The confused debate on “big tech” and free speech. By Matthew Yglesias.

The confused debate on “big tech” and free speech. By Matthew Yglesias. 

January 13, 2021. 

Hey folks. I think if I were in a normal columnist job, I’d be expected to do impeachment takes this week where the goal would be not so much to persuade anyone of anything as simply to restate the views the audience already has, in a better-crafted way one hopes, so as to induce them to share the column on Facebook. Part of my goal here at Slow Boring is to not do that, to instead try to focus on content that’s distinctive. So that’s why I’m not doing a piece that’s all “Trump is bad there must be accountability blah blah blah Republicans blah blah our sacred democracy.”

Instead, I want to talk about the less important but more intellectually interesting subthread of how Trump is being treated by big technology companies.

Big tech and anti-anti-Trump politics

When Donald Trump does something outrageous, two things happen.

One is that people outraged by Trump’s conduct try to do something about it. The other is that elite conservatives are split between those who know that the orange man is bad, and those who know that siding with him is the road to success. This breach inside the conservative world is problematic. Usually the divide can be papered-over by having conservatives agree that the countermeasures taken to Trump’s misconduct are themselves bad, and so now we are seeing congressional Republicans line up against impeaching a president who incited a violent attack against Congress with the purpose of overturning the results of the election.

Fortunately for the United States, Trump has been de facto muzzled — overtly by a group of tech companies, but seemingly also by the broadcast media, although in a lower-key way.

Conservatives of all stripes are mad about this specific sanction on Trump.

And more broadly they are alarmed about the following basic reality: at a time when politics has been more polarized along lines of age and educational attainment, big tech companies are primarily staffed by youngish college graduates with left-of-center politics. These companies compete for skilled labor. And one of the fronts of competition in the tech world is trying to create a sense that your particular company is doing big important things and not just trying to make a quick buck. As Steve Jobs famously asked John Scully when trying to recruit him, “Do you want to sell sugar water for the rest of your life or come with me and change the world?”

This is alarming to conservatives and they like to whine about it. Whining is fine, but in keeping with the old joke about a liberal being someone unable to take his own side in an argument, I think some people on both the far-left and center-left are too eager to bend over backward to these whines.

There are some incredibly legitimate public policy questions to ask about large technology companies and their regulation. But progressives ought to demand that conservatives embrace real policy ideas to address legitimate problems and not just do one-off whining. We already know that Facebook executives changed generally applicable algorithm features to improve the performance of conservative websites. And when Twitter considered using their anti-ISIS algorithms to crack down on white nationalists, they backed down after realizing those algorithms would end up flagging some mainstream conservative figures. That’s because, as I wrote yesterday, the line between the mainstream right and the fever swamps is porous.

It’s absolutely the right of Facebook and Twitter executives to facilitate misinformation and racism on their platforms in order to curry favor with Republican Party politicians. But progressives shouldn’t encourage them to do that. And skilled technology workers should ask themselves if they are proud to work at companies that facilitate misinformation and racism on their platforms in order to curry favor with Republican Party politicians, or are they ashamed? I think it should make them ashamed. But they should be proud to have stood up to Trump this January.

Free speech and platform regulation

When you start thinking about a regulatory framework, it becomes important to distinguish between two different things that “free speech” can mean.

There is “free speech” as a dispositional preference for freewheeling discussion verging on cacophony, with not too much attention paid to whether people are offended, or to whether the content of the speech is accurate.

There is “free speech” as a legal concept about private entities’ rights to make their own decisions about what opinions they prefer to express to the public.

When I was in Maine over Christmas, I visited Blue Hill Books as I always do and browsed around before picking up the Historical Atlas of Maine. I took note of what was in their new non-fiction section, and to the extent that the selections touched on politics, it was a decidedly left-of-center set of books — no Liberal Privilege by Donald Trump, Jr. At the same time, I will say that the shop’s current owner seems to stock a more normie liberal set of books than I remember from 15 years ago when there was more radical or leftist stuff.

Either way, I think everyone understands that forcing Blue Hill Books to stock more conservative books would be a huge abrogation of free speech rights. It’s Samantha Haskell’s store and she can do whatever she wants with it. Do I wish she had One Billion Americans on the shelf? I sure do. But freedom matters more.

It’s important to note that this is not just a weak claim that the first amendment doesn’t apply to private actors. It’s true that you don’t have a constitutional right to a book deal, or to have your book stocked in a bookstore, but there’s a strong claim here: bookstore owners have a right to decide which books they want to stock. That is a core first amendment protection. The government can’t make you put Yglesias on the shelf if you don’t want to. It’s your store!

What is the antitrust issue exactly?

Ah, but these companies are monopolies!

This is a claim which seems to appeal not only to conservative opportunists, but to leftists who like to call things monopolies.

But what does Twitter have a monopoly on? Only about 22 percent of Americans use Twitter at all, so the network effects can’t be that strong. Even Facebook, which clearly has stronger network effects, charges an annual price of $0 to users, which suggests that in fact the company feels a decent amount of competitive pressure. A real natural monopoly, like an electrical utility, faces basically no competition, and in theory its owners could soak the customer base to achieve massive excess returns. That’s why a centerpiece of the regulation of electrical utilities is to regulate how much they are allowed to charge.

Now don’t get me wrong, I think antitrust thinkers have raised several good issues about the technology industry:

A fair amount of tech M&A activity — Facebook acquiring Instagram is the poster child for this but Google buying Doubleclick is also a good example — arguably has anticompetitive motives in a way antitrust authorities have not been sufficiently attentive to.

There is a strong argument at the margin that Apple’s App Store policies — in particular disallowing apps that inform you of your ability to sign up and buy things on the open web — are anticompetitive.

There is maybe a case that Google-style conglomeration of web search with all these other functions is bad because it allows Google to favor itself.

I am not particularly fond of the argument that Amazon shouldn’t be allowed to combine retail with Amazon Web Services, or that there’s a problem with Amazon mixing first-party and third-party merchandise in its store, but there are credentialed people who advance these views.

But two critical points about this.

None of these concerns apply to Twitter, which does not dominate any markets or pose any kind of antitrust questions.

None of this is about direct government regulation of content moderation policies (indeed little of it is even relevant).

I think the sensible view about content moderation at Facebook is that Facebook managers need to reconcile competing considerations.

A platform full of trolls and white nationalists is going to be unpleasant and will likely alienate users, but overly aggressive moderation will itself alienate users. To secure talented engineers Facebook needs to persuade potential workers that Facebook’s mission of connecting the world is serving good ends and not just inciting genocide in Myanmar; but to be broadly acceptable to the mass public, Facebook’s conception of what “good ends” are needs to be wider than the personal political views of its staff. Facebook’s executives and shareholders, regardless of their political opinions, are rich people who derive large material benefits from Republicans winning elections and cutting taxes; but they also operate globally, and have to deal with a huge range of different governments whose concerns range from the basically valid (Germany has all kinds of speech restrictions on Nazis) to the not-at-all valid (the Vietnamese government wants Facebook to help crack down on dissidents).

Balancing a complicated set of interrelated tradeoffs is the sort of thing that market competition is good for, so while I think there’s a strong case for more vigorous antitrust measures to ensure that the tech market remains competitive on both the labor and product sides, it doesn’t validate direct regulation of Facebook’s day-to-day operations — and certainly not Twitter!

“Like a utility”

Another way of thinking about this, that I see primarily from people on the left, is that we ought to regulate some subset of technology companies in the spirit of how we regulate electrical companies and landline telephones:

Taking a deep breath here, one genuine oddity to me of the way this conversation has played out is that the United States does not currently regulate Internet Service Providers as utilities. Or rather, a company like Comcast is regulated like a utility in its guise as a cable television provider but not in its guise as an ISP. The same is true for mobile broadband.

That was a big part of the substance of the “net neutrality” controversy that you may recall from a few years ago.

Democrats wanted internet service to be governed by a similar regulatory framework as applies to traditional telephone service. That meant a non-discrimination principle, where ISPs had to treat all bits equally, in the same way that all phone calls go through the copper and switches regardless of what the conversation is about. Democrats argued that the very limited level of competition in the market for stringing fiber optic wires to people’s houses meant this bottom level of the internet stack needed utility-style regulation. Republicans argued that allowing more flexible and creative business models would spur more investment, and that one or two ISPs per neighborhood, plus as many as four mobile providers was plenty of competition — now down to three because Trump allowed T-Mobil and Sprint to merge.

Whatever you make of this argument, I think it’s basically undeniable that wired broadband internet is a lot more “like a utility” than Facebook is. The idea that you would skip over the most basic, least-competitive layer of the stack and apply neutrality principles to social media companies rather than to ISPs seems crazy.

Progressives, to their credit, at least want to apply utility-style regulation to everything. Conservatives seem to me to be engaged in special pleading — they don’t believe middle managers at Comcast and Verizon are liberals so they can do what they want, but they think Silicon Valley workers are too leftwing and need to be regulated.

From either point of view, however, simply saying “like a utility” doesn’t really explain what regulation you want.

Incompatible regulatory aims

As I alluded to above, from my perspective, the core vexing thing about this analogy is that the main goal of utility regulation is to prevent price gouging.

The worry is, that because electricity is extremely valuable, the owner of the electrical grid will be able to charge staggeringly high prices. That will crush household disposable income and also place a massive burden on business in international competition. A rival firm might see the incredible profits to be made in owning an electrical grid and decide to build a competing one. But that costs a ton of money, and the electrical monopolist could always cut prices midway through your buildout and ruin your business to preserve its monopoly.

We have a few different models for avoiding that in the United States, but one of them is the regulated electrical utility and it works like this:

On Tuesday, the Pepco filed a request with the Public Service Commission of the District of Columbia asking permission to boost rates by $66.2 million. “The increase to a typical residential customer’s monthly electric bill would be $7.54,” said Pepco region president Donna Cooper.

“Approval would allow us to recover a portion of the funds that we have invested in the electric distribution system in the District of Columbia.”

Pepco spent almost $90 million last year to improve customer service, as well as system safety and reliability. By the end of this year, the utility expects to spend more than $96 million on things such as new control systems that can isolate problems and restore service, and upgrading or replacing old equipment. Pepco also says that this year it has replaced more than 25 miles of aerial cable and replaced or installed another 11 miles of underground cable.

You have to ask permission to raise prices, with the price increase tied to specific benchmarks about investment in infrastructure and services. The regulators’ mandate is to balance the public’s desire for electricity to be inexpensive with the public’s desire for the electrical grid not to be broken; and also allow the companies to be lobbied both by shareholders who want high profits and workers who want high pay.

Absolutely none of this is applicable to Facebook, so when people say they want to regulate the company “like a utility,” they are really under-specifying what they mean. We regulate lots of industries in different ways, and for different reasons.

And it seems to me there are roughly three different objectives that various critics have in mind for social media companies:

Make them do more to crack down on abuse, misinformation, and extremism.

Make them foreswear censorship (i.e., the opposite of 1) and act as an open forum.

Make them stop relying on algorithmic content promotion and targeted advertising.

The big problem here is you can’t do (1) and (2) simultaneously. It can sometimes seem like there is an alliance of “tech critics” to be found among Democrats and Republicans, but the Ds and the Rs are generally calling for opposite things. There are some people on the far-left who agree with the Rs, but that’s because Ds and Rs have different relationships to their respective ideological fringes. The boundaries between the far-right and the mainstream right are porous, and Republicans generally oppose crackdowns on far-right content. The far-left mostly attacks Democrats, so Democrats have no problem with muffling far-left content; far-left people also know that, so they side with Republicans on this issue — part of the general issue in American life where the far-right, center-right, and far-left all see the center-left as the main enemy.

To me (3) is the most intriguing option, as it would basically kneecap the Facebook and YouTube business models. That seems fine to me. Facebook is pretty bad, and people are happier when they quit. The diversion of ad dollars from supporting journalism to supporting social media enterprises strikes me as bad for society. And it’s not as if all the Facebook engineers would be unemployed if ad revenue flowed back to journalists—their skills are very much in-demand and they’d just go do something else.

Tulsi Gabbard and Paul Gosar introduced a bill to do (3) by stripping companies of Section 230 protections if they use targeted ads or non-chronological algorithms. You’d essentially tell Facebook (and Twitter) that they have to go back to old-fashioned reverse-chronological feeds, where you see everything the people you opt into following posts and nothing else. Maybe you could see what they retweet or like as well, but only in reverse chronological order. And you could also make a common carrier rule (Gabbard and Gosar don’t do this), as we do for the phone company, where nothing gets censored and also nothing gets promoted, which would satisfy (2) as well.

What do conservatives want?

My big hesitation about fully embracing this proposal is I’m genuinely not sure how it would work in detail — like what exactly does the rule say?

But if someone could work it out in a tractable way, I think that might be a reasonable means of achieving conservatives’ stated goal of safeguarding free speech from big corporations. That said, this is not actually what conservatives are calling for; the reason they aren’t is that, despite all the whining about censorship, Facebook’s algorithmic content promotion actually massively favors right-wing figures like Dan Bongino and Ben Shapiro.

When you look at conservative commentary on the Trump freeze-out, it’s remarkably devoid of actual policy ideas.

Vivek Ramaswamy and Jed Rubenfeld in the Wall Street Journal assail big tech, and then conclude simply that “aggrieved plaintiffs should sue these companies now to protect the voice of every American—and our constitutional democracy.” In other words, they’re not calling for any new legislation at all, just for people who are mad to bring lawsuits that they’ll lose.

Similarly, Rachel Bovard in Newsweek makes a persuasive case that the companies who have moved against Parler are being a bit hypocritical in their application of their stated goals and policies. But these are private companies, not courts of law. They’re not actually required to follow their own policies or apply them consistently. For her part, she’s also testified in front of the House to ask that the phrase “otherwise objectionable” be struck from Section 230, thereby making it so tech companies can only remove content that is “obscene, lewd, lascivious, filthy, excessively

violent, [or] harassing” if they want to keep their Section 230 protections. However, I don’t really think this solves the problems, since tech companies can just decide that conservatives are “excessively violent” and we’re back where we started.

My conclusion is that conservatives don’t have a coherent policy agenda here. They know that algorithmic social media is good for the right and they don’t want it to go away. But they also know that most of the workforce at these companies are liberals, so their conscience may compel them to make changes that are less favorable to the right. They don’t want to see that happen. Their solution is to work the refs. It’s like how conservatives will whine endlessly about “the liberal media” but never bother to actually create a serious newsgathering operation.

If conservatives are sincere, they can prove me wrong by trying to work out a viable proposal for something like a common carrier rule that would eliminate both the threat of censorship and also the endless propagation of Bongino content.