, and not only is much of the software that makes the web work available under the free software license he created, but Stallman’s work has profoundly affected the way people think about culture and the law. Larry Lessig has credited Stallman for inspiring the ideas behind Free Culture and Creative Commons, a book and an organization, respectively, that aims to create a culture free of legal restrictions.

Stallman’s free software crusade is a quintessential example of the kind of liberal project I wrote about in my last post. The free software movement is philosophically individualist; it’s devoted to enhancing users’ freedom to do as they please with the computers they own. A world dominated by free software is one in which users have more freedom—more opportunity to do as they please without interference from third parties—than one dominated by proprietary software. And in my experience, the free software movement is second only to the libertarian movement in its obsession with freedom as its central value. One of the reasons I like the free software movement is that, like libertarians, they talk about freedom incessantly.

And as I’ve argued before, I think libertarians have particular reasons to celebrate the free software movement. When Stallman became alarmed at the rise of proprietary software, he didn’t run to Washington seeking legal privileges for free software or legal restrictions on proprietary software. He didn’t start lobbying for subsidies or tax breaks. Instead, he rolled up his sleeves and started building an alternative. Stallman’s work is the kind of entrepreneurial success story that libertarians love.

Unfortunately, not all libertarians see things this way. Free software has attracted a considerable amount of misguided criticism from libertarian and free-market intellectuals. In my next post I’ll take a close look at the latest example and argue that it illustrates the perils of libertarians confining themselves to a “thin,” state-centric conception of individual liberty.

My friend Kerry Howley’s has a fantastic essay about the relationship between individualism and libertarianism:

I call myself a classical liberal in part because I believe that negative liberties, such as Min’s freedom from government interference, are the best means to acquire positive liberties, such as Min’s ability to pursue further education. I also value the kind of culture that economic freedom produces and within which it thrives: tolerance for human variation, aversion to authoritarianism, and what the libertarian economist F.A. Hayek called “a preparedness to let change run its course even if we cannot predict where it will lead.”

But I am disturbed by an inverse form of state worship I encounter among my fellow skeptics of government power. This is the belief that the only liberty worth caring about is liberty reclaimed from the state; that social pathologies such as patriarchy and nationalism are not the proper concerns of the individualist; that the fight for freedom stops where the reach of government ends…

“True libertarianism is not cultural libertarianism,” the philosopher Edward Feser wrote on the paleolibertarian website LewRockwell.com in December 2001. This statement was immediately preceded by a call for the stigmatization of porn, adultery, divorce, and premarital sex—in other words, an argument for a particular kind of culture. Feser claimed that small government and an ethos of “personal fulfillment” were incompatible, and he argued for the former over the latter. In the guise of an attack on cultural libertarianism, Feser demanded that libertarians espouse different patterns of cultural behavior.

As it turns out, all libertarians are cultural libertarians. We just don’t share the same agenda. Some prefer to advance their agenda by pretending it doesn’t exist: that social convention is not a matter of concern for those who believe in individual liberty. But when a libertarian claims that his philosophy has no cultural content—has nothing to say, for instance, about society’s acceptance of gays and lesbians—he is engaging in a kind of cultural politics that welcomes the paternalism of the mob while balking at that of the state.

Two other commentators, Todd Seavey and Daniel McCarthy, square off against Kerry, arguing that it’s a mistake for libertarians to concern themselves with forms of oppression that don’t involve the state. Sure, Warren Jeffs, the head of an oppressive Fundamentalist Church of Latter-Day Saints community, may have been a bad guy, they seem to say. But there’s nothing distinctively libertarian about criticizing his misdeeds, provided that the members of his community were there voluntarily.

Libertarianism is commonly described as a political philosophy that favors eliminating “force” from human relationships. Unfortunately, I think the libertarian movement has inherited from Rand and Rothbard an allergy to giving deep thought to the question of what force is and why it’s bad. Rather, the definition and wrongness of “force” is taken as a self-evident “non-aggression axiom.” And all libertarian conclusions are said to follow from the axiom: just figure out who is forcing whom (almost always the state is doing the coercing), and make them stop.

Unfortunately, when you adopt a political philosophy that fits on a postcard, it leaves a vacuum that is filled by whatever inchoate political prejudices you held before you picked up that copy of Atlas Shrugged at the age of 19. And if your political beliefs say that nothing besides state coercion matters, that means you’re likely to underweight the importance of injustices perpetrated by parties other than the state. Like Kerry says, there are many social pathologies that can’t be traced directly to a particular action of the state. The Rothbard/Rand formulation of libertarianism provides no real guidance on how to think about these topics.

Kerry points us toward a more promising approach that views libertarian politics as one facet of a broader liberal worldview. This worldview starts with a commitment to individual liberty for all, and then looks for projects that will enhance the liberty of individuals. Limiting the power of the state, which frequently impinges on individual freedom, is one such project. But there are many others. Fostering a culture of gender equality, so that women do not face hostility for choosing careers that are considered non-traditional for their gender, is a liberal project. Ditto for opposing prejudiced attitudes toward gays and lesbians.

Liberal projects also don’t have to be explicitly political. Expanding literacy and access to information is a liberal project, because better education gives people greater autonomy. Andrew Carnegie’s 19th century effort to build libraries around the country was such a project. Wikipedia and the Internet Archive are modern-day heirs to this tradition. They’re working to achieve the liberal dream of making the world’s information is freely available to everyone.

There are also liberal projects that are narrowly focused on combatting particular types of oppression. Battered women’s shelters ensure that women in abusive relationships have the opportunity to leave them. Anti-scientology activists help to publicize the authoritarian character of scientology and helps scientologists who want to escape do so.

Now, to be clear, none of these projects are libertarian, as such. It’s possible to be a libertarian and believe that, say women should get back in the kitchen, and gays should get back in the closet. Theoretically you can be a libertarian scientologist, or a libertarian who thinks libraries are a waste of money. But I think that if you’re a libertarian because you care about liberty, rather than simply being a libertarian who hates the government, then you ought to be interested in a broad array of liberal projects. It’s good and important to criticize the state when it does illiberal things. But non-state actors do illiberal things too, and our silence and inaction on those topics contributes to the perpetuation of those injustices. It’s a cramped kind of liberalism that only cares about threats to liberty that come from the state.

Last year I did a three part feature on the future of driving, and how the emergence of autonomous navigation systems could change society. Computer science researchers have always demonstrated prototypes of cars that can drive without human intervention, including following traffic laws and interacting correctly with other drivers. However, these systems still need work to make sure they can handle tricky situations like pedestrians and inclement weather.

One of the big questions is how the first self-driving cars will be introduced on the road. Obviously, a car company that releases such a product before it’s fully ready would be in a world of hurt. And there’s a serious risk that a premature release will lead to ill-considered regulations that prevent the emergence of self-driving technologies that would save lives in the long run.

Brad Templeton has argued that self-driving vehicles will probably sneak up on us gradually, as car manufacturers gradually add new sensors and intelligent navigation features to their vehicles. The first such technologies—cruise control and antilock braking—have been with us for a couple of decades. Now we’re starting to see more ambitious navigational features. For example, I happened to see an ad for the latest Mercedes vehicles. Their features include:

“Blind spot assist” which has sensors to monitor the car’s blind spot and alert the driver when a car is in the way.

Adaptive cruise control, which monitors the distance to the next car and automatically adjusts speed to match.

Night vision, which uses an infrared camera to give the driver an enhanced view of the road ahead.

Attention assist, which triggers alarms if it detects that the passenger is falling asleep.

Now, obviously these don’t add up to a self-driving car. But you can easily see where this is going. Combine the adaptive cruise control and the blind-spot detection, and it’s not too hard to imagine a comprehensive crash-avoidance system, which detects imminent accident conditions and takes control (slamming on the brakes, swerving, etc) in time to prevent disaster. Indeed, we may reach the point where the safest thing to do in the face of an impending accident is to let go of the wheel and let the car drive itself to safety.

This interview with my friend Reihan Salam is worth a read. Reihan is generally regarded as a conservative, but he is one of those guys who seems to have read absolutely everything, and he seems to incorporate a little bit of everything he reads into his own views. Thus, he describes himself as something of a libertarian neocon:

The boundaries between the various traditions are, by nature, extremely ill-defined. I definitely identify with a lot of the 1970s-era neoconservatives like Irving Kristol, who embraced certain aspects of the welfare state while also emphasizing the limits of what planners and governments can realistically achieve. Kristol was a great admirer of Jane Jacobs, who led the fight against Robert Moses’s various efforts to raze old neighborhoods in New York city in the name of “rational urban planning,” in the process obliterating flourishing institutions that were hard to discern from a planner’s-eye-view.

I’m a deep believer in the power of bottom-up institutions that are controlled by families and individuals. I worry about sapping authority from ordinary people because I think it undermines our capacity to experiment and to create new and better solutions to social problems. I definitely think the state can and should help give people the tools they need to improve their lives — and sometimes this just means the state can tax some people and give other people money. But again, we have to have a clear sense of what governments can and can’t do well.

In this regard, I’m definitely a great admirer of Hayek, who was a believer in the power of decentralized approaches to achieving social goals. That’s not to say that I’m a doctrinaire libertarian. I recognize that, for better or for worse, richer societies demand more in the way of social protections — actually, Hayek understood this too, and he noted that the basic social minimum guaranteed by the state would vary across societies according to overall wealth and thus overall expectations.

People think of neocons and libertarians as very much at odds, not least because the former are very interested in the idea of promoting “virtue,” in a non-sectarian way. While I think that neoconservatives have taken lots of wrong turns — if we can even talk about this as a coherent category outside of foreign policy — I tend to think of the 1970s-era folks as social policy realists.

One of the embarrassing oversights on my own intellectual resume is that I’ve never actually read the work of the early neocons. I’m associated with an institution whose president uses neocon as a term of abuse, so my reflexive impression of neocons isn’t especially positive. And my view of them hasn’t been improved by the fact that the neocons’ signature policy achievement, the war in Iraq, hasn’t exactly been a shining success.

Still, one element of neocon thought that I do think is worth taking seriously is the point that after a certain amount of time has passed, government programs become deeply woven into the fabric of a society. It’s one thing to argue that, ex ante, the creation of Medicare was a bad idea. But Medicare is now a settled part of the American economic system. Clearly, reforms are needed to ensure the program doesn’t bankrupt the country. But I think libertarians who advocate the immediate abolition of Medicare are guilty of the same kind of top-down hubris that they attribute to those who advocate dramatic expansions of the role of government in health care. Either reform is likely to have unintended consequences, and it’s prudent to approach such reforms cautiously and—if possible—pursue them incrementally.

This, I think, is the hidden wisdom of the now-famous “keep your government hands off my Medicare” comment. Ordinary voters are not policy wonks, and they don’t tend to think about issues in the same crisply ideological terms as policy wonks. But they have a gut-level understanding that dramatic changes in the health care system have the potential to disturb a health care system that’s currently serving seniors pretty well. From the perspective of the average senior, there isn’t that much difference between Medicare and Aetna—they’re both huge bureaucracies that decide which kind of health care you’re allowed to have. Seniors understand that legislation means significant changes, and since they’re relatively happy with the status quo, they’re understandably suspicious of proposals to change it.

A friend was kind enough to send me this post about bottom-up thinking in development economics:

What must we do to end world poverty? There has been a search for sixty years for the right answer. Now most economists confess ignorance how to raise the rate of economic growth — how to progress more rapidly towards development and the end of poverty.

To get out of this dead end, I would respond to this question with more questions.

First, who is “we”? It seems like whoever “we” are, “we” must have unconstrained power to implement “the answer”, so “we” sounds like authoritarian leaders (national autocrats or World Bank officials dictating conditions).

Second, are “we” going to allow poor people to choose their own paths? Of course not, because “we” already know the “right answer” for them.

So this question only makes sense in approach to development that is authoritarian and paternalistic, using Top Down Planning, which in fact has been the prevailing – but unsuccessful – approach to development for six decades.

The paradox of development economics is that Development does NOT require any one person (Expert, Leader, or Aid Official) to have a comprehensive understanding of how to achieve Development (sort of like how evolution managed to happen on its own before Darwin).

I’m not an expert on development economics, but this seems like generally the right way to think about the subject. Any development solution that requires millions of people to conform to somebody’s official plan is likely to be failure.

I have mixed feelings about the news that my former colleague and co-blogger Adam Thierer is the new president of the Progress and Freedom Foundation. On the one hand, I know he’ll do a great job for PFF. He did a great job running Cato’s tech policy program when I was a staff writer there, and he’s done a lot of great research for PFF since he went there.

On the other hand, running a think tank probably leaves him a lot less time to do actual writing and research. And nobody covers Adam’s core subjects—free speech and child protection online—quite the same way Adam does. Adam’s also the backbone of TLF, producing excellent content including book reviews, rankings, and 3000-word posts on libertarian philosophy. So I’m hoping Adam finds a way to run PFF without entirely abandoning the stuff he’s done so well over the last 4 years.

A couple weeks ago, Mike Masnick pointed to an interesting post-mortem of the Rocky Mountain News by John Temple, a former RMN staffer:

We knew the web was a place we needed to be, but we didn’t have a clear strategy, mission, or objective. It was a complement to the paper, as we said in our initial “about us” page. Which brings me to lesson 3. You have to have a strategy, and you have to be committed to pursuing it. We perceived the website as a newspaper online, as a complement to the paper, not as its own thing. That’s not a strategy. Senior management’s focus in the 1990s was on keeping the newspaper alive. Again, to be clear, that’s understandable, at least to a point. We were fighting for our lives, and the money then lay in print. We didn’t understand the web or new technology, and didn’t have the time to learn much about it. We weren’t a consumer-driven company, except that we knew that our priority was to get papers on the porch on time in the morning. Otherwise, we feared our subscribers would switch to the competitor.

The thing that I found puzzling about this is that they actually did have a web strategy. In fact, they had several of them. Temple tells us that, at several points over the last two decades, the Rocky unveiled new web-based initiatives. Indeed, the Rocky launched its first online edition in 1990, with a pre-web dial-up service. The service didn’t seem to be improving the company’s bottom line, so it was shut down after a few months. This pattern was repeated in the following years: new websites would be launched, and each time management would discover that the web didn’t generate significant revenue and decide to devote the bulk of their resources to the print edition.

Temple portrays this as strategic fecklessness, but I think a more charitable explanation is more plausible: the Rocky actually was a consumer-driven company. Consumer-driven companies focus on the needs of their paying customers, and to this day, a newspaper’s paying customers are predominantly print subscribers and print advertisers. It would have been crazy for the Rocky to ignore its paying customers and bet the farm on a new technology with no obviously prospect of generating significant revenue. Experimenting with new technologies made sense, but it also made sense to scale back those experiments when they failed to pan out.

Temple seems to believe that a better web strategy could have saved the Rocky, but I don’t actually see much reason to believe that. The New York Times and the Washington Post were both early leaders in online content. They have the luxury of being national brands, and they’ve both built fantastic websites with millions of readers. Yet that hasn’t stopped either one of them from experiencing the same precipitous financial decline that has afflicted the rest of the newspaper industry. The harsh reality is that no one has figured out how to transform a daily newspaper into a profitable online business.

Temple seems to feel there was more the Rocky’s management could have done to forestall the paper’s demise. And obviously with the benefit of hindsight we can look back and point out individual tactical mistakes. But I suspect that ultimately, the Rocky was crushed by the same broad and near-irresistible economic forces that is crushing its sister publications across the country. Its management appears to have done more or less what it should have. That they didn’t find a profitable web strategy isn’t necessarily the fault of those managers; perhaps there was simply no web strategy to be found.

My friend Sarah makes a point I’m embarrassed I didn’t think of myself:

Bad charters don’t stay open indefinitely because no one has to enroll. Not everyone will necessarily pull kids out of a bad charter, but new people will be less interested and will go other places.

Most bad charters see they’re starting to lose people and then imitate the more successful charters, and improve. It’s a really small percentage (I believe Caroline Hoxby says ~5 percent) that never get with the program.

So… closing bad charters is built into the bottom-up charter school process. It doesn’t have to come in the form of a “Close thou evil charter school!!!” order from above that some charter critics envision.

One of the difficulties of many performance-oriented school reform proposals, such as merit pay is that it’s difficult to measure teacher and school performance quantitatively. So if some public official is in charge of identifying the under-performing charter schools and shutting them down, there’s a risk that they’ll use flawed quantitative measurements and shut down schools that have strengths that aren’t reflected in published statistics—maybe teachers are spending a lot of time helping kids with difficult family situations that will benefit them in the long run but don’t show up in this year’s math test. So a more effective approach is to decentralize the decision about which schools grow and which ones fail by letting the parents pick. Parents see their kids every day, and so they’re likely to have a pretty good idea of whether the school is serving their own kids well. (Obviously some parents lack the education, time, or interest to judge effectively, but these parents are in the minority) And in the aggregate, whether more or fewer parents want to send their kids to a school is likely to be a pretty good indicator of whether the kids are doing a good job.

Also, to address another point made by a couple of readers: the phrase “cheap failures” is infelicitous in this context. I definitely don’t mean to suggest that a failed charter school is painless for the kids in that school. Obviously, it can be quite damaging. Rather, my point is that the charter school model is relatively cheap (in terms of damage to kids’ educational futures) compared to the major alternative: public schools whose kids often have no realistic alternatives, and which are never shut down regardless of how badly they do. If a school is going to fail (and unfortunately, it’s inevitable that some will) I want them shut down in a reasonable amount of time to limit the number of children who are harmed. No plan can guarantee that no children will be harmed, but charters reduce the number of children who get harmed relative to the system that existed before they came on the scene.

Some excellent bottom-up thinking on education policy from Matt Yglesias:

There’s substantial variation in the performance of different charter [schools]. What you need to do is identify schools that consistently perform poorly and shut them down. Then you create space for more effective models to replicate themselves and also for new ideas to be tried out. The promise of charter schools is that by allowing more experimentation we’ll find some good models. But it’s not as if public education in the United States currently achieves some theoretical maximum of badness—with experimentation we’re also discovering bad models. You’ve got to shut those models down, while at the same time curbing state legislators’ tendency to impose arbitrary numerical caps on the total quantity of charter schools. We should let a thousand flowers bloom and then kill 20-30 percent of them if they turn out to look ugly.

I’ve written before about the importance of cheap failures for innovation. Charter schools are a strategy for improving education by making it cheaper and easier to create new schools and experiment with new educational techniques. But this strategy will only work if it’s also relatively easy to shut down schools once they’ve failed to pan out. If charter schools are allowed to stay open indefinitely, regardless of performance, then it’s totally unsurprising if charter schools fail to out-perform traditional public schools. Innovation requires experimentation, competition, and a willingness to cut our losses once we’ve determined that a particular experiment isn’t working.