Incumbents, Not Startups, Are Regulatory Arbitrageurs

Plus! Limits to Automation, Silver Lake, 1830, 1930, 2030, Globalizing Cybersecurity

This is the weekly free edition of The Diff, the newsletter that tracks inflections in finance and tech. Last week’s subscriber-only posts:

And as a reminder, the price of a subscription to The Diff rises to $20/month or $220/year at midnight tonight. Current subscribers will not be affected, so sign up today to pay the current price of $15/month (25% off) or $150/year (32% off).

Incumbents, Not Startups, Are Regulatory Arbitrageurs

A fun thought experiment is to ask: what if the new product had been invented first? Suppose Juul existed, under its current regulatory regime—with national restrictions on flavors, local bans on e-commerce, and restrictions on advertising and coupons. All prudent in light of uncertainty: dozens of people died from other vaping products using illegal adulterants, nicotine is addictive, and with a new product people put into their bodies, there’s a risk that it will have unforeseen side effects.

Imagine that in this world, some entrepreneur decides to buy the raw material that Juul processes and vaporizes, and then convinces customers to wrap it in paper, set it on fire, and inhale the result.

This practice would obviously be seen as dangerous, and would be significantly restricted or subject to an outright ban. As the American Cancer Society notes, “Tobacco smoke is made up of thousands of chemicals, including at least 70 known to cause cancer… The levels of many of these substances [in e-cigarettes] appear to be lower than in traditional cigarettes, but the amounts of nicotine and other substances in these products can vary widely because they are not standardized. The long-term health effects of these devices are not known, but they are being studied.”

Clearly, restrictions on e-cigarettes make sense. But it’s equally clear that the mix of chemicals in a burning leaf is much harder to monitor and control for quality than a mixture of glycerol, propylene glycol, nicotine, benzoic acid, and flavoring agents.

Imagine the pitch to the FDA:

“What does it do?”

“The same thing as a vape. It gets nicotine into your bloodstream, which is a stimulant.”

“Nicotine is addictive. Is your product addictive?”

“Oh, yeah, a lot more, actually. The MAOI inhibitors in some of the other ingredients make it way more habit-forming.”

“Other ingredients? How many does your product have?”

“Uh, thousands? It’s hard to keep count.”

“Are they safe.”

“Only a few dozen of them are known to cause cancer.”

“Can’t you take out those ingredients?”

“We’d have to genetically engineer a new plant to get to a ‘maybe’ there.”


Of course, there’s a decent chance that if cigarettes had never been invented, vaping wouldn’t be legal, either. The US government is very skeptical of novel substances. They even tried to crack down on caffeine in the early twentieth century, leading to one of the best-named legal cases of all time.

Other companies can tell a similar story. If Uber had existed before cabs, the idea of cabs would sound ridiculous: it’s Uber, except that there’s a limited number of accounts, you have to buy one from someone else, the rates are standardized, you can’t review your driver, your driver can’t review you, and there’s no app. (The taxi system is especially bad because it’s a fairly unpleasant, unskilled job, but a job that requires a heavy upfront investment, so medallion owners end up deeply in debt.)

It’s a story that’s much older than apps. Interactive Brokers had one of the greatest engineering-based comeuppances of all time: in the 80s, their predecessor company pioneered electronic trading, hooking a computer up to their Nasdaq terminal to automatically enter trades. The industry association that ran Nasdaq informed him that he needed to manually enter trades on a computer—so he built a robot that typed the trades in. There’s one good reason to ban automated trading: it’s a lot more efficient than the other kind, and Nasdaq’s other members didn’t like the competition.

There are a few reasons startups look like regulatory arbitrages:

  1. Many of them do violate laws, at least initially. There are a lot of rules out there. Plenty of startups are started by idealists who can’t imagine that anyone would ban what they’re doing. But they run afoul of broadly-scoped laws. Different businesses operate under different legal norms; in software, nearly everything is legal by default; in finance or healthcare, it’s best to assume that everything you want to do is illegal, and then see if you can figure out an exception.
  2. In many cases, the place where they scale the fastest is a legal gray area, or something actively illegal. In Juul’s case, clearly many of their users were underage, and they didn’t do much to address this. (From a strict rule-enforcement perspective, the fact that cigarettes smell bad and that smokers smell like cigarettes is a benefit; it means detection is easier. On the other hand, one lesson the music industry learned the hard way is that it’s suboptimal to treat every customer like a potential criminal.)
  3. The PR hype cycle.

The hype cycle is a strong and underrated force. Most startups, at launch, get fairly positive PR: it just feels wrong to bully a tiny upstart. But if a company gets uniformly friednly coverage, that means any bad news is an interesting scoop. In business journalism, as in wrestling, a Heel Turn is good for ratings.

Journalism has a challenging supply chain: most blogs optimize for quantity. TechCrunch posts 235 articles a week, The Verge 204, Mashable 100, Yahoo Tech 374, Business Insider a gobsmacking 1,108. It’s hard to have that many good ideas in a week, but the PR industry is there to help, with prepackaged stories, charts, and narratives.

Startups do PR—one of the services that early-stage investors provide is introductions to friendly journalists, and another service they provide is enforcing embargoes on stories to maximize the PR punch. But big companies tend to have much larger, more experienced PR departments, so they’re a lot better at getting their narrative across. Those big companies also have big legal departments, so they can quickly identify rulebreaking by competitors.

If all this sounds like an endorsement of startups ignoring regulations, it’s not. Or, if it is, it’s a qualified one. Laws need to adapt as circumstances change. A law that mandates notification by mail should be updated after the invention of phones, and again after email, and again after text messages. But when an established company operates within an obsolete legal structure, it has an incentive to preserve margins by keeping that legal structure obsolete. Startups, meanwhile, can bootstrap their way into getting the legal code they want by skirting the rules early on, and then going legitimate once they’re well-established.

In a sense, this entire process is an information-discovery process. When the iPhone came out, nobody understood that taxi medallions were obsolete. It took ride-sharing companies to do that. But if companies wantonly disregard the rules, they’ll impose costs on everyone else. Novel companies really have two regulation-driven tasks:

  1. Work with governments to craft a set of rules that make sense in light of changing technology. (Tech company lobbyists are, of course, very biased. But the lobbyists on the other side are biased, too, and it’s not immediately obvious that one side should be more morally pristine than the other.)
  2. Work like a government on their own platform, providing basic government-like services—contract enforcement, dispute resolution, safety—to users.

The boundary between what the private sector should do and what the public sector should do is not purely a matter of ideology. Even an apolitical technocrat would have to change their opinion as circumstances changed. But one of the ways we find out which laws need to be updated is when a new company gets a lot of happy customers and a lot of happy investors because it turned out to be breaking the rules.

A Word From Our Sponsors

Build vs Buy Decisions

Every business will eventually face a decision like this – “Should we custom-build this technology or should we buy something off the shelf?” There is no catch-all answer, and like most things, the truth lands somewhere on a contextual spectrum. But the right solution is usually “both". The trick is knowing what tools to leverage and how to combine them. This is the art of creative assembly, a term coined by the team at Versett. They work with CEOs to help launch new capabilities, businesses and apps from scratch. Every month, millions of people use the products Versett has built and their team has spent the last seven years tracking the results. You can think of them as the anti-McKinsey – specialized practitioners with hard-won digital experience.

Have a question you are mulling over internally? Readers of The Diff can text Jonathan at 1-403-477-0184 to ask anything related to business and technology or read Versett’s thesis to learn more.


In yesterday’s issue I noted a strange-looking tax provision in ZoomInfo’s S-1. I’ve been informed by a few readers that this is standard when a company reorganizes into an Up-C structure, as ZoomInfo did. For more on the Up-C structure, see this.


Limits to Automation

The Information (subscription, worth it) has a review of Apple’s efforts to automate manufacturing, and where they failed. In 2011, Foxxconn’s CEO suggested that the company would have 1 million robots by 2014. Today, they’re at 100k. The Information mentions two culprits:

  1. Miniaturization: if you have a task that requires heat and pressure sensors packed into a small device adjacent to moving components, the best tool for the job is the human hand. This might be a local maximum, but it’s a fearsome one: electronics mass production coincided with a dramatic increase in China’s available labor supply, so the entire process got optimized around affordable factory workers. As it turns out, some of those workers' functions simply can’t be replaced by a machine.
  2. Short product cycles: humans can be retrained faster than robots can be rebuilt, and robots are less fault-tolerant. This points to another reason that robots might eventually contribute more to Apple’s manufacturing: the device cycle is slowing.

(As a side note, The Information’s reporting on Apple’s supply chains has been stellar. They’ve covered lots of the operational details that don’t make it into 10-Ks and earnings calls, but that are essential to get right if the phones are going to get made.)

Silver Lake

In April, I noted that “Silver Lake’s involvement in multiple deals suggests that somebody there is positive on the online travel space as a whole.” The FT has profiled that Somebody, their co-CEO Egon Durban. Two things that stood out from the article:

  1. Durban has shown an occasional preference for making investments in companies run by very well-networked CEOs. Endeavor Entertainment was a bad bet on paper, but being the first call Ari Emanuel returns is certainly good for deal flow. (For an example from another firm: one reason A16Z got the Clubhouse deal was that they got more celebrities to test it out.)
  2. Silver Lake’s travel investments are, in part, a pharma bet: “Privately, Mr Durban has said that he thinks other investors are discounting the promise that pharmaceutical companies can successfully develop and distribute a vaccine for Covid-19…” If you’re bullish on a vaccine, don’t invest in vaccine companies; the economic outcome will get decided by the media and the government, not by the consumer surplus from ending a pandemic. Betting on travel is the right way to do it—assuming it’s the right bet. (Perhaps Silver Lake’s next big deal will be recapitalizing WeWork: WeWork’s model is to cram more people into smaller offices, so even the mildest social distancing protocols are brutal for them.)

In other Silver Lake news: Reliance Jio, which they recently invested in, has raised yet another giant round, this time from Abu Dhabi’s Mudabala sovereign wealth fund. Silver Lake has done a deal with Mudabala before, investing in Manchester City. And Mudabala and Silver Lake were introduced by… Ari Emanuel.

1830, 1930, 2030

Derek Thompson has a great summary of the current economy:

The COVID-19 crisis is simultaneously thrusting Americans into the pre-urban homestead economy of the 1830s, re-creating the Depression-era joblessness of the 1930s, and pulling forward the virtual economy of the 2030s. We are living in the weirdest economy ever.

That’s exactly right. One specific phenomenon it explains is tech companies' outperformance. Since their valuations are based on highly uncertain results in the distant future, any event that brings the future closer is disproportionately good for them.

Google Dissolves the Webpage

Google’s mission is to organize the world’s information. That used to mean sending users to whatever web page was most likely to answer their question, but increasingly means aggregating the data and displaying it to users directly. (A search for “weather 10016” used to bring up Now it brings up a weather report in the search page itself.) They’ve taken that a step further, testing out sending users directly to the part of the page that answers their question. This is a tiny tweak, but a telling one. It’s Google’s way of saying that the right way to “organize the world’s information” shouldn’t treat the webpage as the basic unit of content.

Globalizing Cybersecurity

I wrote last month about the peculiar economics of cybersecurity. Since the Internet is global, but hacking laws are national, some countries have an incentive to tolerate a domestic hacking industry as a source of soft power. But finance is also international, and the EU plans to take advantage of that by sanctioning Russian hackers. It’s a start, but not definitive, as many Russian nationals seem to revel in not caring about sanctions (in a profile, Russian propaganda wizard Vladislav Surkov said “I see the decision by the administration in Washington as an acknowledgment of my service to Russia. It’s a big honor for me. I don’t have accounts abroad. The only things that interest me in the U.S. are Tupac Shakur, Allen Ginsberg, and Jackson Pollock. I don’t need a visa to access their work. I lose nothing.”).

Private Equity in 401Ks

Private equity is, among other things, the only way underfunded defined-benefit pensions can plausibly claim to meet their obligations. Now, 401Ks can get in on the action. Target-date funds can now invest in private equity, which will allow them to theoretically offer higher returns. AQR has made a compelling case that private equity funds' excess returns come from inaccurate markdowns when the market drops—the industry, in the aggregate, is worth the fees but not worth substantially more than the fees. That appeals to the cynicism of defined benefit pension fund managers, but has a straightforward appeal to retail investors, too.