Taking Children Seriously (TCS) is a parenting philosophy by David Deutsch and Sarah Fitz-Claridge. It advocates a non-coercive, anti-authoritarian, fallibilist approach. It draws inspiration from Karl Popper's fallibilist epistemology and from libertarianism. Deutsch was my mentor. I learned a lot about TCS and was a fan. TCS was largely abandoned by its founders 20 years ago and became a pretty dead movement. Some recent attempts to revive it do not appear to be going well and I'm unaware of any TCS advocates who are open to debate today.
This essay discusses philosophical errors in TCS theory, some of which are related to Critical Fallibilism, the epistemology I developed to improve on Popper's Critical Rationalism. While I don't recommend people try to follow TCS parenting practices, I still think TCS is intellectually interesting, partly because of its connections with Critical Rationalism, and partly because I still agree that conventional parenting is flawed, irrational and mean. Like many reform movements, TCS was right about more of its social critiques than its solutions.
TCS Introduction
TCS interprets disputes between parents and children as disagreements about ideas rather than as disobedience, misbehavior, mistakes, human nature, sin or wickedness. TCS says the parent is fallible and may be wrong, and disagreements should be approached rationally with truth-seeking methods that are not biased in parents' favor. Punishment is a clearly irrational way to deal with disagreement; debate is more appropriate. A parent shouldn't be an authority with the final word who can get his way at any moment just because he says so and he's physically stronger.
TCS rejects forcing children to do anything they don't want to. TCS is highly critical of schools and recommends something along the lines of extreme, pure unschooling (which is like homeschooling but without a set curriculum or anything else resembling school). TCS also criticizes parenting behavior that's illegal to do to other adults, such as hitting someone or physically moving them to a different location against their will. TCS also had some weird, deeply problematic ideas, including about doing some things to children that are currently illegal, e.g. TCS favored abolishing age of consent laws. There are old TCS discussion emails, including by Deutsch, that are dismissive of the danger from child predators, but I think that attitude is factually mistaken.
TCS rejects the goal of passing on all of the parents' ideas to the children. Instead, it says children should be free to choose what to agree and disagree with. In this way, it's expected that each generation will reject and correct errors from the previous generation and humanity will rapidly improve. TCS had a tendency to think utopia was within reach in just one generation if people would listen. Anyone who rejected TCS was seen as standing in the way of a better world by intentionally choosing to be an irrational child abuser. Despite advocating more gentle parenting towards children, TCS was quite harsh with parents and would call them abusers for doing normal, conventional, socially acceptable actions. TCS even compared mainstream parents to slaveowners since they don't give their children total freedom, but that comparison is an error.
TCS advocates dispute resolution by common preference finding. The basic idea is to use creativity to find a solution that everyone is happy with, not coerced by. It's like seeking mutual consent plus more (begrudging consent isn't good enough). TCS's focus was on specifying what attributes a good solution would have (no one is coerced) and on arguing that that goal is possible not impossible, but TCS didn't give enough practical guidance (there was nothing like a common preference finding flowchart to follow). I did develop a common preference finding method myself (which I think is a good start but inadequate to enable most people to do it).
Coercion
TCS redefined the word "coercion" to mean:
the psychological state of enacting one idea or impulse while a conflicting impulse is still active in one's mind.
This is an inherently binary definition: either a person is in this state or they aren't. This definition doesn't specify any amounts or degrees. And TCS advocates sometimes denied that any coercion is less bad than any other coercion, which fits with a binary view. In line with that attitude, Deutsch responded (mirror) to an article about traumatic teen wilderness camps by denying that those camps are worse than other schools (even though the article talks about prosecutors filing over 100 charges of physical and sexual abuse for one of the camps).
One of the major, original ideas of my Critical Fallibilism (CF) is to evaluate ideas in a binary way, with criticisms either refuting or not refuting ideas. CF rejects credences, rejects the concept of arguments strengthening or weakening ideas by degrees, and rejects the approach of evaluating ideas by how good they are then reaching a conclusion based on which idea has the highest evaluation. CF puts a lot of effort into discussing what binary ideas are, how they're different than matters of degree, how to work with them effectively, how to avoid potential problems with them, how to get benefits from them, what degrees still can be used for, etc. It's a major change to approach epistemology in a binary way and it requires a lot of explanations and methods to make it work and replace everything that it's logically incompatible with.
TCS brought up but didn't answer some potentially confusing issues. If coercion is binary, then whipping a child isn't more coercive than yelling because there's no such thing as more coercive. So is whipping worse in some other ways? Should we be looking at other issues besides coercion? But TCS focuses heavily on coercion. If whipping isn't worse, what does that mean for how we live or how we judge parents? Whipping could be replaced with something worse, and yelling with something milder (TCS sometimes used frowning as an example of coercive parental behavior), and the same questions still apply.
When you approach an issue in a binary way, and it's counter-intuitive to most people, a lot of analysis and explanation is needed to help people make sense of it, which TCS didn't provide. TCS didn't say their coercion definition was binary, let alone explain how to deal with that. I searched over 49,000 TCS emails and only found two short conversations that discussed coercion being potentially binary. There was one in 2002 and I brought it up myself in 2004. TCS emails began in 1996 and the TCS founders never discussed coercion being binary.
Can CF's analysis of binary evaluations of ideas be used for TCS? I think it's relevant and useful, but you'd have to change it and add some new stuff to try to make it work for TCS. I haven't done that so I don't know how well it would work out. There are significant differences between binary evaluations of ideas as refuted or non-refuted and binary evaluations of states of mind as coerced or not coerced. A refutation means an idea fails at a goal, and we can consider multiple goals and give an idea multiple evaluations (one per goal), so within CF's binary approach we can differentiate one idea as better than another, even though they both have the same evaluation for one goal, because they have different evaluations for other goals. TCS's goal with coercion is total avoidance, whereas with CF it's fine if an idea is refuted for some goals: you can still use it for other goals. (In fact, for any idea, you can specify many goals which it fails at.)
Fallibility and Mutual Consent
TCS seeks mutual consent as a way of dealing with fallibility. People can be wrong. Even when you're really confident, you might be mistaken. If you ignore critics without their consent, you may be wrong and stay wrong even though a better idea is available.
Why does consent matter? If someone thinks they have an important correction of an error you're making, which they want to share, it's intellectually dangerous to ignore that. Consent means that someone agrees with you (they aren't a critic) or they are fine with you not changing your mind at this time (so e.g. they don't see their correction as important or don't want to make the effort to share it).
TCS correctly points out that children are people with ideas who may be correct when they disagree with their parents. TCS also has arguments about how children are selective about what they disagree with their parents about, and this selectivity can make up for their ignorance, allegedly giving children around a 50% rate of being correct when they disagree with their parents. It can of course be much higher or lower than 50% depending on the child and the parents. But the common assumption that the parent is right 99% of the time is unreasonable. Fallibility isn't just a technical point about edge cases; it's pretty common that parents are wrong when they disagree with their children. (I don't agree with the 50% claim but I do agree that parents are mistaken reasonably often.)
Note that the parent being mistaken doesn't mean the child is correct about what they want to do. It's often easier to be right about a criticism than a solution or proposal. It's often easier to point out errors than avoid errors. Just because a child correctly recognizes a flaw in a parent's idea doesn't mean the child's alternative idea is correct or better. It's common that the initial ideas of the parents and children are all incorrect.
Note: This section about fallibility uses my own way of explaining TCS. I think I made presentation improvements while staying accurate to TCS's claims.
Handling Fallibility
Seeking mutual consent runs into some difficulties. It can be too time consuming to answer every critic who wants your attention. And after you answer, they might still disagree and want more answers. They might have reasonable followup points, or they might be a beginner who is effectively seeking free tutoring disguised as critical debate, or they might be an unreasonable person who doesn't listen well.
The mainstream, conventional answer to this is to legitimize ignoring dissent when you judge that engaging with a critic isn't the best use of your time. You estimate which critics have the most promising points, if any, and ignore any person or idea that you want to ignore. This basically enables an unlimited amount of bias. I consider it one of the largest errors in our society which is systematically making all of science and academia much less productive or rational.
TCS's answer is, at least with your own children, to basically spend unlimited time and energy addressing disagreements. Just keep at it until you get mutual consent about what conclusion(s) to reach or what to do.
TCS and Deutsch didn't give clear guidance on how to handle fallibility regarding online debates with strangers. I personally tried to follow their principles by seeking mutual consent to end debates. I tried to be open to criticism from anyone. I tried not to unilaterally judge that a conversation should end. I tried not to be dismissive of ideas even if they sounded dumb to me because, as a fallibilist, I didn't want to risk being wrong about that decision to be dismissive. I even engaged with ideas that I'd already addressed in the past because I could potentially be mistaken that it's a repeat of a past idea, rather than a similar idea that's actually correct. Or even if some ideas are identical to ones I refuted in the past, this person could have an additional argument that the past person didn't which changes the final conclusion.
Although my debates were time consuming, overall I found them productive and I learned a lot. While I'm sure there was room for improvement, engaging in a lot of debate was valuable for me. However, after around ten years, the value of my debates started noticeably going down for me. Debates started getting too repetitive, both about the specific ideas (like hearing the same old pro-induction or anti-fallibility arguments again) and about the meta patterns (like people making logic errors, failing at reading comprehension, or quitting in the middle).
I developed a solution which I called Paths Forward. The basic idea is to have error correction mechanisms other than continuing to participate in a discussion. That let me decline or end more discussions because I knew that, even if I was wrong about the value of that discussion, there was still a way I could be corrected. Having backup plans for error correction enabled me to opt out of far more discussions without the typical risk of being and staying wrong. I've found this highly effective at protecting my time and energy. I've concluded that it's unnecessary for intellectuals to be bad fallibilists; they should have error correction mechanisms like a debate policy.
Error correction mechanisms, beyond the first one of using your good judgment, require some degree of autonomy or independence to be effective. They should be external things, generally written down in advance. They need the ability to overrule your judgment or exert control over you. If they couldn't, then they wouldn't be able to correct you when your judgment is mistaken. This is similar to a king giving up some arbitrary power to have written laws which even he must follow. Error correction mechanisms are powerful when you genuinely give up some power or authority.
For me, having error correction mechanisms is far less work than putting basically unlimited energy into critical discussions and debates. However, most intellectuals are used to arbitrarily ignoring whatever they want with no transparency. For them, this approach requires them to do more work (and give up some power), which many of them don't want to do. I got the effort required to be rational down from unbounded to maybe 10 hours a month, which I thought was a really good accomplishment. But most intellectuals spend 0 hours a month on being rational so 10 hours still sounds like too much work to them. Also, I think a lot of intellectuals avoid error correction mechanisms because they're making a lot of errors and would in fact be corrected, which they would find embarrassing and unpleasant (they don't consciously view it this way, but they intuitively avoid the sorts of debates they would lose).
The basic issue is how to say "no" without being an authoritarian, irrationalist or infallibilist. It's very important to have good answers to that. Being unable to say "no" is extremely harmful. Being authoritarian, irrationalist or infallibilist is also bad.
So the mainstream approach to fallibility is to be arrogant and have inadequate interest in error correction. And the TCS approach, at least with one's children and ambiguously with everything else, is to put unlimited energy into dealing with one's fallibility. TCS, by advocating mutual consent (or more) as the goal, is demanding people put more energy into error correction on any issue as long as a counterparty isn't satisfied. This is impractical. The mainstream is right to reject this. Neither side (TCS or mainstream) has any practical ideas for how to rationally handle fallibility and error correction in a limited, finite amount of effort. In that context, while I sympathize with TCS's aspirations, I think the mainstream view is better than TCS's view: arbitrarily rejecting error correction (with people encouraged to do their best to be rational and avoid bias, which isn't very effective) to protect people's autonomy is better than placing an unlimited burden on people.
As a fallibilist philosopher, I worked on solutions to the problem of dealing with fallibility well using limited effort. My solutions were developed primarily in the context of online debate between public intellectuals. I think significant modifications would be needed to apply them to parenting. But I think that's what should be done to create a good, new parenting philosophy: adapt CF's error correction mechanisms and develop new ones for interactions between parents and children. I don't expect the results of this project to be an incremental improvement on TCS. I expect it to create a new, separate approach to parenting that takes some inspiration from TCS (particularly the idea of applying Popperian fallibility to parenting). I haven't done this. I haven't worked out the details. Perhaps this project would fail for some reason. But I think I figured out a good lead for how to develop a good parenting philosophy.
Authority and Responsibility
In general, we pair authority with responsibility. If you were responsible for a decision made by someone else, it wouldn't make sense and would be unfair to you. Whoever has control over something is also the person responsible for the outcome. Imagine if a company hired two people: one whose job is to make decisions and another whose job is to be blamed if the decisions are bad. That wouldn't make sense and would be especially unpleasant for the person being blamed for things he had no control over.
TCS takes authority away from parents and tells them to defer to their children when they fail to resolve disagreements through rational discussion and mutual consent:
where there is no common preference found, the parent must self-sacrifice ... Occasional failures, or even frequent minor failures, to find solutions, are probably inevitable, and we endorse parental self-sacrifice as the best way of making them less harmful and less frequent.
TCS also said parents shouldn't self-sacrifice, self-sacrifice is coercive, common preference finding should fail infrequently, people should not ask "What if?" or "Who should rule?" type questions, people should focus on solving underlying root cause problems instead of on solving the immediate problem, people should focus on problem solving not assigning blame, and that asking about what happens when finding a common preference fails shows that you have the wrong attitude and therefore will be unable to find many common preferences. This is confusing. TCS didn't have a good, clear answer here.
The law and society in general hold parents responsible for parenting outcomes. If a child doesn't brush his teeth and gets cavities, that is the parent's fault. If a naked toddler wanders outside in the snow and freezes to death, that is the parent's fault. If the child asked the parent to let him go in the snow with no supervision, and the child kept insisting and wouldn't listen to the parent's concerns, and the parent followed TCS's advice and deferred to the child, that is still the parent's responsibility even though TCS gave the child the authority to get his way and make the decision.
This isn't just the mainstream opinion. TCS also holds parents responsible. If you defer to your child and then he gets a bad outcome, TCS blames you (you should have given your child better advice and warnings, and been more persuasive).
You may think toddlers wandering around in the snow is absurd but TCS's founders communicated very clearly that TCS allows no exceptions and that parents must be open to common preference finding, and acknowledging that they may be mistaken, even in worse scenarios than that. Parents who said they definitely wouldn't let their kid do something really dangerous were called irrational, mocked, and pressured to keep an open mind and seek a common preference instead, and told to never ever use their authority to get their way if common preference finding fails.
Splitting up authority and responsibility, so that one person makes a decision and someone else is responsible for the consequences, is an error. It doesn't work in companies, families, or anywhere else. If you want a rational family, you need to help parents be more rational, not take away their authority (while having no way to take away their responsibility). While you can put some limits on parental power (and society does, e.g. laws against child abuse), parents need to learn to use their power rationally, not try to stop having the power that is paired with their responsibility as parents. Parental responsibility is a burden which parents took on and must use their power to handle well (and should gradually transfer as children grow up and approach independence).
To use their power rationally, parents must handle their fallibility well. They must know that they may be mistaken even when they feel confident that they're right. Neither arbitrarily dismissing disagreements nor putting unlimited energy into discussing disagreements is a good approach. So my abstract suggestion is that parents approach their fallibility with error correction mechanisms similar to how CF advises public intellectuals to have debate policies.
I think one reason TCS didn't want to give parents permission to ever use their authority without their child's consent was because TCS was too controlling towards parents. TCS's founders cared what parents did and pressured them to change their behavior to better suit the preferences of the TCS founders. This is problematic even when the TCS founders have good preferences (like that children shouldn't be spanked or yelled at). It doesn't adequately respect the freedom and autonomy of parents. It tries to transfer some authority from parents to TCS leaders without transferring corresponding responsibility.
Parents and Culture
TCS says parents and children together control children's outcomes. And children aren't born bad. Therefore, if parents changed their behavior and were great, their children would definitely turn out great, and if many parents followed TCS then society would soon be great.
I now think culture is more powerful and influential than parents in many ways, and that even great parents could easily raise mediocre children who accept lots of mainstream errors.
Parents can be more influential than culture by keeping their children extremely isolated, but they shouldn't. Merely homeschooling and being anti-social isn't even close to isolating enough. I mean more like being in a cult or living off-grid with the children not having internet access or any friends with internet access.
I think if you want a better world, you need to figure out how to deal with culture. TCS tried to be a shortcut to a rational world without figuring out how to help adults become rational, just by having adults change their parenting behavior so that the next generation would be rational. I don't think that shortcut can work.
This is related to another mistake TCS made, which was to view coercion as intentionally caused by other people. The idea was that if parents just stopped coercing on purpose then children wouldn't be coerced. I think self-coercion and unintentional coercion are both common and that cultural ideas can cause coercion.
Conclusion
TCS defined coercion in a binary way. TCS didn't recognize this, acknowledge the many difficulties with it, and provide appropriate explanations and analysis to address those difficulties. CF uses a binary approach to evaluating ideas, but it clearly says so, addresses many concerns, and offers a lot of analysis and explanation.
TCS's attitude to fallibilism puts a basically unlimited effort burden on parents (and potentially on everyone, throughout their entire lives, if they apply the principles universally). Putting such a big burden on parents is really harmful and I think this did harm a lot of parents without them recognizing what the source of the harm was. So I think my identification of this issue is important for people who were involved in TCS (and hopefully intellectually interesting for others). And dealing with fallibility is an important philosophical issue that I'd already worked on abstractly and regarding public intellectuals and debates. A middle ground is needed between unbounded effort to handle fallibility or legitimizing arbitrarily dismissing criticism and requiring zero effort. Working on handling fallibility with limited effort in the parenting context, including applying my work about error correction mechanisms to parenting, is a potential way to develop a successful rational parenting philosophy.
Transferring authority from parents to children, without transferring corresponding responsibility, doesn't work.
Culture can be more powerful than individuals including parents.
These are complex ideas and this essay has focused on high level summary. To understand in more detail, you could read my many other essays about TCS, epistemology, fallibility, Critical Rationalism, Critical Fallibilism, decisive arguments, Paths Forward, debate policies, debate methodology, error correction mechanisms, etc.
Messages