Like a lot of other people, I’ve been pondering the question, Why does the Android ecosystem suck? TechCrunch’s MG Siegler has written a couple of posts presenting a conventional view: It’s the carriers’ fault. The openness of Android allows the carriers to load the phones with crapware.

My point is that the same “openness” that Android users are touting as a key selling point of the OS could very well end up being its weak point. If you don’t think Verizon, AT&T, T-Mobile, and Sprint are going to try to commandeer the OS in an attempt to return to their glory days where we were all slaves to their towers, you’re being naive.

Google had other plans that were allegedly foiled by the carriers. They were planning to offer the Nexus One as an unlocked phone for $99.

You see, Google’s grand mobile plan had just one pesky problem: Google doesn’t run their own cellular service. They need the carriers’ support in order to make their phones work. Otherwise they’re just handheld WiFi devices (an idea Google also toyed with).

And Google had another problem. Android was already in full swing at the time, and while the Nexus One was going to be a different kind of phone, they still needed full carrier support for all their other phones. They simply couldn’t afford to piss off the carriers who could effectively destroy the platform they had built. So Google backed down.

…

Yeah. That wasn’t Google’s original plan.

Their big back-up plan to revolutionize the industry was to sell the device online. The carriers went along with that, likely knowing it would flop. After only a few months, the entire thing fell apart.

…

Earlier today, Robert Scoble painted a similar (though much more brief) picture over Twitter of the story I just told. “My conversation last night with a Google VP confirmed that they threw their principles under the bus in order to gain Android market share,” he tweeted this morning. “What did the Google VP say? They learned from Google Nexus One that carriers hold all the cards. They had to play ball with them,” he continued. Bingo.

Does this story make sense? I’m not sure it does. At minimum, it can’t be a complete or primary explanation of the Android problem.

The first major problem with the conventional view is that the wireless industry is actually pretty competitive. The carriers are doing all they can to take business from each other. People look at the 2-year contracts that characterize the industry and assume that it is anticompetitive lock-in. But the existence of long-term contracts is not sufficient to demonstrate lack of competition; competitive markets sometimes feature long-term contracts to penalize inefficient switching, haggling, and opportunism, and of course, the contracts help finance the phone subsidies.

Siegler implicitly concedes that the industry is competitive today when he refers to the carriers’ desire “to return to their glory days where we were all slaves to their towers.” Undoubtedly, the carriers would like to make us all slaves to their towers, but when you have three competitors trying to undercut you, that is a difficult thing to do. Why wouldn’t one carrier simply run TV ads saying, “We install less crapware than the other carriers?”

The second (related) problem with the conventional view is that it is trivially possible to buy (or get for free) contract-free SIM cards from T-Mobile and AT&T. If Google wants to subsidize a $99 unlocked phone, there is nothing stopping them. T-Mobile even charges less for service if you are not on a contract (since it is not subsidizing your phone). I don’t want to automatically dismiss the testimony of Scoble’s Google VP, but this explanation is hard for an economist to swallow.

If the carriers are not to blame for the sad state of the Android ecosystem, then who is? It’s hard to put this delicately, but…I think you people are. Look, in a competitive market, firms basically produce what consumers want. When there are fixed costs (hello phone development), not every niche gets served by the market. Firms sacrifice the inframarginal customer to go after the marginal customer. Who is the inframarginal customer? Nerds. People who love the Internet so much that they are going to buy a data plan no matter what. Who is the marginal customer? The customer that is on the fence between a smartphone and an old-school phone.

I know it is hard to believe, but the evidence leads us inexorably to this conculsion: the marginal customer loves crapware. Crapware is shiny, loud, gaudy. It is the “America’s Next Top Model” of software. People love it. They also love QWERTY keyboards.

The reason for the iPhone’s success is not that Apple is able to strong-arm the carriers. Apple is able to strong-arm us (or rather, you people, the marginal customers). Apple says, “We know that you think you want QWERTY keyboards, the TouchWiz interface, and an antenna that works, but you’re wrong.” And Apple is right. The carriers, on the other hand, give us what we (again, the marginal customers) think we want.

When something sucks, it’s tempting to look for a villain, and often the villain is a shadowy, secretive, and wealthy corporation. People like this story. It’s less sexy to ask if fixed costs plus bad tastes are at play. And often they are. The Android ecosystem sucks because we suck.

Let’s start with the most important question: why did Google decide to start being evil? People seem to actually be asking this childish question. The answer, of course, is that good and evil is not a useful framework for analyzing Google’s actions (though if they open concentration camps I will take this back). Google is motivated by profit. It faces incentives. I outlined Google’s strategy for profit-maximization in A Theory of Google. The basic conclusion of that post is that Google benefits from widespread, cheap, and high-quality access to the Internet.

If that’s true, then why doesn’t Google support net neutrality for wireless providers? <sarcasm>It’s almost as though they haven’t given this any thought.</sarcasm> Except that their chief economist is Hal Varian, who is one of the top scholars of the industrial organization of information-intensive markets and coauthor of one of the seminal books of the field, Information Rules. Varian and his fellow Googlers must have some reason to believe that net neutrality could hinder the development of the wireless Internet (though it appears not all of the rank-and-file are on board).

The first step to understanding the economics of net neutrality is to recognize the large fixed costs that accompany any network industry. The presence of large fixed costs means that the simple price-equals-marginal-cost condition for efficiency no longer applies. If all customers were charged MC, the firm would go out of business. It could not cover its large fixed costs. Even if the costs were sunk, the firm would “go out of business” at the margin, refraining from adding capacity on which it would only lose money. In general, large fixed costs imply that price and/or quality discrimination is a necessary feature of an efficient equilibrium (that is, if consumers do not all have identical demand). Read Michael Levine to see how this is the case even in competitive markets!

Another feature of industries with high fixed costs is that they tend to be monopolized or at least highly concentrated. Economists use the term “natural monopoly” to refer to those cases in which the monopolization is due purely to fixed costs and not to any coercive factor. In fact, traditional ISPs, in addition to being natural monopolies, are also coercively monopolized due to municipal franchises that grant them exclusivity. They therefore do not face even potential entry into their markets. A monopolist ISP might favor its own properties on the web, which is what worries net neutrality advocates. But if the monopolist ISP is free to charge whatever prices it wishes for its service, it can’t gain from pushing its own properties, or at least not at the consumer’s expense. Its incentive is to make the Internet as valuable as possible for its consumers so that it can maximize its profits on its monopoly. Remember the logic of double marginalization. If the municipal franchise results in regulated prices, then the monopolist ISP may have a strong reason to favor its own content. It leverages its monopoly position to reap profits through unregulated content rather than regulated Internet service.

A third factor intrinsic to Internet service is congestion. Transferring data on a network reduces the ability of others to do the same. This is a negative externality that can be remedied through a Pigovian tax, or better yet, through a Coasian solution. After all, property rights are well defined and transactions are already occurring. The externality can be resolved by a change in the terms of the contract.

The challenge, then, if you are socially benevolent, is to find a way (1) to efficiently incentivize investment in Internet service infrastructure, (2) to minimize the ill effects of the tendency of the industry to be monopolized, and (3) to reduce congestion, thereby making existing bandwidth capacity maximally valuable. This is not easy. The efficient solution would be something like the following. Consumers would pay for Internet service in two parts. First, they would pay an access fee, which varies from consumer to consumer in proportion to how much they value the Internet. Second, they would pay for the data they consume on a metered basis, with peak rates being higher than off-peak rates to efficiently allocate traffic. There would be no restrictions on the price or quality of service, though violations of service agreements would be prosecuted as fraud. Because the value of Internet service to consumers vastly exceeds the fixed costs associated with running an ISP, my intuition is that all monopolistic municipal charters should be abrogated and all markets contestable.

If that’s the ideal world, it’s not clear whether net neutrality brings us closer to it or further from it. Because we do not observe the ideal pricing structure, net neutrality regulations hamper firms’ ability to ease congestion by de-prioritizing what they believe is the lower-value traffic (remember, if optimal pricing exists, congestion is self-regulating). On the other hand, because some firms are coercive monopolies and face regulated pricing, net neutrality can improve welfare by taking away an inefficient monopoly rent.

Perhaps the most subtle way that net neutrality could be harmful is by aiding collusion between ISPs. If the firms have a sunk investment in infrastructure, regulations that make it more difficult to recover the value of new investments will discourage entry and expansion. Existing firms can carve up the current market and keep prices artificially high.

Google’s position, that traditional ISPs should be regulated and wireless ones should not, is defensible. Competition is more vigorous in the wireless sector than in the wired, and pricing is less regulated. Furthermore, the wireless industry is further from optimal capacity, so we ought to be sensitive to the incentives to invest.

I don’t mean to endorse Google’s proposal; rather, I wish to suggest that critics take the time to learn something about what it is they are criticizing. It’s dismaying to me that so many non-economists think they understand the effects of net neutrality. Skim some of TechCrunch’s recent posts on the topic: they are, frankly, asinine. It reminds me of a quotation from Murray Rothbard:

It is no crime to be ignorant of economics, which is, after all, a specialized discipline and one that most people consider to be a ‘dismal science.’ But it is totally irresponsible to have a loud and vociferous opinion on economic subjects while remaining in this state of ignorance.

Let me preface my argument by reiterating that there is a lot about Facebook that I don’t like. I hate the “walled garden” approach, and would prefer that decentralized protocols like those used in Google Buzz take off. I’m not a cheerleader for Facebook or its strategy by any means. Nevertheless, I don’t think Internet users have much to fear.

The way to start thinking about network effects is to think about fax machines. A fax machine is absolutely useless if you are the only one who has one. It’s only when other people have fax machines that they become useful. A fall in the price of fax machines has two effects. First, it will induce people to buy more fax machines—this is the ordinary demand effect of moving along the demand curve. Second, because people are buying more fax machines, fax machines become more useful, which increases demand for fax machines—the demand curve shifts out, amplifying the effect of the price drop on quantity. Since the effect is amplified, the true demand curve, the one that takes this into account, is very “flat” or elastic.

Facebook accounts are like fax machines—they are only useful if other people have them. There is the same positive feedback effect of price on quantity. As a thought experiment, imagine that Facebook started charging $10/month for access. My intuition is that many people get more than $10/month of value out of Facebook, and therefore would be willing to pay the fee. However, a lot of marginal users would drop the service. The fact that a lot of users would drop the service would make Facebook less useful to those people who remained; they may no longer get $10/month worth of value out of Facebook if half their friends weren’t on it any more.

I’m not suggesting that Facebook is going to start charging a fee for access. But nevertheless, elastic demand plays a role in how it relates to its users. For instance, one way of cashing in on the service’s popularity would be to plaster it with interstitial ads. This would be kind of like charging a “price” for the service. Why doesn’t Facebook do this? Elastic demand. And elastic demand places limits on the amount of “evil” that Facebook can do, or at least the amount of wealth it can transfer from its users to itself.

Finally, think about network effects and monopoly. Goods and services that exhibit network effects are going to tend toward monopoly (at least if the network effect applies to the good or service directly, and not the protocol, as with fax machines). We should not be surprised that Facebook is becoming huge; if it wasn’t huge, something else would be huge. That’s just how these industries work. But with very elastic demand, having lots of users doesn’t translate into a large monopoly rent. And meanwhile, Google has a strong incentive to ensure that the Facebook tax is not very high.

So my advice is stop worrying and enjoy your consumer surplus. Facebook is not good or evil—it is profit-oriented and faces a serious demand constraint due to the nature of its own product.

Double marginalization is pretty easy to understand. Imagine that people travel by boat down a river from point A to point B. There is nothing useful in between these points; the only point of going down the river is to get from A to B. Between A and B, the river is divided into two segments, each with a separate owner, each of which erects a toll booth. People who are considering going down the river make their decision to do so on the basis of the sum of the tolls. As a result, if the upstream owner raises his toll, it has a negative effect on the downstream owner in the form of decreased traffic and revenue.

If the downstream owner also owns the upstream segment, its profits will be more than the combined profits of the separate segment owners. In addition, under joint ownership, the total amount of tolls will be lower, so consumers of the river will be better off. But here is the key to understanding Google: if the downstream owner can destroy the upstream owner’s toll booth, it has the same effect as if he owned both segments. In either case, there is just one toll booth, and the amount of the toll is set to the profit-maximizing amount.

Think about the Internet as composed of two complementary goods, broadly defined: access and content. Assume, quite reasonably, that Google has some market power in content monetization. Other companies have market power in the access industry. When most people buy a computer, they pay the Apple or Microsoft tax (or both). To get online, you pay your local (usually monopoly) cable company. On your cell phone, you are locked into a two-year contract.

Content monetization is the downstream industry, and access is the upstream industry. Google doesn’t have to take over the upstream industry to increase its profits; it just has to destroy their toll booth. By disrupting complementary industries and making them more competitive, Google is increasing their profits in their downstream industry.

A lot of Google’s projects make sense when viewed from this angle. Google Docs is an attempt to reduce the Microsoft tax. Android is an attempt to reduce the Apple tax. Google’s participation in the spectrum auction a couple years ago (and lobbying of the FCC to require open access) was an attempt to reduce phone carrier market power. Google Voice is the next iteration of that attempt. Ditto for selling phones without contracts. Google Buzz is an attempt to disrupt Facebook’s market power. Google’s recent plans to build a fiber network is an attempt to go after local ISP monopolies.

The point is that Google doesn’t have to dominate any of these industries to be successful, provided that they dominate content monetization. They merely have to make these industries more competitive, lowering the barriers to consuming a lot of content online.

The broader lesson is that monopolies will provide public goods in complementary industries, meaning that they are not as economically harmful as a static analysis would suggest. This is something that policymakers and government agencies should keep in mind as they prosecute firms for antitrust violations.

While I still find the Buzz user interface to be inferior to Facebook’s, it is important to look beneath the UI to see what Buzz is all about: open protocols. You can visit the Buzz API page to see what protocols are in use. The three that I find most interesting are PubSubHubbub (a.k.a. PuSH), Salmon, and WebFinger.

PuSH is about distributing content in real time. Any subscriber or client that is PuSH enabled can subscribe to any publisher that is PuSH enabled and get new content as soon as it is available. I recently outfitted my blog with PuSH, which means that new posts show up immediately in Google Reader and in Buzz.

Salmon is about unifying comment streams. If you spend time on both Google Reader and Google Buzz, you will notice that comments on one service get transferred to the other service. Imagine this happening across the whole web. That is the point of Salmon.

WebFinger is a way to get profile information and other data from an email address. If your email provider supports webfinger (just Gmail so far), someone who knows your email address can access whatever profile information you want to be made public. This means that disparate web services can find each other, so your Twitter account can discover your Flickr account.

Putting these protocols (and others) together, it becomes clear that Buzz is not just another social network. It is a recipe for a decentralized social network. Once the Buzz API is complete, if Yahoo! wanted to, they could launch a social network that would completely integrate with Buzz. So could anyone else. The result will be a less monopolized social networking experience, which will ultimately be a better social networking experience.

Dear Facebook: 1994 called, and they want AOL back.