My Insanely Long Field Guide To Common Carriage, Public Utility, Public Forum — And Why The Differences Matter.

Once upon a time, social conservatives used to be major allies on both limiting media consolidation and on net neutrality. Why? Because they recognized that if you had a handful of corporate gatekeepers controlling access to the marketplace of ideas, they could easily get shut out. Market forces being market forces, companies pressured to censor unpopular or controversial speech and views will do so. Add to that the belief on the part of conservatives that they face ideological bias from the “mainstream media” or “Silicon Valley,” and you had many conservatives back in the day who stood shoulder to shoulder with us back when I was at Media Access Project to oppose Powell’s efforts to relax media ownership rules in 2003 and who opposed Congress’ first attempt to gut net neutrality — the COPE Act — in 2006.


Then came the 2008 election and the Tea Party blowback of 2009-10. Net neutrality became a red team/blue team issue and even social conservatives who had previously supported net neutrality went silent on the issue.


Ironically, now that Republicans dominate all branches of government, conservatives are once again discovering the value of common carriage and government prohibition on any sort of interference with conduits of speech — at least with regard to social media platforms like Facebook, Youtube and Twitter. Why? As conservatives have once again discovered, if companies retain the right to exert editorial control based on content, they will get pressured by the market and government to use that editorial discretion to censor “harmful” speech. That, and the perception that Silicon Valley has a distinct liberal bias, have prompted some in the conservative movement to rediscover the idea that common carrier regulations actually protect and promote free speech and are not a regulation of speech. Because without access to the public square — whether the real life public square or its digital equivalent — your freedom of speech is simply a freedom to whisper to yourself.


I am happy to agree that the time has come to consider whether social media platforms — and other essential elements of communications such as operating systems, DNS registration, or content hosting — should have non-discrimination obligations consistent with our traditional concepts of common carriage. I believe this would also have the salutary effect of protecting companies from liability or social pressure by taking away their discretion. After all, we don’t see anyone demanding that the major mobile providers stop providing cell phones to white supremacists or that broadband providers block subscribers from accessing websites like Daily Stormer. The public accepts that these companies have no choice, because they are common carriers and must serve everyone equally as a matter of law. By contrast, we have seen successful campaigns to pressure DNS registrars to refuse to host the Daily Stormer domain name, Cloudflare, which itself decided to stop servicing Daily Stormer after Daily Stormer claimed that Cloudflare’s decision not to suspend service constituted an endorsement, posted this excellent blog post on why their actions should make people very uncomfortable.


So this should be a great time to reforge the Left/Right alliance on media diversity and government regulation to prevent private censorship, right? I hope so. Unfortunately, this very important conversation keeps getting muddled for two reasons.


1) People keep confusing the concept of “common carriage” with the concept of “public utility.” The differences actually matter a lot, despite 15 years of anti-net neutrality advocates muddling the two.

2) The most active proponents of using government regulation to prevent private censorship on the conservative side are pretty much treating common carrier regulation as a form of revenge porn rather than as a serious public policy debate. “Oh, you don’t want me? You want to break up with me? Well I’ll show you! I’ll make it so you have to carry me!” Indeed, since 2006, when Google (to my considerable annoyance) became the poster child for net neutrality for opponents and a trade press obsessed with treating every policy debate as an industry food fight, the debate about common carrier obligations or non-discrimination obligations or even privacy has always triggered a “but what about edge providers? Waaaaahhhhh!! Regulate them! Regulate them!”


Now I should make it very clear that I can find plenty of progressives who have conceived passionate hatreds for “Silicon Valley” platforms for various reasons, and who also get confused on the concept of “public utility.” Additionally, I can find at least some conservative free market types who understand why we need to regulate things like Internet access differently than hosting services or social media. But it’s conservatives lusting to regulate “Silicon Valley” that have been getting the headlines, and are driving the discussion among Republicans in Congress. Plus I’m getting tired of being asked the same stupid questions by the same folks on Twitter. So I’ll call out the conservatives howling for Silicon Valley blood by name.


Anyway, because whether and how to regulate various parts of the Internet supply chain (or, if you prefer, ecosystem), I will try to explain below why common carriage obligations, such as network neutrality, are different from public utility regulation (even though most utility providers are common carriers), which is different from natural monopoly regulated rate of return/tariffing/price regulation. I will briefly explore some of the arguments in favor of applying some sort of public forum doctrine or common carrier obligation to social media platforms, and — because this invariably comes up in telecom space — why platform or other infrastructure providers are not and should not be covered by Title II or the FCC, even if we agree they should have some sort of public forum or even public utility obligations.


More below . . .

Lets begin with a review of the three concepts that really lie at the heart of this debate as a policy matter: Common Carriage, Public Utility, and Public Forum Doctrine.


Common Carriage.  You can get a more scholarly version of this with plenty of footnotes from this 1994 article by Eli Noam. Contrary to what people who came to this via the net neutrality debate think, common carriage is a purely economic doctrine that goes back about 500 years. It has its roots in the common law around carters (people who took goods from one place to another) and public houses. For trade to work, goods had to flow freely from one place to another. That required carriers of goods and places to stop on the road overnight. Carters and publicans (the people running public houses), recognizing their importance in the trade food chain, began charging people based on the value of their cargo, or accepted payments from one trader to exclude a rival, thus cutting off competition. So Parliament passed laws that certain businesses essential to trade had to treat similarly situated customers the same way. Carters were no longer allowed to pick and chose whose goods they would carry. Inn Keepers would no longer get to charge a different fee for the night depending on the value of a merchant’s goods.


Over time, the doctrine evolved in common law. As described in National Association of Regulatory Utility Commissioners v. FCC, 525 F.2d 630 (1976), the critical case on this in telecom, there became two ways of being treated as a common carrier. First, the law could require you by law to act as a common carrier. Second, you could hold yourself out to the public as serving anyone who shows up, asks for service, and gets basically the same terms as anyone else. There are a bunch of other qualifications and things which I won’t get into, because I’m not here to explain why your average hotel is a common carrier but your average restaurant isn’t. What’s important for purposes of this summary is that common carriage is not particularly about speech or competition. It doesn’t just apply to public utilities. It’s an economic regulation that we apply when policy demands that a service act in a neutral way.


Public Utility. While most common carriers are not public utilities, just about all public utilities are common carriers. A “public utility” is a service so critical that we, as a society, decide that everyone needs to have access to it at a reasonable rate. That fits a very small class of goods and services. Generally, we think of things like electricity, heat in the winter, public transportation (which shades more toward the traditional idea of a “public convenience” — a thing that government should do because it is good for the public, rather than a necessity, but lets not argue that here).


You can see a lot more about my opinions on public utility and why it is critically important in this speech I made to the Personal Democracy Forum back in 2015, and this blog post of mine from 2010 on why we should classify broadband as a Title II service. Again, I’m just trying to summarize things quickly so we can get to the actual discussion.


Critically for our conversation, the government’s obligation under public utility far exceed the general concerns over common carriage. Also, you can have common carriage obligations without being a public utility. For example, cable operators are not generally considered “public utilities.” But states can require cable operators to provide service to the entire franchise area on non-discriminatory terms, a quintessential common carrier obligation. By contrast, the whole point of the Communications Act of 1934 was to recognize that everyone needed access to communications services like the telephone and the telegraph in order to participate in society. This included making sure the rates were “just and reasonable” and that the communications networks met important standards of service and reliability.


It’s certainly true that, for much of the 20th Century, the idea of public utility got bound up with ideas of “natural monopoly.” But public utility does not depend on monopoly or even lack of competition generally. It has to do with the important nature of the service, and the general government responsibility to make sure that everyone has some kind of access. This is why we subsidize power and telephone service to high-cost rural areas, and why we offer subsidies for poor people — something we don’t do for common carriers.


Public Forum. I spent a bunch of time discussing Public Forum doctrine in this post about the Supreme Court’s decision in Packingham v. North CarolinaIndeed, this post includes most of what I wanted to include in Part II.


To refresh people’s memory without them clicking through to read the post. Public forum doctrine holds that certain venues are so closely tied to public expression that members of the public have a right to use the forum to express their views. While the archetypal example is the soap box in the public park, public forum doctrine can apply to private property if it meets the criteria for a public forum. (See Pruneyard Shopping Center v. Robbins.) The Supreme Court has applied similar concepts in electronic media (see, e.g., Red Lion Broadcasting v. FCC), even if declining to embrace the public forum doctrine in toto (see, e.g., Arkansas Education Forum v. Forbes).


In Packingham, the Supreme Court invalidated a North Carolina law which made it a crime for anyone on the state’s convicted sex offender registry to use any social media platform. The majority relied on public forum doctrine, identifying “the vast Democratic forums of the Internet in general, and social media in particular” as subject to public forum doctrine. (The concurrence would simply have found the existing law failed to pass First Amendment scrutiny without reliance on public forum law.)


As one conservative commentor has already noted, it doesn’t take much of a leap from the majority decision in Packingham to find that Facebook, Twitter, and other social media services have the same obligations under public forum doctrine as private streets or shopping centers. This would mean that, even though these are private services, they could not arbitrarily exclude members of the public based on the content of their speech.


So to review what we are talking about here. We have common carriage, which is an economic regulation that requires a provider of services to serve all customers and to treat all classes of similar customers the same. This may range from a general duty to very detailed regulations. But while the conduct regulated may facilitate speech (e.g., carrying mail, offering voice service), the regulation does not have anything to do with speech intrinsically. We have public forum doctrine, which is all about where the public have a First Amendment right to access a particular location — and, after Packingham, broadband and social media platforms. Unlike common carriage, this is all about speech, and is an example where the right of free speech by members of the public actually takes precedence over the private property right. Finally, we have public utility, services considered so essential that the government has an obligation to ensure that all people have reasonable, affordable access.


As a coda, I’ll mention the big favorite in policy  circles right now, antitrust. There is a lot of interest in addressing basic questions of antitrust and how various digital platforms gain and maintain dominance and how this impacts related lines of industry in a very complicated dance. Antitrust regulation can impose various kinds common carrier-like solutions or common carrier-like obligations to stay out of trouble. For example, the requirement to license essential patents under fair, reasonable and non-discriminatory terms (FRAND) looks a heck of a lot like a common carrier-type obligation. The mechanical licensing scheme for music operates under a long-standing consent decree from the Department of Justice against ASCAP and BMI.


So we may have a service or product or behavior where our worry is purely a competition issue that antitrust can deal with, rather than a broader policy issue where we need some form of access regulation or non-discrimination regulation for policy reasons other than (although often in addition to) antitrust. That doesn’t mean antitrust is better or worse than any of the other options. As always in policy, it should be about using the right tool to get the right result. (This goes back to my longstanding theories on how we ought to do policy and stuff, which I won’t bore you with here.)


Why Do We Care? Isn’t This All Just About Net Neutrality Anyway?







It was the great good fortune/bad luck of history that net neutrality was the first taste of the debate we were always gonna have to have in society about how broadband changes everything, just as the debate about regulating railroads was only the first debate in how we needed to adapt as a society to the industrial age. Like the railways, actual broadband access is the necessary component in all these things. As the most important element, the necessary precursor to caring about anything else because if we didn’t have a critical mass of people online services like Amazon and Cloudflare wouldn’t matter, it was natural that we were going to face the essential questions about the appropriate regulatory regime before we really got focused on anything else.


It helped, of course, that we started with the basic transport capacity regulated as a telecom service, so that things like DSL were actually tariffed under existing public utility regulation and the question presented was not “gosh, what’s this broadband thing” but “should cable modem service be classified as a telecommunications service like DSL, or should it be something else — and if so should we reclassify DSL to something else.” That made it a lot easier to start the debate (going all the way back to 1998, when the debate was about “equal access” by resellers to cable modem service and years before Tim Wu would invent the term “network neutrality”) than over services like Google and Facebook (which, of course, did not even exist in 1998).


Likewise, opponents of any sort of common carrier obligation have characterized even the most basic non-discrimination/common carrier obligations as “heavy handed, public utility regulation” that the entire anti-net neutrality movement now believes that common carriage/non-discrimination= public utility regulation. So, annoyingly, this entire extremely important discussion starts totally effed up and confused, with people saying things like “we should regulate Amazon like a public utility” which makes as much sense as “we should regulate Amazon as a bluebird of happiness” or “I want an app to cook my dinner.” WTF does it mean??!! Why don’t you talk about what problem you want to solve, what remedy would work, and then worry about whether you are justifying it under public utility theory or common carriage or antitrust or whether we just need a sector specific law to solve a sector specific problem?


The fact that we have been having the net neutrality debate for so long, and the fact that opponents of net neutrality have been targeting any company that supported net neutrality (particularly Google, even though Google has been a fairly passive player since their disastrous attempt to compromise with Verizon in 2010), has muddied the waters and created this automatic — and very dangerous — link in the minds of people in Policyland between net neutrality and the debate we now need to have about access to certain digital services and platforms. Bluntly, we would need to have this debate about whether and how to regulate these platforms and services now even if the FCC had decided cable modem was a telecom service back in 1998 when this first came up and the issue were long dead and buried. We are having these debates because these services have become critically important to our economy and our lives.


Yes, many of the same basic questions come up in these debates (as I will elaborate below) because they are the questions we should ask in general. In fact, as I noted above, Eli Noam went through a lot of this analysis back in this 1994 paper when we were debating the 1996 Telecom Act and Noam accurately predicted that the emergence of cable broadband services as private carriage threatened to undermine the entire basic principle of common carriage in communications and why that would be a Very Bad Thing. Now we are having a similar problem


OK, So It’s Not Just Net Neutrality. But Why Does it Matter If We Call It Common Carrier Or Public Utility or Sector Specific Regulation Or Antitrust? 


This matters because how you shape and conceptualize something has real world implications when we write it into law. If we justify basic common carriage on Google search, Facebook, whatever as public utility, one of two things ends up happening. Either we end up dragging in a lot more baggage than we need in trying to address the complicated issues around whether and how to regulate these platform services, or we end up gradually redefining the entire idea of public utility as some form of common carriage. Do we really want a world where “public utility” simply means that the provider needs to make it available to anyone with access and enough money, as opposed to the existing government obligation to ensure that everyone has access under reasonable conditions? Alternatively, are you prepared for the government to subsidize server farms for Google and Facebook because they constitute an essential service and therefore the public has an obligation to support the needed additional capacity?


To use a policy analogy, this is like when people want to extend copyright to address problems like revenge porn, because copyright offers such a power set of tools to beat up people accused of infringing that people trying to stop revenge porn say “hey, I want the same power to beat people up! Let’s amend copyright law to give victims a copyright interest because that will blast away all the First Amendment or due process issues!” But doing that would further distort copyright and further perpetuate the problems we already have from extending copyright well beyond its traditional limitations.


Fine. Now That We Have Our Terms Straight, Can You Tell Me How To Regulate Google And Facebook?


Not yet. First we need to ask a bunch of important questions to figure out what is the actual problem we need to solve and how we solve it. Imagine the following conversation.


Patient: It think I have cancer. I’d like you to give me some radiation please.

Doctor: Ummmmm . . . Did you have any particular kind of cancer? It sort of matters.

Patient: Dude! Stop stalling. You are just trying to keep me sick. I want some radiation right now and none of your snooty elitism.

Doctor: Did you want any particular type of radiation treatment or should I just blast you with gamma rays so you can develop super powers?


Sigh. Fine. What Are The Questions We Need To Ask First?


Again, good policy should start with the services themselves, not the players. This also requires not just the “edge” v. “core” debate so popular in net neutrality and in “policy-as-revenge” discussions. Is Cloudflare — which provides hosting services for websites — an “edge” service or a “core” service? What about operating systems? For all the blather about “Google” as an arch-villain coming from the anti-net neutrality crowd, shouldn’t we distinguish between the operating system for mobile (which is pretty much a duopoly with Apple), Search (where they maintain the largest market share), access (where they are intensely tiny, but still Title II providers), and so forth? While “whatever you do to AT&T, do it to Google” may make something of a talking point for people who find actual policy discussions too taxing, it makes a lousy basis for policy if you want to actually write laws and regulations and so forth.


So when we start asking questions like “is it time to regulate edge services?” Or even “while I hate Nazis, it worries me that a handful of companies can just decide to cut a group of people off from participating online,” understand that you are asking the wrong question. I will refer you to works of my old mentor, Professor Robert Seidman, to elaborate more thoroughly. Instead, here are my basic questions to try address these issues. (Don’t worry, I’ll start applying them to specific platforms in a bit.)


Is This More About Speech, More About Economics, Or More About Other Values?


These aren’t mutually exclusive. My unfettered access to Youtube or ability to get found on a search engine clearly implicate speech, but they also have economic consequences. But I can approach speech concerns on platforms like YouTube or Facebook or Twitter, for example, in ways that are different from pure business economics and concerns about competition or “innovation” (which is a fancy way to say competition).


But there are also concerns that don’t neatly fall into the pure business of pure free expression but are vitally important. If a particular segment of society (people of color, rural people) don’t have access, do those people suffer a significant disadvantage. Conversely, does the rest of society as a whole lose something if we don’t provide the service. Is the service necessary for quality of life? For economic opportunity? We provide everyone with free K-12 education because we have made a determination as a society that (a) we are all better off if everyone is educated, (b) those who can’t get access to education are at a severe disadvantage, and (c) we have an idea about ourselves as a society that all Americans should have the right to basic education so they can live full, meaningful lives. We have stated as a society that all Americans have a right to clean, drinkable public water — which is why we invest millions of dollars when publicly shamed into action by the failure to provide clean water. This is different to our concern about either free speech or competition.


Is it about the service or about the size?


People care about being kicked off of Facebook because Facebook is a huge market. Merchants care about access to Amazon or Ebay because that’s where the customers are. But are we worried that exclusion from similar platforms might cause people harm as well? When we regulate taxicabs as common carriers, it is not about competition. We do so because once a customer is inside the cab, they are stuck (this is the idea of “termination monopoly”) and not in a position to negotiate a rate. If you doubt this, I invite you to allow one of the “unlicensed limousine” drivers who you can find by some train stations or airports put your bags in his trunk and then negotiate a rate.


Is there something about the service that makes it behave in an unusual way?


Networks have their own special economics. This makes a cable network or a railroad line very different from a steel mill or a supermarket. Are there particular barriers to entry (for example, the need to acquire a licenses that the government restricts and auctions away)? How easy or hard is it to block competitive entry? What about more complicated economic questions such as two-sided markets, behavioral economic issues, or various forms of network effects? Some or all of these may be present.


How important is the good or service to related markets? 


As noted above, common carrier law started with the recognition that any business that wanted to engage in trade — whether wool merchants, wheat farmers, or weavers — needed reliable transportation from one market to another. Is the good or service something so important to a broader line of commerce that it constitutes an “essential facility?” To whom? What if exclusion simply makes things harder, but not impossible? I can advertise goods even if I am excluded from Google or Youtube, but it would clearly put me at a competitive disadvantage no mater what business I’m in.


How easy or hard is it to translate the answer into actual policy? Does the proposed solution actually address the problem?


When someone says “let’s regulate search like a public utility,” I have absolutely no clue what that means. Does it mean that we are going to spend millions of dollars to make sure that all Americans have it? Even if we say “let’s have search neutrality,” what does that mean? If it means “don’t favor your own stuff over non-affiliated products,” that’s fairly straightforward — even if we quibble around various edge cases. Folks like Tim Wu have done tests to determine if search engines such as Google favor their own products, demonstrating we can have rules and test for violations of broad principles. But should “search neutrality” mean “search engines can’t learn based on an individual’s behavior?” That would undermine the entire point of a search engine — which is to allow people to find results they consider relevant and useful. How do we balance between the interests of a user, who wants the search to learn their specific preferences v. competitors and speakers who want an equal opportunity to be found?


Do we need any sort of action at all? And at what price? Sometimes, the cure really is worse than the disease, or the disease cures itself over time. I have often accused opponents of worshiping the Gods of the Marketplace and of elevating their criticisms regulation to the level of religious dogma. But it would be equally foolish and dogmatic to ignore legitimate concerns about regulatory capture, public choice theory, the ability of incumbents to manipulate rules to create barriers to entry, and other potential costs of any particular regulatory action or non-action. It is unfortunate that the residents of Econ Cloud Cuckoo Land have given cost/benefit analysis a bad name by using it as an ideological tool and devotional exercise to the Gods of the Marketplace. But even when it is clear that regulatory intervention is necessary, we ought to consider how to achieve the desired results most efficiently, and how to minimize the likelihood for mischief. As I like to say, public policy is about maximizing the likelihood for good outcomes while minimizing the likelihood of bad outcomes.


Are We Finally Ready To Talk About Regulating Facebook Or Google Yet? 


I know, everyone wants to jump to the juicy stuff — especially the people who think this is about revenge for net neutrality bwahahahahaha. But yes, now that we have some idea of what we’re talking about, and the right questions to ask, we can start looking at what in the Internet ecosystem needs what kind of regulation.


So Do You Support Regulating Google Or Facebook Or What?


I think it’s time to have the conversation. I can’t pretend I know enough yet to pull a full blown regulatory regime out of my ass. But I think at this point it’s reasonable to start breaking down the various services that we consider essential for a variety of reasons and asking the questions I’ve outlined above. I also think that once we run through the analysis, it becomes pretty clear why broadband access stands out from just about every other piece in the equation. However central any individual piece in the ecosystem may be, the broadband access provider is critical to providing that service, as well as every other. Amazon may be essential as a shopping hub, but I can’t get there without broadband access. I can’t reach Facebook to conduct my speech, or Google to conduct my search. My ISP can block my access to any website or service as easily as denial of hosting by companies like Cloudflare. On top of that, the access to pure transport of information is critical to any other service from accessing my smart home to playing with my connected sex toys.


But lets start running through the analysis.


How Do We “Break Up The Ecosystem?”


Lets start by recognizing that this is “activities based” rather than “status based,” to borrow language from the AT&T Mobility v. FTC case. Contrary to the “regulation as revenge” school, which of necessity views this as “punishing Silicon Valley” for its supposedly liberal bias. But just as network neutrality doesn’t apply to Comcast’s broadcast or movie businesses, any regulation of search or social media platforms wouldn’t apply to other businesses offered by the companies. Similarly, just as Google Fiber is a Title II telecom service, and regulation of hosting services, for example, would apply equally to AT&T’s hosting services.


It’s important to understand this not simply to get the policy right, but to get rid of the rather pervasive notion among those who see this as part of the “net neutrality” fight that this is all collectively “the Internet” and we can just impose the same regulation on everyone because “Internet.” (No, I’m not making this up, the idea that all these services are the same because “Internet” is actually a selling point.) No one outside the Washington DC Beltway (with the exception of some of the economists in Econ Cloud Cuckoo Land) has any trouble distinguishing Amazon from Verizon, or even distinguishing Verizon’s Yahoo! affiliate from its mobile services. Likewise, no one would imagine that we could impose the same regulations on airplanes and UPS because “meat space.” Yes, there is overlap between airlines that carry passengers and companies that deliver packages (both use airplanes, both must abide by basic principles of common carriage), but no one tries to book themselves from DC to California as a UPS package (although at this point that might be safer and more comfortable than United).


Fortunately, the recent incident with the Daily Stormer and the treatment of neo-Nazis and other white supremacist groups as a reaction to the events in Charlottesville provides a useful way to break out the Internet ecosystem and how it works.


The Internet Ecosystem Post Charlotte System.


Following the “Unite The Right” Rally in Charlottesville, various Internet companies have taken action to terminate business with the most blatant white supremacist groups. As this piece in Slate observed, “What’s new about that latest group of bans is that, rather than Facebook, OkCupid, or Airbnb revoking individual and group accounts, the internet’s gatekeepers are now kicking out whole organizations.” For our purposes here, lets break it down (and note what is not included as well).


Facebook, Redditt and other social media cites have been terminating groups, fora and communities. (Source: Slate piece above and here.)

Paypal announced it won’t service hate groups. (Id. and here.)

Domain name hosting companies such as Go Daddy and Google have refused to host the domain name for The Daily Stormer, effectively kicking it off the Internet during the periods it is without hosting services. (Sources: here and here.)

Cloudflare, an infrastructure company that handles hosting and cyberdefense, terminated its agreement with Daily Stormer after The Daily Stormer claimed that Cloudflare’s continuing to provide service constituted an endorsement. (Here and here.)

Google removed the ap, a social networking ap popular for hate groups worried about being kicked off more mainstream social media. (Here.) (Apple never permitted in its ap store prior to CVille. so both major ap stores now ban

Interestingly, there is no indication that anyone is calling for cell phone companies to cancel the service of identified white supremacists or white supremacist organizations — or block their last mile Internet access.


Extrapolating out, we can loosely classify the relevant ecosystem into the following categories. 1) Places where people get together to discuss things and possibly also carry on commerce (Reddit, Facebook); 2) services that make it possible to speak directly to the public on the Internet (name hosting, cyberdefense); and 3) services that enable speech but in ways that are not unique to the Internet (electronic funds transfer, ap store). We also have a class of services that are accepted as too important to cut off or embedded so deeply in the infrastructure that people don’t think about them: last mile connectivity and search.


All this has certainly had an impact on the ability of the “alt-right” to speak — either to each other or to the public — and tho carry on commerce. It is also of note that a relatively small number of companies, working privately, are able to achieve a level of censorship and eliminate businesses from the stream of commerce in a fairly thorough fashion. It is potentially possible for various of these websites and speakers and aps and businesses to find alternatives. But doing so is costly and inefficient. For better or worse, The Daily Stormer became a lot harder to access, it became a lot harder for those behind it to make money from it, and supporters of its viewpoint found it a lot harder to openly discuss it and organize effectively. Whether we like the outcome in this case or not, it provides a stark illustration of why we would want to have rules in place to prevent these companies from having such outsized impact on speech or the economy.


But it also illustrates why trying to impose the exact same rules at every layer would not make any sense, and would not serve our purposes. Lets take it a category at a time. We can see that when it comes to the places where speech takes place, such as Facebook and Reddit, that this start to look a lot like “public forum.” We care about the ability of speakers to speak and organize freely, even when their views are unpopular. We recognize the need to police conduct so that one set of speakers cannot drive out another via harassment or punish those actually planning criminal activity. We have plenty of analogs in meatspace to the anti-harassment policies used by social media platforms. But we get understandably nervous when behavior considered legal and acceptable becomes punishable online without any sort of process.


This looks a lot like traditional “public forum” analysis. First, we don’t care about the size of the speaker. Reddit has just about no market power in the social media market in the classic antitrust sense. But eliminating speakers and communities on Reddit can have as crippling an impact on a community as elimination from Facebook. What matters is that these are services where people go to exchange news and views and speak to each other and the public at large. The value of the service, and our concern over censorship, is inextricably bound up with our concern over speech. That suggests that this is a question of public forum doctrine.


By contrast, services like web hosting and ap stores may enable speech, but they generally do a lot more. They are critical to commerce as well as speech. We don’t tend to think of them as necessary for speech itself, but for the ability to speak freely without reliance on an available platform such as Facebook or Reddit. And again, the question of size seems less relevant to the nature of the service provided. To the extent that regulation is needed here, we would be looking more at common carrier/non-discrimination type regulation. Our concern is not simply that particular viewpoints will be excised, but that — in the words of Matthew Prince, Cloudflare’s CEO: “Literally, I woke up in a bad mood and decided someone shouldn’t be allowed on the Internet. No one should have that power.”


And, as Prince points out in both that article and this thoughtful blog post, regulations such as common carriage protect companies as well as customers and commerce generally. As I have observed on many occasions, when you have editorial discretion people will expect you to use it. Prince notes that it will now be harder for Cloudflare to resist requests from governments to refuse to service other “bad” websites or services or to shield them from government action. After all, are Nazi’s worse than people accused of sex-trafficking? Or terrorism? Or of “incitement” to various crimes? The potential to abuse the infrastructure is as powerful for government actors as it is for private actors. Who decides which speakers or businesses are worthy? Obligations which impose due process considerations and make it harder for both private actors and government actors to discriminate both protect speech and make commerce — including those businesses on which speech depends — more reliable and dependable.


Sector Specific Regulation, Antitrust and Public Utility


You will notice that for each sector — direct speech such as social media, support for speech and commerce generally such as webhosting, and direct commercial activity that is only incidentally related to speech such as electronic payment processing or access to app stores — we have similar questions but very different answers. Solutions that make sense in the context of direct speech, such as public forum obligations, don’t make sense in the context of web hosting or app stores. The broader economic concerns around self-dealing and profit that motivate antitrust may apply to any of these businesses, but our concerns are not limited to market power. If we believe that we social media have become too important to public debate to allow the provider to arbitrarily kick people off or favor one sort of speech over another, then we care about smaller social media platforms such as Reddit, LinkedIn and Twitter as well as caring about Facebook.


So where does that leave public utility? Again, I stress that “public utility” is a much broader and more powerful concept than simply “common carriage.” Are any of these services so critical to our daily lives that there is  risk we might die without access, or that lack of access effectively cuts us off from fully participating in modern society? I would argue “no.” And that is why we are much more comfortable with cutting neo-Nazis off from Facebook than we are with cutting them off from all phone service and broadband service.


Cut someone off from broadband access and you cut them off from all the services listed above as essential — and many more relating to common human relationships like staying in touch with family or doing your job. It’s like cutting neo-Nazi’s off from electricity so they can’t spread their evil poison. It’s incredibly effective precisely because it is so difficult to participate in society without access to electricity, let alone communicate with others or run a business. That’s why we treat electricity as a “public utility.” That’s why we make it so hard for the power company to cut off your service when you don’t pay your bill on time. That’s why we as a country are willing to spend so much money collectively to make sure that everyone who wants it can access electricity. Yes, the Amish can live without it — but the whole point for the Amish is that they do not want to participate in modern society.


Broadband, as I wrote back in 2010, is part of our critical infrastructure as a nation. We as a nation are looking at spending billions of dollars to make it available to everyone, because it is so critical to living productively in the 21st Century. However important Facebook or Cloudflare may be in the broader world, it doesn’t rise to the same level as basic access to broadband service. And that’s why we care about broadband not simply as a common carrier, but also as a public utility.


What About Google Search?


I didn’t discuss Google Search because there doesn’t seem to be any effort to push Google (or Bing, or any other search engine) to degrade or remove white supremacist results. but Google Search is no different from any of the other services, in that the same questions apply. It is pretty clear that search is critical to using the Internet effectively, and businesses live and die by their search results. It is also clear that, while vitally important to speech, search engines are not speech per se.


It is hard, however, to know precisely what people mean when they say they want “search neutrality,” or other ways of framing the question that make it clear this is about punishing Google for its perceived sins rather than actually giving a crap about policy. Some applications of the term are straightforward — don’t favor your own products over those of others. Don’t take money to subtly alter search results and make them look “organic.” As I noted above, experts can test for these things and we could impose rules on Google (if we are only concerned about antitrust) or all search engines (if we think the sector is important), that would reflect these fairly straightforward principles.


What gets harder is that we actually want search engines to learn our individual preferences. Odds are good that when I type in “Guardians of the Galaxy” to a search engine, I’m looking for reviews, criticism or parodies of the movies. My son is more likely to be looking for material relating to games and the comic books. We want the search engine to learn our individual preferences over time, so it can serve us better. That’s what makes a good search engine. But doing so has costs to merchants and others who are trying to optimize their products to attract our attention. Whose preferences should the law favor, and who decides? Do we ignore the customer preferences out of concern for the macroeconomic impacts and the fear that Google or other search engine designers can discriminate in the bad way when we use a “consumer welfare” standard? Or do we maximize the ability of the provider to custom tailor the experience? If we have a different answer for Google than for other search engines, then we are probably talking about antitrust. if we think this ought to apply to all search engines, then we are looking at something sector specific, but probably not common carrier as the concept of common carrier doesn’t make much sense here in the traditional sense.


In any case, the idea that we should port regulations suited to search to other “edge providers” or broadband access providers because “Internet” is a good way to make really bad policy.




So here we are, 7000+ words later, and I haven’t directly answered whether I agree with all these conservatives about regulating “edge providers” or “Silicon Valley” or not. Hopefully, however, I’ve explained how we ought to talk about it intelligently. I can’t say I’m 100% certain about what sort of regulations ought to go where. I do know that we ought to be looking at this through the lens of public policy, not the lens of revenge or even simply the lens of antitrust. Finding that broadband access itself is a public utility was comparatively easy. It displaces a public utility (voice service), and provides the necessary infrastructure on which all these other services and all this speech depends. If that does not qualify broadband access as a public utility, then I don’t know what could. But is Facebook or Google on the same level? I confess I’m skeptical that, for all their obvious importance, they rise to the level of “public utility.”


Questions about the best way to protect speech and whether services are so central to our economy that they require sector specific regulation, or so central to our lives that we consider them public utilities, go back centuries. We should consider these questions with the seriousness they deserve, rather than try to shoehorn them into categories that don’t fit. We don’t need to find that Facebook or Google are “public utilities” to impose necessary sector specific regulation on social media, or search, or web hosting. We should debate the need for regulation and what regulations are appropriate on the merits, on a sector specific basis, because good public policy demands it.


Stay tuned . . . .

One Comment

  1. Dammit, Harold, here you are giving thoughtful, nuanced views, when everyone just wants “answers” 🙂

    Thanks for the nice piece.

    How would some of this change when viewed with a lens outside the US? Where market power, affordability, etc. are different? When the government itself is a major provider of commerce and ICT?

    Also, a lot of people believe “terms of service” is what allows folks to be kicked off platforms/ICT – have people ever dug enough into that? We have lots of horror stories of Paypal cutting off people’s accounts when *someone* thinks they are wrong/bad/evil and then uses the fig leaf of terms of service to kick them off.

Comments are closed