All Articles

How to Regulate Big Tech (and Why it’s Failed So Far)

Reading Time: 13 minutes

Table of Contents

Today’s big tech companies have too much power — too much power over our economy, our society, and our democracy. They’ve bulldozed competition, used our private information for profit, and tilted the playing field against everyone else. And in the process, they have hurt small businesses and stifled innovation.

Elizabeth Warren, “Here’s how we can break up Big Tech

I agree with Warren. So do most people – a 2019 survey indicated that two-thirds of Americans (on both sides of the aisle) think we should break up big tech companies. Google, Facebook, Amazon, and Apple’s monopolistic abuses of their market power are clear, whether it’s through failing to shepherd their harvest of our personal data, squeezing out competitors through predatory pricing, or in the literal words of Biden this week, “killing people” by allowing the spread of misinformation. They’ve been doing this for a long time and everyone knows it.

So why hasn’t anything changed?

In fact, every time regulators act, the companies’ stock prices go up (Facebook example, Google example, Amazon example). How can we be failing this badly?

Many just chalk it up to the greed and incompetence of our politicians. After watching their performance in Zuckerberg’s 2018 senate hearing (below), it’s easy to see why. But even in the hands of more digitally savvy, regulation-forward politicians (like AOC, who put Zuckerberg in the smoker), regulation has stagnated.

“Senator, we sell ads.”

I believe it is because most regulation relies on the concept of a monopoly, and our current understanding of monopolies is irrelevant in the digital age. Any solution rooted in that understanding will fail.

Instead, I’m going to lay out my digital-first perspective on how we should think about regulating big tech, with a focus on a new concept: digital public goods.

Why Big Tech isn’t Big Steel

monopoly: the exclusive possession or control of the supply of or trade in a commodity or service.

Oxford Dictionary

Typical monopolies are characterized by a single supplier. Andrew Carnegie’s U.S. Steel famously controlled steel supply. Here in B.C, ICBC controls the supply of car insurance. These are problematic because when there is just one supplier, there are no market forces holding them in check. If they raise the price, provide terrible customer service, or squeeze their suppliers, there’s not much anyone can do, since there are no alternatives. Hence, dumpster fire.

However, monopolies have been largely under control in North America for about 100 years. Thanks to the 1890 Sherman Antitrust Act, there is a clear path to breaking up big, non-digital monopolies. This act is still referred to in today’s tech lawsuits, but with much less success. Is that any surprise? The world of business has changed radically since 1990, let alone 1890.

The reason it doesn’t work is tech companies don’t have typical monopoly power.

For one, their markets are near-impossible to define. Does Facebook have a monopoly over social media? No, in fact, it’s slipping behind upstarts like TikTok. Online advertising? Nope, under 20%, and that would be even lower if you take the overall advertising market.

What about Amazon? They have a 37 percent market share of eCommerce, which is substantial. But their biggest competitors are Walmart, Target, and other large physical retailers. If you include those stores’ in-person presence, Amazon has just 5% market share.

In the past, regulators could address this by identifying the physical region that a company has a monopoly over. But on the Internet, regional borders are irrelevant. All Internet-based companies compete with every other Internet-based company in their industry around the world. You can’t draw a border on a domain name.

The challenges go beyond defining the market, as well. They also have no supply control. Google, for example, has a 93% market share of online searches. That’s certainly a monopoly, right? Not by the book, because searchers can choose to go wherever they want.

Unlike U.S Steel, which was literally the only supplier of steel, Google competes with dozens of other search engines. Some competitors, like Bing and DuckDuckGo, are full-featured and well-funded. Google has no way of stopping you from using them. It can’t be a monopoly if there are plenty of alternatives that users are free to use.

So how did big tech companies end up so powerful and profitable? The answer is they aren’t monopolies; they’re aggregators.

Their power is explained by Aggregation Theory, coined by the writer Ben Thompson, to whose blog this post owes a great debt.

Aggregators: The Tech Non-opolies

Here’s Aggregation Theory in a nutshell: in pre-Internet times, distribution was expensive and controlled through intermediaries. This refers to the distribution of anything: steel (warehousing, shipping, etc), information (newspaper printing, hosting a cable TV network, etc), learning (opening classrooms, printing textbooks, etc), reaching a friend (phonebooks, landlines).

The Internet reaches everywhere, anywhere, nearly for free. This drastically reduces the marginal costs of distribution and shifts every business’ addressable market from “local/regional” to “the entire world”.

Now, instead of a teacher only being able to reach their classroom, they can reach anyone in the world. Instead of just being able to chat with people in your postal code, you can chat with anyone in the world, and it’s no more expensive if they’re in Taiwan or two blocks away. Instead of only being able to do business with your local steel supplier, you can source your steel from anywhere in the world.

This created a massive shift in the value chain. “Integrated” companies that succeeded by owning both product and distribution suffered. Think: newspapers, travel agents, cab companies.

The Internet also created a new challenge for consumers: when I have access to all the options in the world, how do I find the option that’s right for me?

Enter, aggregators: companies that use algorithms to round up all the options in the world and find the best one for you.

  • Facebook aggregates and serves you the best content.
  • Amazon aggregates and serves you the best products.
  • Google aggregates and serves you the best answers to your questions.
  • Airbnb aggregates accommodation.
  • Uber aggregates transportation.

The list goes on. Understanding this shift in market dynamics is crucial to understanding how to regulate them.

Credit: Stratechery

Why don’t we just break them up? (Or tax them more? Or give them more rules?)

Typically, people suggest we do one of the 3 things above, but they are all destined to fail as they are rooting in old ways of thinking about monopolies. In fact, they’d actually hurt the Internet in the long term.

Break Them Up

Aggregators win by providing users with the best experience through algorithmically serving the highest-rated, most relevant options. This has incredible lock-in: once you’ve felt the magic of having your own personalized and pre-vetted slice of the world at your fingertips (at a fraction of the cost), you’re not going to go back. Think of Uber versus cab-hailing, or Google versus looking through an encyclopedia/phone book. This makes users of these products extremely loyal.

So these companies don’t control supply, they control demand; through the quality of their user interface, selection of content, lower prices, or smarter results.

This combination of high loyalty and reliance on comprehensive selection makes aggregator markets ‘winner-takes-all’ – there’s a reason no users are clamouring for another Facebook, another Google, Airbnb, et cetera. At best, these markets can be oligopolies with a few massive players, but never truly competitive markets. And that’s why breaking them up will fail.

If we split up Facebook and Instagram, for example, people will end up just going to one or the other – or a new entrant like TikTok will fill the void, and it will command just as much market power as Facebook does now. Systemically, nothing will change. Unwinding acquisitions (such as FB buying IG) also sends the wrong signal to startups – acquisitions by big companies are important to the startup ecosystem.

Tax/Fine Them

Once aggregators reach a certain size, they reach ‘escape velocity’ through their network effects. This entrenches them and their war chest. Take Facebook, for example. The more people are on Facebook (or Whatsapp, or Instagram), the more useful it is for everyone on it. So the site gets better for its users as it grows, increasing loyalty and growth. As more users stay longer, they see more ads. And since those ads have virtually no marginal cost and a worldwide addressable market, the business ends up spitting out insane amounts of cash, and you end up with one of the richest companies in human history.

It’s hard to comprehend how much cash they have. At the end of 2020, Facebook had 62 billion dollars in cash, which on a GDP basis, is enough to buy the entire country of Croatia and have billions left over. Google has 135 billion.

This allows them to maintain their market position through sheer force of cash. As we know, they are able to buy their competitors outright, or effectively buy markets with cash. See: FB paying creators $1 billion to use their products.

More importantly, their cash piles mean that even historically large taxes or fines are just rounding errors on their income statement. They’ve done the math and know it’s better to get sued than to do anything to hurt their position as the aggregator. Instead, the taxes/fines would just discourage new entrants that would have to play by the same rules without the war chest.

Change the Rules

Rules-based regulations that force big tech companies to play nice often come at a greater hidden cost. I’ve seen this play out most directly through the ongoing war for Internet privacy standards:

  • The EU’s GDPR regulations forced companies to ask for user’s consent before tracking them. But it has turned the Internet into a hellscape of pop-up requests for users and made running compliant analytics totally inaccessible to small businesses.
  • Regulators are asking that companies scale back their user-level interest targeting. But Google’s proposed solution, the FLoC, is creates as many privacy issues as it solves.
  • Apple has bragged about how its iOS14 iPhone privacy policy updates protects their users. But they’ve hurt the ability of tens of thousands of small businesses to advertise profitably. And it also forced app advertisers to shift their budgets to Apple’s search ads, making it an expansion of monopoly power, not a regression.
    • They’re also tracking all the same creepy stuff as before, they’re just keeping it within Apple, not sharing it. Don’t believe me? Go to Settings>Privacy>Location Services>System Services>Significant Locations. Then click the locations. They’ve got a record of everywhere you’ve ever been, and if you’ve got an Apple Watch, what your heart rate was when you were there.
Privacy: When everything you do is tracked, but only Apple gets to access it.

Although efforts to create new privacy standards should continue, a ‘rule-making’ approach to regulation will hurt new entrants and consumers more than the aggregators themselves. Aggregators will simply introduce new, arcane methods (like FLoC, cookie management, iOS14) to achieve their goals while following those regulations, and those methods will entrench their market position.

When it comes to privacy, this trade-off between decreased market competitiveness and increased rules may be worth it. But more generally, rules are not a solution for aggregator market power.

How to Regulate Big Tech: Digital Public Goods

Instead of applying the old, monopoly-based methods above, I propose we focus on defining a new framework for regulation: digital public goods.

Public goods are resources that affect the general wellbeing of the population, so need to be managed equitably, with government involvement. In Canada, drinking water, healthcare, clean air, and law enforcement are all public goods.

I propose we choose 3 digital public goods to regulate: cloud computing, user-level behavioural data, and search interfaces. Each of these is necessary for digital wellbeing and requires its own type of regulation.

Cloud Computing

Cloud computing power is the quiet cash cow behind many big tech companies.

What is it? In short, if you want to create an app (say, a to-do list), that app’s code has to exist in ‘the cloud’, so other people are able to access it. But there need to be literal computers that host the cloud for your app to live on.

These computers live on massive, dystopian server farms across the world. These farms are almost 90% owned by the big 3: Amazon AWS, Microsoft Azure, Google Cloud.

It’s highly profitable. AWS recorded a $13.5 billion profit last year.

Cloud computing is a foundational part of the piping of the modern Internet. Companies (and users) don’t really have the option to opt out of it. Imagine if Canada’s clean water was in the hands of just a few private companies, and they were racking up billions in profit?

Not only are these companies profiting off of something everyone needs, they’re also using it to fund market dominance for the rest of their businesses. Amazon can afford to compete on ever-lower prices, not by “being customer-obsessed” like they say, but by using their cloud computing profits to cover their losses from selling under-priced goods.

The solution to this is my most radical one: nationalize cloud computing. Canada and the U.S should forcibly acquire the big 3’s server farms and continue to serve their customer bases for a fraction of the cost.

AWS et al. can continue to offer more premium services on top of cloud computing, such as machine learning and database management, but the raw horsepower of computers needs to become a public good.

User-Level Behavioural Data

Every day, I scroll through Instagram, and it learns a little more about me. What I like, what I click, what I buy, what I ignore. It then uses this data to serve me targeted ads. As everyone knows, this is a little creepy, and often a gross breach of trust by the aggregators.

Most solutions so far have focused on how much tracking we can turn ‘off’. But it ignores a few key realities:

  • Advertising powers a lot of the world. It pays for our TV, sports, blogs, news, affordable direct-to-consumer brands, and more. Tracking makes advertising better, allowing for more of all those good things. Turning tracking off will hurt all the industries that rely on advertising.
  • As discussed above, turning ‘off’ tracking risks just signalling to the aggregators to find another way to track the same things.
  • The aggregators have a massive head start in ad technology. If we reduce tracking moving forward, their existing algorithms and history of data will make it impossible for new ad platforms (or social networks) to compete.
    • I’m seeing this play out in real-time. My agency usually just advertises on Google and Facebook, but we’re currently testing ad campaigns on Pinterest and TikTok. So far, not great – with new privacy rules, it is much harder to define the audience we need and justify our spend to clients.
    • The 2nd-order effect of this is a less fun Internet – if new social platforms can’t run profitable ads, they simply won’t exist, and we’ll be stuck with Facebook/Instagram’s doomscroll forever.

So simply reducing what we track will hurt more than help.

Additionally, great digital marketers know that any individual type of data isn’t that useful – the most effective advertising harnesses big data for ‘lookalikes’: giving ad platforms ‘seed’ lists, such as your customer or website visitors, and simply telling the algorithm “find more people like this.” We don’t really need all that personal data, we just need a nice big dataset to allow the algorithms to do their thing.

Instead, I propose that individual sites can track user consumption data for advertising, but they can’t host it themselves – all user-level behavioural data must be passed directly and exclusively to an open, government-regulated advertising API.

There would be one master database that would house all your information: what IG posts you reacted to, what Spotify playlists you have on repeat, what TikTokers you follow, what YouTube videos you watch, what you purchased on Amazon. All that data is already out there. By putting it in the hands of a government group, we would create transparency and standardization about what data is ethical to capture and what isn’t, taking that decision out of the hands of the private companies.

Unlike most proposed solutions, it would also make digital advertising better, not worse. All digital ad platforms would have to feed their data into it, but they would also be able to use all that data for lookalike targeting. Instead of algorithms finding “similar audiences” based only on what those people did on a single platform – say, Facebook – they would be able to run correlations based on all types of different factors to figure out the most likely people to buy. The result would be algorithms that are better than ever at targeting customers, with less access to personal data than ever.

It would also force social networks to compete on the merit of their actual network instead of relying on their advantage in data. My agency could create a general ideal lookalike audience, then run ads to target it on TikTik, Pinterest, Snapchat, or wherever else we think our audience is. We wouldn’t be locked into advertising on Facebook just because we need Facebook’s data.

Search Interfaces

When you Google something, the result is called the SERP, or Search Engine Results Page. This is the single greatest advertising placement in history.

An example of a SERP. Note that my screen is entirely ads.

Every minute, millions of people start their shopping journey on Google. Millions more search for new apps to download on Apple and Android. This completely changed advertising.

In the old world, you had to find your audience and hang around long enough in their head until they were ready to buy. In the new world, you can show up right at the perfect moment when they’re ready to buy. It’s incredible.

If anyone’s wondering: yes, it still works. I can confirm this as the owner of an agency that spends hundreds of thousands of dollars there every month. It works because people trust search engines’ algorithms to recommend them the right options.

Unfortunately, search engines haven’t always earned that trust. They’ve used their aggregator position to bend the results to their favour. Sometimes this is subtle, sometimes not so much:

These companies argue that their search results are their own purview – grocery stores, for example, have long guided their customers towards their own private-label products and no one ever said boo. But these companies are aggregators, not grocery stores. They have a unique, dominant relationship with their consumers, so their search results need to be considered a digital public good.

As you can tell from the coverage above, I’m not the first person to propose it, but I will say it loudly: we should regulate how aggregators show search results.

One of the reasons SERPs are an effective place for regulation is that unlike most of the Internet, they don’t escape definition. They are pages that show results based on a query.

Although defining the exact regulation is beyond the scope of this post, I would focus the definitions on 4 key points:

  • Reduction of opportunity for “manual” (done by human) adjustments to results
  • Limits on ad units relative to organic (non-ad) units
  • Limits on ‘self-serving’ algorithmic weights, such as ones that would favour first-party results
  • Creation of a governing body for auditing search results, possibly within the FTC.

Why wouldn’t this work?

The strongest argument against regulating based on my concept of “digital public goods” is government incompetence. The government has a poor record in managing large technology projects and “all the smartest tech people are in the private sector”. There is lots of truth in that.

But I’d argue that despite their failings, the government is actually the absolute most qualified group to manage these digital public goods:

  • Cloud computing is comparable to a natural resource – water, energy, gas and many other natural resources are run by the government or under heavy government control.
  • Consumption data is highly sensitive personal information – the government already has highly sensitive information on us. In fact, their data trove would make a digital marketer squeal: through census, taxes, licensing, and more, they know our marital status, income, and interests. Giving them more information is scary, but at least there are checks, balances, and security measures that private companies simply don’t have.
  • Search interfaces are about protecting consumer choice – the government has long had a hand in regulating consumer choice. For example, their mandate on warning labels on cigarette boxes.

Giving control of these public goods to the government isn’t perfect. But like democracy itself, it’s our least-worst option for a healthier market and more just society.

Leave a Reply

Your email address will not be published. Required fields are marked *