Closely related to the dictator's dilemma is "authoritarian blindness," where an autocrat's censorship regime keeps them from finding out about important, socially destabilizing facts on the ground, like whether a corrupt local official is comporting themself so badly that the people are ready to take to the streets:

pluralistic.net/2020/02/24/plu…

2/

The modern Chinese state has done more to skillfully navigate the twin hazards of the dictator's dilemma and authoritarian blindness than any other regime in history. Take Xi Jinping's 2012-2015 anticorruption purge, which helped him secure another ten year term as Party Secretary.

3/

Xi targeted legitimately corrupt officials in this this sweeping purge, but - crucially - he only targeted corrupt officials in the power-base of his rivals for Party leader, while leaving corrupt officials in his own power base unscathed:

web.archive.org/web/2018122216…

How did Xi accomplish this feat? Through intense, fine-grained surveillance, another area in which modern China excels.

4/

Chinese online surveillance is often paired with censorship, both petty (banning Winnie the Pooh, whom Xi is often mocked for resembling) and substantial (getting Apple to modify Airdrop for every user in the world in order to prevent the spread of anti-regime messages before a key Party leadership contest).

5/

But there are many of instances where China spies on people but *doesn't* censor them, even when they express dissatisfaction with the state. Chinese censors allow a surprising amount of complaint about official incompetence, overreach and corruption, but they completely suppress any calls for *mobilization* to address these complaints. You can be as angry as you want with the government online, but you can't call for protests to do something about it:

science.org/doi/10.1126/scienc…

6/

This makes perfect sense in the context of "authoritarian blindness": by allowing online complaint, an autocrat can locate the hot-spots where things are reaching a boiling-over point, and by blocking public manifestations, the autocrat can prevent the public from turning their failings into a flashpoint that endangers the autocracy.

7/

In other words, autocrats can reserve to themselves the power to decide how to defuse public anger: they can suppress it, using surveillance data about the people who led the online debate about official failures to figure out who to intimidate, arrest, or disappear.

8/

Or they can address it through measures like firing corrupt local officials or funding local social programs (toxic waste cleanups, smokestack regulation, building schools and hospitals, etc) that make people feel better about their government.

9/

Autocracy is an inherently unstable social situation. No society can deliver everything that everyone in it desires: if you tear down existing low-density housing and build apartment blocks to decrease a housing shortage, you'll delight people who are un- or under-housed, and you'll infuriate people who are happily housed under the status quo. In every society, there's always someone getting their way at the expense of someone else.

10/

Obviously, widespread unhappiness is inherently socially destabilizing. After all, no society can police every action of every person. From littering to parking in disabled parking spots, from paying your taxes to washing your hands before serving food, a society relies primarily on people following the rules without even though their face little to no risk of being punished for breaking them.

11/

The easiest way to get people to follow the rules is to foster a sense of the rules' legitimacy: people may not agree with or understand the rationale for a rule, but if they view the process by which the rule was decided on as a legitimate one, then they may follow it anyway.

12/

This legitimacy is a source of social stability. Sure, your candidate might lose the election, or the government might enact a policy you hate, but if you think the election was fair and you believe that you can change the policy through democratic means, then you will be on the side of preserving the system, rather than overturning it.

13/

A democracy's claim to legitimacy lies in its popular mandate: "Sure, I don't like this decision, but it was fairly made." By contrast, a dictator's legitimacy comes their claims to *wisdom*: "Sure, I don't like this decision, but the Supreme Generalissimo is the smartest man in history, and he says it was the right call."

14/

You can see how this is a brittle arrangement, even if the dictator is a skilled autocrat who makes generally great decisions: even a great decision is going to have winners and losers, and it might be hard to convince the losers that they keep losing because they deserve to lose. And that's the *best* outcome, where an autocrat is *right*. But what about when the autocrat is *wrong*?

15/

What about when the autocrat makes a bunch of decisions that make nearly everyone consistently worse off, either because the autocrat is a fool, or because they are greedy and are stealing everything that isn't nailed down?

16/

Every society needs stabilizers, but autocracies need more stabilizers than democracies, because the story about why you, personally, are getting screwed is a lot less convincing in an autocracy ("The autocrat is right and you are wrong, suck it up") than it is in a democracy ("This was the fairest compromise possible, and if it wasn't, we need to elect someone new so it changes").

17/

The Snowden revelations taught us that there is no distinction between commercial surveillance and government surveillance. Governments spy, sure, but the most effective way for governments to spy on us is by raiding the data troves assembled by technology companies (for one thing, these troves are assembled at our own expense - we foot the bill for this spying whenever we send money to a phone or tech company).

18/

The tech companies were willing participants in this process: the original Snowden leak, about the "PRISM" program, showed how tech companies made millions of dollars by siphoning off user data to the NSA on demand:

en.wikipedia.org/wiki/PRISM

19/

It was only later that we learned about another NSA program, "Upstream," through which the NSA wiretapped the tech companies' data-centers, acquiring *all* of their user data, and then requesting the data that interested them through PRISM, as a form of "parallel construction," which is when an agency learns a fact through a secret system, and then uses a less-secret system to acquire the same fact, in order to maintain the secrecy of the first system:

eff.org/pages/upstream-prism

20/

Upstream vs. PRISM

After Edward Snowden leaked NSA slides in 2013, two names became synonymous with the agency’s vast online spying powers: Upstream and PRISM.

^{Electronic Frontier Foundation}

Upstream *really* pissed off the tech companies. After all, they'd been dutifully rolling over and handing out their users' data in violation of US law, risking their businesses to help the NSA do mass spying, and the NSA paid them back by secretly spying on the tech companies themselves! That's a hell of a way to say thank you to your co-conspirators.

21/

After Upstream, the tech companies finally started encrypting the links between their data-centers, which made Upstream-style collection infinitely harder:

arstechnica.com/information-te…

22/

Yahoo will encrypt between data centers, use SSL for all sites

CEO Marissa Mayer promises Yahoo will be locked down by March 2014.

^{Sean Gallagher (Ars Technica)}

But that hardly ended the mass surveillance private-public partnership. Congress continued to do nothing about privacy (the last federal consumer privacy law Congress gave Americans is 1988's Video Privacy Protection Act, which bans video store clerks from telling newspapers about the VHS cassettes you take home) (we used to be a *country*).

23/

That meant that tech companies could collect our data will-ye or nil-ye, and that data brokers could buy and sell that data without any oversight or limitation:

pluralistic.net/2025/02/20/pri…

24/