In that same vein, there're questions like, "If you discover someone has violated this policy, will you be able to stop them from continuing to do so?" For example, the US routinely indicts North Korean hackers, but unless those hackers visit a place that the US can inveigle into arresting and extraditing them, it's a mostly symbolic gesture:

https://www.justice.gov/usao-cdca/pr/3-north-korean-military-hackers-indicted-wide-ranging-scheme-commit-cyber-attacks-and

3/

One undertheorized aspect of administrability is "fact-intensivity"; that is, are there difficult, fact-intensive questions that need to be answered in order to determine whether someone has violated the policy?

Think of probate: probate is often a lengthy and expensive process, especially if the deceased is "intestate" (has no will). To probate an estate, all the deceased's assets have to be cataloged and assessed, claims of heirs and inheritors have to be evaluated, etc, etc.

4/

People spend a lot of time and money creating wills and family trusts largely to answer these questions when they're easiest to resolve (when you're still alive and can clearly express your preferences), because it's even more expensive and time-consuming to answer these questions when you're not around anymore to weigh in on them.

5/

As complex and time-consuming as managing your estate can be, there's nothing wrong in theory with having a complicated, careful process in place for dealing with it. Taking care of your loved ones and disposing of your assets is something that's worth getting right, and people have all kinds of highly individual preferences for this that requires a lot of flexibility in the system.

6/

Making a system that's very customizable but also robust against fraud (or even honest mistakes) requires a *lot* of administrative superstructure to hold it all together.

And besides, probate isn't something we have to do very often. After all, most of us will only die one or fewer times. It's not like we have to figure this stuff out every day. It's the kind of thing you can do every couple of decades, over several hours, spread out over weeks.

7/

*Frequency*, then, is the enemy of *fact-intensivity*. If you had to do probate-level form-filling to buy a cup of coffee or pay your electricity bill, that would be *nuts*. For one thing, it would be full employment for lawyers - and it would cost so much that by the time you got to the cafe or the gas-pump, you'd be too broke to actually complete the transaction.

8/

This comes up a lot in discussions of tech policy, because once you computerize something, you can start to do it *very quickly*, which means that policies that added, say, a 1% admin overhead to a task before it was digitized can add up to a 1,000% overhead once it's digitized.

The best example of this is copyright: copyright is the most fact-intensive doctrine you deal with on a day to day basis.

9/

Technically, *conclusively* determining whether you have the right to forward an email could take a lawyer a whole day. Sure, *most* email forwarding is "fair use" (that is, it fits into one of copyright's "limitations and exceptions"), but any decent IP law prof could come up with ten email forwarding hypotheticals in ten minutes that could occupy a whole fourth-year IP law class for an entire semester.

10/

One of the reasons copyright is so fact-intensive is that it was designed to be invoked infrequently. We're talking about a legal regime that was designed to answer questions about book and music publishing (and then adapted for other kinds of media), and even the most prolific publisher or label is going to deal with double-digits' worth of new works per season.

11/

Meanwhile, the people working at that same publisher are likely forwarding hundreds, if not *thousands* of emails per *day*. If the publisher's copyright lawyers had to review every one of those forwards, they would never publish another book. They would go bankrupt.

Obviously, that's not how things work.

Why not, though?

12/

Well, mostly because we just pretend copyright law isn't there. To the extent that we do acknowledge the potential for copyright liability from everyday activities that no one ever asks a lawyer to sign off on, we manage that liability through shitty, one-sided contracts.

13/

You have undoubtably clicked on dozens of agreements this year wherein you warranted that nothing you were doing violated copyright law (a neat trick, given that you probably have *no idea* whether any of the activities you routinely engage in could violate copyright) and further, you indemnified someone else for "all costs arising from any claims" associated with your activity.

14/

That's an *unbelievably* shitty, one-sided clause for you to have "agreed" to, since "any claims" includes *claims with no merit* and "all costs" includes "money we paid someone who brought a bullshit claim to just go away."

15/

In other words, you routinely click through these nonsense "agreements" where you promise to give *every cent you have* to anyone who wants it, if the company that made you click through that bullshit decides to promise some deranged rando a million bucks to settle their wild accusation that you violated their copyrights.

16/

For complicated reasons, we're not all drowning in copyright lawsuits all the time, but if someone *really* wanted to fuck you up and they had deep enough pockets, they could use the fact that you're a giant, routine copyright infringer (just like everyone else) to *wreck* your life for *years*.

So obviously, it would have been better if we'd done some major refactoring of copyright law once the internet came along.

17/

My preferred fix? Carve out activities unrelated to the media industry's supply chain from copyright *altogether*:

https://pluralistic.net/2023/10/21/the-internets-original-sin/

Copyright isn't the only fact-intensive doctrine that's challenged by the cadence of digital life. The internet lets us do a *lot* of things, *very* quickly, meaning that even small factual questions pile up beyond any reasonable capacity to resolve them.

18/

Take the debate over content moderation and hate speech. Hate speech and harassment online are serious problems and they disproportionately affect people who are getting the shitty end of the stick in the offline world, too. The legacy platforms obviously don't give a damn about these people, either.

19/

So it's tempting to attempt to use policy to solve this real problem. Even if the US wasn't being run by a trollocracy, this would probably be a nonstarter in America, because hate speech is protected by the First Amendment, and purely speech-based harassment is hard to punish without falling afoul of 1A.

20/

But other countries - notably the EU - are having a go at it. I think this is a doomed effort - but not because hate speech isn't a serious problem! Rather, because hate speech regulations are *very* fact intensive, and hate speech is *very* common. Frequency is the enemy of fact-intensivity.

Say the EU creates a rule requiring platforms to take reasonable measures to prevent hate speech.

21/

This requires

1. arriving at a common definition of hate speech;

2. adjudicating whether a given user's speech rises to that definition; and

3. determining whether the platform's technical measures were "reasonable."

This is the work of months, if not *years*. And hate speech happens hundreds of times *per minute* on the big platforms. It's just not an administrable policy.

22/

Now, just because policy isn't administrable, it doesn't follow that there's nothing to be done. There's other ways to give relief to the targets of harassment and hate speech. To get to those ways, we have to ask ourselves *why* people who are tormented by trolls stay on the platforms that expose them to abuse.

23/

There are plenty of extremely wrong explanations for this floating around. One is that Mark Zuckerberg and Elon Musk are Cyber-Rasputins who can hypnotize us into using their platforms even if we don't like them, by "hacking our dopamine loops." This is a very silly explanation: everyone who's ever claimed to have perfected mind-control was a liar and/or deluded:

https://pluralistic.net/HowToDestroySurveillanceCapitalism

24/

Another is that people are lying (possibly to themselves) when they say they don't like being harassed on legacy social media platforms. This theory - from neoclassical econ - is called "revealed preferences," and it holds that people whose actions go against their stated preferences are "revealing a preference" for the thing they're doing.

25/

This is the sort of thing you end up believing in if you incur the kind of neurological injury that arises from pursuing an economics degree, which causes you to be incapable of reasoning about (or even perceiving) power. "Revealed preferences" tells you that if someone sells their kidney to pay the rent, they have a "revealed preference" for having one kidney.

26/

Thankfully, there's a much simpler explanation for people's continued use of platforms where they are subject to abuse and harassment. It's this: the only thing worse than being a member of a disfavored minority who is subject to abuse and harassment is being a member of a disfavored minority who is subject to abuse and harassment *who is also isolated from your community*.

27/

Leaving Facebook or Twitter means leaving behind the people who comfort and support you when you are subject to abuse. The more abuse and discrimination you face, the more that support matters, and the harder it is to leave that community behind. You love your community more than you hate Zuck or Musk, so you stay, because as much as you love them, it's transcendentally difficult to coordinate a mass departure for somewhere else.

28/

This is called the "collective action problem" and it's a regressive tax on the most abused platform users and communities.

*This* is a problem we *can* solve with policy! We can mandate that platforms support interoperability, so that when you leave a legacy platform like Twitter or Facebook for a modern platform like Mastodon or Bluesky, the messages addressed to you on the legacy platform are forwarded to your new home.

29/

That way you can have the people you love without the platform you hate.

This is a very administrable policy. The main lift is figuring out the nuts and bolts of interoperability, and while that's a big technical project, it's the kind of thing you only have to do once or twice.

30/

Then, if a platform fails in its duty to forward your messages after you leave, it's very easy for a regulator to determine whether it's violating the rules - they just have to send a message to your old account and see if it shows up for your new account:

https://pluralistic.net/2022/12/19/better-failure/#let-my-tweeters-go

A hate speech policy is hard to administer because it requires resolving a bunch of fact-intensive questions.

31/

A "right to exit" policy replaces all those fact-intensive questions with a bright line policy ("if you don't forward your former users' messages, you are guilty"), which can be administered at high speed.

32/

Whenever a fact-intensive policy that regulates an infrequent activity fails because the activity becomes more frequent, you have two choices: you can either slow down the activity, or you can replace the fact-intensive questions with bright-line tests that can be resolved much more quickly.

But more often, we fail to do either, and everything goes very badly indeed.

33/

That's more or less what's happened with "merger scrutiny," the part of antitrust law that lets competition regulators (or competitors) block or put conditions on mergers that involve large firms.

In these merger scrutiny cases, plaintiffs who challenge a merger are expected to resolve a bunch of *extremely* fact-intensive questions. Fail to resolve any of these questions and the merger goes ahead.

34/

The most pernicious fact-intensive question that arises in antitrust cases is "market definition." That's pretty much what it sounds like: "What market is this company doing business in?" If you can prove that the companies in a proposed merger are in the same market, then it's a lot easier to prove that allowing the merger would reduce competition.

35/

The problem is that "market" is a very slippery concept. As Tim Wu describes in his excellent book *The Age of Extraction*, "market definition" creates a near-infinite amount of wiggle-room:

https://www.wired.com/story/tim-wu-age-of-extraction/

When Wu was serving in the Obama FTC, he had a front-row seat for Google's acquisition of Waze.

36/

Now, obviously these companies are direct competitors, but the Obama administration wanted the merger to go through (it was dominated by people who thought monopolies are efficient and didn't want to do their jobs). So these officials decided that Google Maps' market was "finding out where you are" and that Waze's market was "getting you somewhere." It was really that stupid.

37/

Writing for the *Law and Political Economy* project, Hal Singer explains how the fact-intensive nature of the "market definition" question makes it virtually impossible to prevent market concentration and abuse of dominance:

https://lpeproject.org/blog/the-market-definition-trap/

From Livenation/Ticketmaster to Paramount/Warner Brothers, the "market definition trap" leaves the public virtually defenseless before efforts to reorganize the economy into extractive, rapacious cartels.

38/

In a recent interview with the Do Not Pass Go podcast, Paul Crampton (Canada's recently retired top competition judge) talks about the tsunami of mergers that Canada's Competition Bureau is expected to oversee:

https://www.donotpassgo.ca/p/inside-canadas-competition-court

Fact-intensive market definition questions can't possibly be resolved at the pace of mergers. That's because companies' preferred growth strategy is combining, rather than competing.

39/

There's plenty of political problems with merging Paramount and Warner, but there's also a huge *economic* problem, because these companies are direct competitors who will soon operate as a single firm.

The M&A industry has staged a denial of service attack on its regulators, accelerating the pace of mergers involving large firms far beyond the ability of a regulator to resolve the fact-intensive questions these mergers raise.

40/

They've flooded the zone, and after the mergers go through and the companies start abusing their customers, workers and competitors, these same market definition questions bedevil any attempt to rein in this abuse of dominance.

Singer makes some excellent suggestions for legal reforms to resolve this, moving some of the fact-intensive questions to bright-line ones, such as "whether the challenged conduct injured workers, consumers, or some other counterparty."

41/

This is the right approach. As we plan for a future in which legislatures recognize the enormous harms that monopolization inflicted on our societies, we need to come up with more bright-line rules for antimonopoly rules. These will lack some of the subtlety that fact-intensive treatment affords, but you can't do fact-intensive adjudication for high frequency activities.

42/