Bottom-up

The Software Freedom Law Center is a public interest law firm that helps free software organizations with legal issues. They’ve got an excellent podcast that dives into detail about the legal issues faced by free software projects. I particularly enjoyed show 0x14 (yes, they number their shows in hex), which talks about free software business models.

There’s a common misperception (shared by some libertarians, unfortunately) that free software is hostile to business and markets. I think this misperception flows from a deeper misunderstanding of how the software industry works. When the typical consumer thinks about the software industry, they think about Microsoft and the shrink-wrapped boxes at Best Buy. Obviously, having copies of your software available for free limits the potential of “sell copies” as a business model. Fortunately for the free software movement, most of the software industry does not work like this. For example, I worked for a firm this summer makes money producing custom software for clients’ websites. Google lets customers use its software for free and sells advertising. Red Hat develops software and uses it to sell services and support. Still other firms write custom “embedded” software: software that comes pre-loaded on hardware products like cars or medical equipment. All of these businesses are totally compatible with a free software licensing, and indeed, free software is becoming increasingly common in all these businesses. The project I worked on this summer, for example, is built with PHP and MySQL, both of which are free software.

In the podcast, Bradley Kuhn makes the point that that the “free” in free software means “free as in market” in addition to “free as in speech” and “free as in beer.” The point of “copyleft” licenses like the GPL is to prevent people from restricting end-users’ freedom to modify and redistribute the software, it’s not to prevent anyone from making a profit in the process.

There’s also an interesting discussion of the relationship between the structure of a free software community and its long-run viability. Some free software communities are built around a single organization. The MySQL community (at least before the Sun acquisition) is a good example: A lot of MySQL development work is done by MySQL AB, the company. Other free software projects, including Linux and Apache, are collaborations among dozens of firms, large and small. Kuhn makes the point that free software communities of the latter type tend to be healthier in the long run. We’re seeing the problem MySQL now: Oracle’s imminent acquisition of Sun (which previously acquired MySQL) raises the distinct possibility that Oracle might do a half-hearted job of developing a product that has the potential to cannibalize its more lucrative proprietary database products. And because MySQL development has traditionally been so centralized, there may not be a ton of other companies ready to take up the slack. Eventually, there would surely be a fork of the MySQL codebase and a continuation of the project outside of Oracle. But problems of this sort aren’t even possible in a decentralized open source community like Apache. Buying and shutting down any one firm in the Apache community would have a negligible effect on the health of the overall project.

One way to look at this is that MySQL lies in the middle of a spectrume of sftware development strategies that range from highly decentralized projects like Apache to highly centralized efforts like Windows Vista. The general trend over the last 20 years has been toward the decentralized, bottom-up end of this spectrum. This shouldn’t surprise us. As we’ve seen in the newspaper business, large, hierarchical organizations tend to have systematic inefficiencies relative to groups of smaller, decentralized firms. The proprietary business model requires software products to be centrally planned by a single firm, so it has to be profitable enough to overcome these diseconomies of scale. Free software can be (although it isn’t always) developed in a decentralized fashion. Freed of the costs and complexities of top-down planning, decentralized free software communities have a tendency to be both more nimble and more frugal than proprietary firms.

This isn’t to say, of course that proprietary software business models can never work. Obviously they can and do on a regular basis. It’s not even to say that free software will generally be more profitable. Indeed, because free software markets tend to be more competitive, the opposite is generally true: Microsoft can charge a lot more for Windows than Red Hat can charge for its version of Linux. But it’s important to distinguish the fate of any individual participant in a bottom-up system with the health of the system as a whole. Like other bottom-up systems, a free software community is robust to the failure of any one firm. In contrast, if Microsoft screws up the next version of Windows, customers are at Microsoft’s mercy. This is good for Microsoft, but it’s bad for Microsoft’s customers. And as a consequence, I suspect the long-term trend will be continued growth of decentralized—which is to say free—software communities at the expense of proprietary software firms.

So there’s no confusion, I think it’s entirely reasonable for ACORN to look the other way if an adult prostitute comes looking for help obtaining housing. But assisting child prostitution is a different story. In latest ACORN video, from San Bernardino, the couple is much clearer that the business involves importing 14-year-olds and letting customers beat them. And the ACORN employee in San Bernardino didn’t blink:

Child prostitution is reprehensible, and it’s inexcusable that the ACORN employee here didn’t throw them out of her office and/or go to the police. And the bit where the ACORN employee “admits” to murder is surreal. I guess I’m glad the police think that she’s a liar rather than a murderer.

Over at Cato’s blog, I follow up on my Monday post on the Google Book Search deal. If you’ll forgive me for quoting myself, my bottom-line point is that the class action mechanism is the wrong way to deal with the many issues addressed in the settlement:

The judicial process works well precisely because it typically makes decisions on a case-by-case basis, fitting the circumstances of each case to an evolving body of precedent. This incremental approach tends to produce a body of law that adapts well to changing circumstances while giving all affected parties the opportunity to have their interests represented. Because different cases are heard by different judges, the mistakes of any one judge won’t unduly influence the direction of the law’s evolution. The class mechanism tends to undermine these beneficial properties of our legal system. Rather than many cases being decided by many judges over a period of years, a class action lawsuit asks a single judge to render justice for thousands of plaintiffs whose individual interests can’t possibly all be represented by the attorneys presenting arguments to the judge. Especially when the proposed class is as large and heterogeneous as the plaintiffs in the Google Book Search case, the class action mechanism demands that the judge to balance the competing interests of thousands of different parties, many of whom have divergent interests. No single person could possibly weigh all the competing arguments in a systematic fashion.

A few years ago I wrote a paper for Cato opposing the DMCA’s anti-circumvention provisions. One of my key arguments was that the courts had done a good job of adapting copyright’s fair use doctrine to rapid technological change, and that Congress’s track record was much less laudable. In a series of important cases, the courts upheld the legality of the VCR, reverse-engineering for interoperability, the MP3 player, and image thumbnails for search engines, while simultaneously (and correctly, in my view) ruling against peer-to-peer services that were simply profiting from the infringing activities of their users. Congress’s track record hasn’t been too hot, as the pair of atrocious bills enacted in 1998 attests.

The difference, I think, is that the courts create law using the incrementalist, precedent-based process we call the common law. Obviously, courts have to rule consistent with existing statute when applicable, but when the relevant statute is vague or silent, then judges have to fill in the gaps with their own judgments. Like any bottom-up system, judge-made law evolves incrementally. Judges are expected to stick as much as possible to applying existing caselaw to the facts in front of them. If the facts are different from those addressed in earlier cases, then the judge extends the principles of earlier rulings to the new cases. This ensures that caselaw remains firmly grounded in reality, because a change in doctrine typically only occurs in response to a change in the “facts on the ground.”

Like any bottom-up system, the common law process also makes use of competition. When social or technological changes force judges to confront questions that have not been previously considered by the courts, judges may not always see eye to eye on the best way to accommodate the law to the new circumstances. In these circumstances, the first few judges to rule in an area may push the law in different directions. Subsequent judges then exercise a kind of peer-review, choosing the past precedents they feel best fit the facts in front of them. Over the course of many cases, individual cases tend to coalesce into a nuanced and cohesive body of law.

This process of judicial competition and peer review is especially important to the Supreme Court, which can’t possibly review every decision by a lower court. One of the things the high court looks for is “circuit splits”—cases where two different appeals courts have developed conflicting bodies of case law. These are good cases for the Supreme Court to take both because it’s important that the law be consistent across the country, and because it saves the Supreme Court work: it can pick and choose between the competing approaches, rather than having to fashion a new doctrine from scratch. I’ve argued that one of the problems with patent law is that Congress foolishly consolidated jurisdiction over patent appeals into a single appeals court. Without competition among circuits, it has been more difficult for the Supreme Court to oversee the Federal Circuit’s work.

The problem with the class action mechanism is that it eschews this kind of incremental, decentralized decision-making in favor of a kind of judicial central planning. Rather than trying to resolve a dispute between a single plaintiff (or at least a manageable number of them), and a single defendant, the judge is now being asked to make decisions about plaintiffs whose interests may be totally unknown to the judge. This undermines a key virtue of the judiciary: its ability to gradually adapt the law to changing circumstances. And it introduces a single point of failure: if the judge screws up, it’s going to be much more difficult to undo the damage.

Reader Rhayader points to the recent ACORN prostitution scandal as an example of investigative journalism being done by people who are not journalists in the traditional sense. The guy who made the video is James O’Keefe, an “activist filmmaker” who has made videos promoting various conservative causes. It’s not clear exactly how Mr. O’Keefe makes his living, but it seems clear that he’s not a journalist in the traditional sense.

Now, I don’t actually think what the ACORN employees did here is all that scandalous. Obviously, people should follow the law, but I don’t think it’s necessarily ACORN’s job to rat out, or even refuse service to, people who it knows are breaking the law. We certainly don’t apply that standard to most other professions. If a doctor has a patient who admits to illicit drug use, it’s actually considered unethical for him to report that to the authorities. This is because if we started requiring doctors to report drug users to the police, the only result would be that drug addicts would stop seeking medical attention. I think the same principle applies to ACORN. ACORN is in the business of helping poor people. Sometimes poor people turn to drug dealing or prostitution out of desperation. I don’t think there’s anything wrong with ACORN turning a blind eye to those kinds of activities.

But in any event, I think this illustrates the extent to which people outside the traditional journalism profession can perform the functions of a traditional journalist. O’Keefe made a video and posted it on his blog. The coverage by mainstream media outlets like the Associated Press has largely consisted of summaries of the contents of the video. The role of professional journalists here is not to go out and track down information; rather it’s to take information that has already been tracked down and delivered by a third party, and summarize and synthesize that information in a way that lets readers quickly make sense of it. This is an important function, to be sure, but it’s not a function that requires large amounts of time or resources.

I don’t think we’ll ever reach the point where all stories are like this, but more and more stories are. Certainly court decisions, company product announcements, athletic events, scientific and medical discoveries, political speeches and debates, are among the categories of news that can largely be covered based on materials readily available online. Summarizing and analyzing a document or video is an important activity (much of my writing for Ars has consisted of summaries and analysis of bills and legal decisions), but it’s relatively simpler than the sort of thing Woodward and Bernstein famously did, where they’d spend hours tromping around Washington digging up documents and interviewing sources on background. The Internet is shifting the balance from this kind of “pavement-pounding” journalism to what Jonathan Chait has dubbed “ass-welt” reporting.

Berin Szoka links to this article by Brad Feld promoting Paul Graham’s idea for a founder’s visa. Under this proposal, there’d be a special route for immigrating to the United States for foreigners who found startup firms. Berin makes the argument well:

What an absurd country we live in: We accept, for better or worse, massive illegal immigration across our porous southern border as a fact of life, but can’t muster the political will to give legal status to the most creative and innovative from around the world drawn to the Land of Opportunity made possible by capitalism. So, being dutiful and law-abiding, these “Talented Tenth” go home to suffer under the dead weight of bureaucracies even more oppressive, incompetent and corrupt than our own. How sad.

I’m completely on board with this in principle, but I’d be reluctant to get behind Graham’s specific proposal because it seems awfully narrow. We should be admitting hundreds of thousands more people to the United States every year; a visa that only lets in 10,000 people a year is a relative drop in the bucket compared with the scope of the overall immigration problems. The H1-B program, for example, admits 65,000 people per year, and that comes nowhere close to meeting the demand for high-skilled workers. If the choice is a founders’ visa or nothing, I’ll take the founder’s visa. But I think narrow visa programs like this can have the same effect on the politics of immigration that narrow tax credits have had on the tax code: many interest groups that would otherwise be supportive of broad immigration reform might decide to focus on defending their narrow exemption, with the consequence that narrow exemptions are all we get. We need an immigration reform that liberalizes and simplifies the immigration rules, not a restrictive immigration system that looks like Swiss cheese.

In addition, I think Graham and Feld underestimate the difficulty of enforcing the “startup” requirement. Permanent residence in the United States is an incredibly valuable asset, and people will go to great lengths to secure it. I suspect there are many foreigners with the necessary connections to borrow $250,000 and draft a plausible-sounding business plan that meets the legal requirements. One can imagine, for example, “startups” that mostly do contract work for large companies, making the startup visa a backdoor substitute for an H1-B visa. Now, I’d like to see the H1-B program expanded, so this doesn’t bother me as such. But it would give critics of the policy a lot of ammunition. And it would raise the possibility of immigration bureaucrats micro-managing startup founders’ business plans, both before and after admission, which would almost guarantee that they’d all fail.

I’d much rather focus on promoting Brad Feld’s proposal to grant permanent resident status to anyone who graduates from an accredited college with a bachelor’s degree or better. (He wants to limit it to computer science, while I’d like to see it applied more broadly) This would affect vastly more people, would be harder to game (creating a sham four-year college in the US is a lot more work than creating a sham startup), and I think it would still be a relatively easy sell politically, since I think voters have an intuitive sense that it’s stupid to educate a bunch of foreign students here and then force them to go be productive, tax-paying citizens in their home countries.

Over at Cato’s blog, I make the case against the impending Google Book Search Deal.

Brian Moore has been doing a fantastic job over at my previous blog, and his latest post on the doctor shortage is particularly good. Brian and I are both married to doctors who are currently suffering through residency, so we have a very personal perspective on what’s wrong with our medical system. Brian’s a little harsher than I would be, but this is basically right:

The medical profession is not very forgiving to part-time employment (there are high fixed operating costs). You can’t phase out slowly like you might be able to with other jobs — and being a physician in your 50’s means you have been working very hard for at least 28 solid years. The “lifestyle” issue is very big for both men and women.

As we value leisure and lifestyle more and more, becoming a doctor is becoming less and less appealing. Not everyone is willing to accept major curtailment of friends, family, hobbies, relaxation, sleep, dating and kids during the best years of their life — and make no mistake, that is what you accept if you are going to medical school and residency. Especially during internal medicine residencies (the one all those “primary care physicians” we want to have will be enduring) you are looking at 70-100 hour work weeks (yes, it’s “capped” at 80 hours officially, but that’s like saying drugs are illegal, so no one will have them).

At the most competitive residency programs, working 100 hours a week doesn’t mean just losing the things above, it means losing things like sitting down, eating, any sleep at all, and going to bathroom when you need to. If you want some scary statistics, look up how many residents abuse alcohol, prescription drugs, are on anti-depressants, are in therapy, etc…

The only reason we’re not facing an even more massive shortage right now is that residency programs are highly incentivized (up to including yearly 6 digit bribes from medicaid/care per resident) to retain residents they’ve conned into coming to their programs, and so have erected very high barriers to exit, or even changing specialties. Plus there are natural barriers, such as the sunk cost fallacy that “all those years would be wasted if I left” and your student loans, which can range up to a quarter of a million or more.

The sharp limits on the number of people who are permitted to be doctors means that they are simultaneously overworked and (after enduring 3-7 years of residency) overpaid. If they were given the option, many doctors would choose to work fewer hours in exchange for lower wages, but the way the medical profession is structured makes it difficult to do that. Meanwhile, there are lots of people, both here and abroad, who would like to enter the medical profession in the US but are prevented from doing so by a variety of legal restrictions. If Obama is serious about “bending the cost curve” over the long run, one of his top priorities ought to be expanding the number of doctors in the US—this could be accomplished both by liberalizing licensure here at home, and by liberalizing immigration rules to allow more foreign doctors to practice here. Of course, he won’t do anything of the sort because the medical profession is well-organized to resist any reforms that might place downward pressure on their wages.

I haven’t been able to take anything Thomas Friedman writes seriously since reading Matt Taibbi devastating takedown of his 2005 book, The World is Flat. As Taibbi puts it, “Friedman is such a genius of literary incompetence that even his most innocent passages invite feature-length essays.” But if Friedman’s columns are rarely enlightening or persuasive, they at least have a lot of potential for humor. His latest does not disappoint:

Watching both the health care and climate/energy debates in Congress, it is hard not to draw the following conclusion: There is only one thing worse than one-party autocracy, and that is one-party democracy, which is what we have in America today.

One-party autocracy certainly has its drawbacks. But when it is led by a reasonably enlightened group of people, as China is today, it can also have great advantages. That one party can just impose the politically difficult but critically important policies needed to move a society forward in the 21st century. It is not an accident that China is committed to overtaking us in electric cars, solar power, energy efficiency, batteries, nuclear power and wind power. China’s leaders understand that in a world of exploding populations and rising emerging-market middle classes, demand for clean power and energy efficiency is going to soar. Beijing wants to make sure that it owns that industry and is ordering the policies to do that, including boosting gasoline prices, from the top down.

It’s true that China’s current leaders aren’t as criminally incompetent as, say, Mao Zedong, but I think “reasonably enlightened” is a stretch. Mostly that’s because I think human rights are important, but it’s also because I’m less optimistic than Friedman about any government’s capacity to “order” technological innovation “from the top down.” Certainly, our government’s own efforts to promote alternative forms of energy don’t inspire confidence.

Anyway, the gratuitous compliment to Chinese autocracy is just a warm-up for the following remarkable analysis of the American political dynamic:

The fact is, on both the energy/climate legislation and health care legislation, only the Democrats are really playing. With a few notable exceptions, the Republican Party is standing, arms folded and saying “no.” Many of them just want President Obama to fail. Such a waste. Mr. Obama is not a socialist; he’s a centrist. But if he’s forced to depend entirely on his own party to pass legislation, he will be whipsawed by its different factions.

Will makes fun of this more effectively than I could:

Only the Democrats are really playing! You might think that would mean they can do whatever they darn well please. But no! The Democrats can’t do anything! Because the other party’s opposition is so effective! So it’s exactly as if there’s just one party: nothing gets done!

My hunch is that the Times’ editors see Friedman aiming the gun at his foot, but watching a man stupid enough to actually pull the trigger is so fun they hate to intervene. That or they’re trying to explode the myth of American meritocracy.

So where were we? Oh, yes: one-party democracy is aggravating because sometimes one party can’t do what it wants because the other party gets in the way. Sooo frustrating!!! Why have democracy at all when all you end up with is a single party stymied by the other one!

One of the reasons I think EFF made a mistake in endorsing new privacy regulations is that I think there’s a huge gap between the sensible-sounding rhetoric of privacy legislation propposals and the details of what’s actually being proposed. The report EFF endorsed offers no fewer than 16 “points” for the design of privacy regulations (although note that “these recommendations are not exhaustive”). Here’s one of them:

“Sensitive information should not be collected or used for behavioral tracking or targeting. Sensitive information should be defined by the FTC and should include data about health, finances, ethnicity, race, sexual orientation, personal relationships and political activity.

I think it’s instructive to try to puzzle out how a regulatory scheme like this would affect sites like Facebook and Twitter. Facebook, by its nature, involves collecting sensitive information about such subjects as when a user logs on, who the user communicates with, what political organizations and causes a user supports, when a user begins or ends a romantic relationship, and so forth. The document advocates that “data collected on users who consent must not be retained beyond a period of three months.” This would apparently require that this kind of information be deleted and re-entered by the user every three months.

The report also demands that “with any change of purpose of the data the individual must be alerted and given an option to refuse collection or use.” Facebook introduces new features that use personal data pretty regularly, so we’re talking about users being repeatedly presented with legalistic descriptions of new features being introduced. The primary result of this is likely to be irritating users, the vast majority of whom will reflexively click the “OK” button without reading the disclosures.

The report would outlaw any “contest that seeks the collection of consumer information in exchange for the chance to win a prize,” apparently even with user consent.

It would also prohibit “behavioral targeting” of anyone under 18. So the Democratic Party, the Catholic Church, and Human Rights Campign, couldn’t buy ads targeted toward 17-year-olds who had identified themselves as respectively, liberal, Catholic, and gay. This seems unduly restrictive.

“Personal data” includes information about “personal relationships,” and the report advocates mandatory advance disclosure of all the ways such data will be used. That would seem to outlaw Twitter’s open social graph APIs, which allows arbitrary third parties to access information about your “personal relationships” without oversight from Twitter.

Maybe I’m misunderstanding some of these proposals; it’s a 13-page document that still manages to be vague about many key details. But I think that’s the point: drawing up a set of rules for the use of personal information by every site on the web today and every website that might be created by someone in the future is a tall order. Websites use personal information in a huge number of different ways; one-size-fits-all rules will inevitably be inadequate. Even in the best case, people crafting a policy like this are going to make some mistakes and create headaches for the world’s web developers. And of course, the sausage-making process on Capitol Hill is not the best case. Maybe the brainiacs at EFF could craft legislation that protects consumers privacy without unduly burdening firms like Facebook. But the people with ultimately authority over the legislation will be members of Congress, most of whom have never used a social networking site in their lives.

The proposal EFF endorsed is based on the “fair information practices” that are beloved by privacy zealots. These rules were designed in the 1970s at a time when the technologies for large-scale data collection was only affordable to relatively large companies and the technology severely limited the types of information that could be collected and the ways it could be used. Given these limitations, it wasn’t crazy to think policymakers could catalog and regulate every conceivable use of personal information.

That top-down world is gone. It has been replaced by a bottom-up world in which the technologies for data collection are cheap and ubiquitous. There’s a lot more data being collected by a lot more organizations in a lot more different ways. It’s not unreasonable to be concerned about the potential for abuses. But it’s delusional to think that we can put the genie back in the bottle. We’re not going to get back to a world where government bureaucrats can prospectively regulate every use of personal data. And legislating as if it’s still the 1970s is a recipe for creating laws that are completely out of touch with the real world.

Over at my previous blogging home, Adam Marcus has a good post enumerating some of the advantages of behavioral advertising:

The problem is that, even if online video services can sell ad time for 3 times as much as broadcasters, because there is almost 15 times as much ad time on broadcast television than online services, the online service will still earn only 1/5 as much revenue as a traditional broadcaster. This is why online video is expected to drive adoption of personalized (or “behaviorally targeted”) advertising: If online video programmers can target advertising to the individual user’s likely interest, rather than to a crude profile of their likely audience, they can generate much higher revenue per ad because advertisers won’t be wasting their ad budgets showing users ads for things they aren’t interested in! The increased revenue for online content providers made possible by targeted advertising is the “mother’s milk” that many websites need to survive.

These are strong points, but I’m not totally sold on this analysis. I think it’s quite possible that as in the cases of the recording and newspaper industries, the future may very well be one in which video content generates less revenue than it used to, and video producers are forced to find creative ways to produce more content on tighter budgets, which isn’t necessarily a bad thing. But Adam’s larger point is surely right: there’s a trade-off here. At least some forms of personalization are likely to increase advertising revenues, and disallowing tracking will mean that content producers will have smaller budgets which will, on the margin, reduce the quality of content. Obviously, some people will prefer that outcome. But others might think the risks of personalization are worth the benefits of more expensive content.

I also think it’s a mistake to focus too much on the financial benefits of personalization. Consumers benefit indirectly from content producers with bigger budgets, but personalization can also offer more direct benefits. If I turn my television to the wrong channel at the wrong time of day, I’m likely to see a lot of ads for incontinence aids, feminine products, children’s toys, and other products for which I currently have no use. If I had a way to signal to the television that I wasn’t interested in these product categories, that would not only benefit the advertisers, but it would benefit me as well, because I’d be more likely to see ads for products I actually care about. Google goes to great lengths to only show search ads it thinks users are likely to click on. This benefits not only Google and its advertisers (who make more money and attract more business, respectively) but it also benefits users themselves, because their time and attention isn’t wasted with totally irrelevant ads as often.

So it’s important to remember that there are trade-offs here. But not only do people disagree on how exactly the balance should be struck, in many cases they even have a clear idea of what the relevant trade-offs will be. There are many possible business models for online video; the best ones probably haven’t been invented yet. If we write privacy regulations now, we not only limit consumers’ choices among existing options, but we may prevent the emergence of new options that could be more congenial to both sides. Not only is it difficult to predict how websites of the future might work, it’s even more difficult to predict how users will feel about services that don’t exist yet. The only way to find out what the options are is to allow experimentation to continue in a decentralized fashion, and to let each user decide for him or herself which options make them the most comfortable.