Summary: the error is assuming that a system of higher surveillance and punishment, necessarily improves overall citizen safety and economic performance (there can be an economic issue as well). That’s not accurate because the people in charge of such systems, and who (possibly illegally) have access to such data, are not necessarily benevolent. The specific best response varies by the measure in question; below are a number of explanations that may apply.
The question is asked in various contexts,
“Why do you oppose public surveillance cameras/wiretapping at will/massing of information into super-databases/badge swiping/tracking of your movements? You obey the law, right? What do you have to hide?”
Plenty – there are a number of significant threats against which personal information must be protected. It could be economic gangsters, it could be whatever evil ruler happens to take power, now or in the future, or it could just be people making mistakes, because the cost of gathering and assessing accurate, vetted information, is high enough that it doesn’t happen.
Collecting information makes a lot of crimes cheaper to commit without providing a compensating crime reduction:
– Identity thieves thrive on a wide variety of personal information. In addition to the normal threats like disclosure of Social Security numbers, they also need information about family members and friends, pictures, and general activities that aids them in their ability to mimic you. Identity thieves do not obey the law, so those of us who do must take actions to reduce the attack surface we expose by disclosure of information.
— This is even worse if these measures capture biometric information; you will watch all your passwords and financial data go on the public Internet.
– Stalkers and criminals for hire such as kidnappers and hitmen use location information about you to commit their crimes; reducing the attack surface by lowering your profile is critical against these threats.
– There are many laws against discrimination based on skin color, age, and other factors that are either tangentially or completely relevant to performance, but in many cases, such as denial of employment due to these characteristics (e.g. HIV status), enforcement of these laws is impractical in the current economic and legal framework. In the context of these flawed systems, in order to minimize the opportunities for adverse actions to be taken against you, information regarding these characteristics must not be routinely exchanged.
– There are also areas such as the underwriting of health insurance where good, proactive acquisition of information such as genetic tests may legally be punished by denial of service or other adverse impact. Whether a citizen abides by the law is irrelevant.
– That information is used by popular (mob-rule) campaigns, either directly to lynch individuals (e.g. India Whatsapp), or to ruin their businesses and employment (Papa John’s founder who was thrown out for saying a particular word in a prep session designed to prevent that person from using that word in the actual press conference). Even if the state does not carry out these actions, you still ruin people’s lives by building these systems.
The collection of information creates new situations where mistakes and ordinary criminal tendencies become magnified:
– A law-abiding citizen also has reason to fear the mistakes of others. When someone’s identity is stolen, bad information gets into personal information databases like ChoicePoint, or arrest records without convictions become easy to access, law-abiding citizens (more people than just the immediately affected person) have to spend a great deal of time trying to correct or expunge these records, if such is even possible. The gathering of such information and its use for high-stakes decisions, like allowing people to get a job or rent an apartment, means that ample measures should be taken to ensure the accuracy of the information. However, even though a citizen may be law-abiding, that does not mean that information brokers and users have sufficient monetary incentive to correct mistakes or to attempt to sort out the conflicting reports. Thus, the law-abiding citizen may wind up paying the price for others’ mistakes when we create these monster databases.
– There is also a small segment of the security forces like the police that is corrupt and willing to frame others. The threat from these people is there regardless of whether any automated surveillance or information-gathering measures are taken. However, with crimes such as photo-enforced traffic violations, if the automated system is assumed to be infallible, then if the criminals gain control of it, they will use it for extortion.
In fact, law-abiding citizens have the most to fear from the technological prison state, because the downside of such constructs disproportionately affects them, vs. a more conventional human patrol system. The technological police state is vastly more effective at detecting thoughtcrime, underclass in high-class neighborhoods, racial composition, and other measurements that allow evil laws to be enforced, and much less effective against typical just law violations like physical damage. As one example: consider a speed camera vs. a police officer. Both can be deployed to fine speeders who more or less obey the law, but only the police officer can chase and stop severely reckless, drunk, etc. drivers. (incidentally this is a core reason why I prefer police officers even for conventional speed traps) Moreover, the police officer could choose to look the other way in a conventional police setup, whereas the technological police state allows for no disobedience to the evil laws.
More generally, we persist in finding ways to frame law-abiding citizens (and not specify safe harbors/proper controls) by wastefully building up the now obsolete concept of political culture where these sources of information are considered reliable indicators of guilt or innocence. Generative AI makes many information sources of evidence essentially non-indicative; provenance of evidence is critical to verifying guilt. However, the point of the technological (vs. the conventional) police state, is to record every activity of society, and then later on assess punishment, since you know where the offenders are. Since all you get out of the technological police state is easily-generated videos, no properly aware jury could ever convict solely on this basis. Hence, even given an assumption that the government is trying to be bound by the law, the technological police state is essentially useless at obtaining convictions; it only has value in cueing police to the scene, where, if those police are trustworthy, they could then discover evidence, etc. Two examples:
– Insider trading. By rough definition, this involves trading on non-public information, and the proof of it relies on the records keeping/call logs/etc. of the people suspected of passing on the information. All of that is pointless if the government cannot be trusted, because the government can generate masses of “guilty” logs at any time. The effort spent in building the credibility of the technological police state is futile when you see e.g. grossly political prosecutions, because no one can trust the provenance of the location data, the texts, etc. More generally, law enforcement cannot punish most insider trading with those means anymore: there have to be far stricter trading controls (e.g. you can only trade for specific reasons which are enforced by specific timings) and far more information transparency (i.e. open books and receivables, etc.), that limit the amount and duration of insider information, and that limit the time windows in which insider traders can exploit their information. Saying “I’m going to force people to use controlled channels and punish people for going outside them” is not going to work in any way, and building a technological police state to try to enforce it, is therefore a fool’s errand.
– Often, the government knows what your DNA is, and they run the breathalyzers, etc. Thus, relative to certain status quos, the technological police state is superfluous to one corrupt cop (and of course corrupt government), if that cop can get at all your damning information (or make it up, since they own the breathalyzers) in one computer browsing session. But, its existence provides a ready explanation for “oh how did the government find out that you were an evil person”. They don’t need to involve other corrupt cops, unreliable jailhouse birds, drug abusers, jilted lovers, political enemies, etc. if all they have to do is play a card with the jury that somehow they detected it in their NYC mass-surveillance super-LCD war room. This means a lot when people like Luigi Mangione (no commentary here on any aspect of his innocence or guilt) are alleged to have been caught on the basis of only this evidence. The same hand-waving that a mass of cameras somehow caught an odd angle on Luigi Mangione, is the same hand-waving that can send all the law-abiding opponents of a regime to prison (e.g. for alleged participation in a violent protest cooked up by the supporters of a regime).
Having raised all these actual and potential harms, we should return to the most correct means of argument, the complete quantitative perspective, such as:
– If you say I want to stop the speeders with a bunch of cameras, how many collisions, injuries, and deaths will I save with these, vs. the alternatives (e.g. actual police officers)/status quo? (remembering humans can also be employed to stop riots, etc.)
– If you say I want to deter and prosecute the murderers by tracking every person everywhere, how many murders will I avoid with these, vs. the alternatives/status quo?
– And how much is all this going to cost me, if I am trying to prevent property crimes?
The answer on all of these types of questions is sometimes “not that much” and usually “negative” when it is weighted against any significant downside risks noted above, and you account for the costs of implementation. Beijing East Asia implements these systems throughout society, and the total policy cost (including the technology cost) is one big reason why they are so poor relative to their manufacturing strength.
The obvious response is “well I’m only going to implement cameras” or “I’m only going to implement search term tracking”, so you avoid the complete implementation of the technological prison state, so avoiding most of the downside risk. This is a meaningful statement; then the question would be, which ones will you implement, and what is the threshold at which the marginal downside risk outweighs the marginal benefit?
Consider the matter from the perspective of how much of the technological prison state already is in place in relatively free Western society:
– Likely almost all of your financial and business records and property are electronic, and you probably are not the custodian/physical protector of most of your property
– You and your fellow citizens have a bunch of computers, cellphones, cars, and other objects, that already form a tracking web, whether or not we choose to directly centralize/maintain records of it
– Your criminal records, academic records, employment history, marriage history, credit history, transaction history when using credit cards and/or online services, demographics like birth certificate, and such electronically are held by various arms of the state and private industry
– Likely your medical records are becoming electronic as well, and placed on the Internet
– Much of your Internet history likely is being saved forever
– The state has many hooks to spy on you at any time e.g. via the Plain Old Telephone System (and the enemy states do too, e.g. Volt Typhoon/Salt Typhoon)
– Even if you don’t use these devices, other people take pictures and videos of you
– Participating in moderately or highly paid work requires the use of and contact with all of these items (only fairly casual work does not)
– Individual businesses and many government facilities run a series of (mostly) closed-circuit cameras and authentication systems
– There are some tolling mechanisms, airline watch lists, and other similar monitoring
What is missing?
– Expanded audio and camera coverage
– Full review of recordings/fully centralized monitoring facilities
– Similarly full monitoring of computing devices/cell phones, including access to encrypted communications
– Some paper records are outside the system
– Casual labor and large parts of travel outside hotels, planes, etc.
– Cash transactions
Most of the above measures, whether it was right or wrong to use them, already were taken because of economic savings and/or practicalities. In particular, it would be very costly to repeal the digitization of the financial system, and the cellphones and computers (plus the cars if they are hooked in) inherently are trackable. Much of your life history has solid reasons to be held somewhere, such as citizenship verification for voting, or who decides what happens with your property or your life when you get sick or die. From an economic perspective, cellphone cameras and microphones will not go away, so the capability for personal surveillance amongst the general public is a given.
Additionally, certain measures, such as your Internet history being saved forever, you may not be able to stop since criminals (e.g. empires like Beijing East Asia) are some of the ones trying to make copies. Nor is it obvious that given the clear benefits of instant access to information and communications, you actually would choose to stop using the Internet because of its clear economic and citizenship (e.g. candidate and history research) benefits.
So where is that threshold/tipping point of downside risk? It can’t be at the point of implementing all these measures, but as you can see, most of the basic mechanisms required to implement the prison state economically are givens or highly desirable of themselves. Only the extent of surveillance, and the existence of encrypted media/off-grid capability “loopholes”, distinguish the technological posture of modern Western society from that of the anti-Muslim/non-Han prisons in East Turkestan.
Here is where I believe the true problem with the recommendations for additional surveillance exists; how could we definitively show, in a manner we both independently could verify, that we haven’t already crossed the threshold where marginal harm outweighs marginal benefit? Given the examples of the Stalinist and Maoist pre-technological police states, clearly the police state can be implemented with informers and “papers please!” Technological analogues of those are your cellphone, your Internet history, your communications metadata, and so on. So the means already are in place for the authorities – it’s a question of their will to use them. That would indicate we already have passed the threshold.
The alternative argument is “we’ve already gone too far” and passed the threshold, so we’re already at the mercy of the authorities, and any additional harm is minimal, and therefore we should implement the remainder of the technological prison state to capture the effectively free marginal benefits. The logic would be sound, but is there significant marginal harm remaining? Consider some use cases that do work under the current normal Western status quo as outlined above, but don’t work within a full technological prison state:
– If you want to kill your rulers, you have some space to organize, some means to communicate, and you probably have some semi-effective means to do so easily within your grasp. Even as apparently one man, Brendan Crooks was a couple inches off from killing the former and future American Emperor in 2024.
– If you want to spread uncensored news and information generally, you have a number of means to enable this (for example, take a look at Cuban thumb drive networks)
– If you want to flee the country or move across legal jurisdictions, you still have some limited capability to move around undetected
Those are the types of measures that can save your life, organize resistance, and that can eliminate clear threats. Hence while it may be that we already have passed the marginal threshold, clearly things can get a lot worse even if we have passed such a threshold.
Hence, at least any such proposed enhanced surveillance measures must be obviously incomplete and relatively easy to circumvent – otherwise an ordinary citizen could not any longer perform those important use cases. If that’s the case, then how effective can those additional surveillance measures be? Obviously their effectiveness is bounded above by the intelligence level and capability of the criminals.
Yet, by historical evidence, typical measures in place in Western countries already catch a number of criminals: cellphones get tracked down, cars are traced within and across state lines by toll booths, police cars, and other cameras, and sometimes even humans are traced out in public, particularly in cities like London where there are significant numbers of public cameras. To what extent are the criminals caught by these measures stupid, and how much stupider are they vs. the average citizen? That’s a hard question to answer because of multivariate, difficult to measure IQ, but in my review of various criminals and their behavior, the average criminal is not that much dumber than the average citizen. Much of the variation is the influence of intoxicants, loss of emotional control, and similar factors that likely would not apply for the use cases above.
Hence, from an empirical perspective, it’s not clear how to determine the raw capability of a criminal. Thus the manner in which I view the situation is:
– Obviously a functional test, meaning you ask a bunch of random citizens to perform the above use cases, is the better approach to assessing the capability
– If you are going to ask a bunch of random citizens to do it, then in order for those things to be viable, you need 80-90% of the citizens to be able to pull it off; otherwise, your information distribution network would fail due to detection by the authorities, and many people would be unable to flee.
— You could accept less on the killing one and maybe still be successful in the end, but note the historical example of Hitler assassinations: a very poor record despite dozens of attempts. So it’s probably incorrect to tighten that one up.
– Because you need almost all citizens to be able to circumvent in these types of situations, and because the criminals aren’t obviously that much dumber than the average citizen, most criminals would be able to circumvent controls in these situations.
In conclusion, the whole thing makes for a very dicey situation, and in that situation, I don’t see how we can reach anything remotely approaching definite agreement on costs vs. benefits on additional surveillance. Hence it’s extremely difficult to spend the money and take the additional risk, in a general situation (vs. say a sting operation).
(There are other problems with the apparatus of the police state like the discomfort caused by the lack of privacy from taking intrusive measures, but it is not necessary to delve into these when the threats previously discussed already indicate the need to avoid the construction of the no-privacy society.)