That's not really how theory works; the claim isn't that this is the single and sole determinant of civil unrest. This Atlantic article[1] looks more broadly at the effect of the young male subpopulation's size on unrest, this[2] paper references the mitigating effect of family formation on individual crime rate, this[3] one both covers unrest due to low family formation (generally through the mechanism of uneven sex ratios, given how out of favor polygamy currently is), as well as effects on individual crime.

That being said, this little excursion into Google turned up some nascent evidence that this historical dynamic is shifting. The modern world is very odd in lots of different ways: wealth, recreational opportunities, the incredibly wide social circles one's exposed to. It's entirely possible that this dynamic has changed, but the research hasn't caught up yet.

[1] https://www.theatlantic.com/international/archive/2014/03/an...

[2] https://d101vc9winf8ln.cloudfront.net/documents/33306/origin...

[3] https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5983495/

But you presented the theory as a causative link between family formation being unavailable for young men and civil unrest.

It's a classic case of correlation is not causation. Yes, there have been societies where men couldn't form families that were violent (like the Wild west in the US) but that doesn't mean that that was the cause of it.

Similarly, many on the left have long thought that porn viewing causes sexual violent crime against women. And there is ample evidence showing that men who view porn are more likely to be sexual offenders than non-viewers. But then came the internet and the amount of porn and porn viewers skyrocketed - but the amount of sexual violence against women didn't - therefore busting the theory.

In the same way, evidence from Japan, China, South Korea and even India means that there can't be a causative link between family formation being unavailable for young men and civil unrest.

Are the plethora of cases where cigarettes did not cause cancer compelling proof that they don't?

Perhaps these counterexamples are evidence of additional factors, in cancer or in unrest, but they do not prove that a causal relationship doesn't exist.

If you had a very large group of heavy smokers and if you could demonstrate that the lung cancer prevalence among this group was not higher than among the general population then yes, that would be compelling evidence against the theory that smoking causes lung cancer.

Say you went to a city with no air pollution, a high rate of exercise, great diets with antioxidants and omega-3s, no asbestos or radon, and no genetic predisposition to lung cancer. And you find that, in this city, heavy smokers have a lower rate of lung cancer than the general population of the world.

Do you conclude that the overwhelming abundance of data that shows that smoking causes lung cancer is wrong? That there cannot be a causative link between smoking and cancer because there is a counterexample where it didn't appear? I would say that these other factors are also having an effect of comparable scale.

Yup, I would conclude that if the heavy smokers in this city have a lower rate of lung cancer than the rest of the inhabitants in that city that smoking doesn't cause cancer.

So you have a comparison between unrest in Chinese people who are starting families and those that are not? You keep trying to push this analogy into a territory where the evidence is a lot stronger than what you've given. You compared the country to other countries, and it's analogous to comparing a city with a higher cancer rate to a city with a higher smoking rate and saying that smoking cannot cause cancer.

Data suggests that social unrest in China has increased in correlation with incel count. I would count that as supporting evidence for the theory above. But you're saying, correct me if I'm wrong, that, because the initial rate of unrest was so low, and the increase didn't push that rate above that of other countries, incels cannot be a causal factor in unrest?

Help me understand where we disagree here. What part of ´inability to start a family->unhappiness->social unrest' do you think is questionable?

You don't seem to realize is that you have a theory X -> Y and I can find some X for which X -> Y is false, then the theory is false.

Regardless of the theory.

I have found such X. In principle you agree, but you want to save your theory so you say "Well, if X and not Z then Y!" and that explains my counter example because my group probably had Z.

Here Z being the racist idea that Asian people are less violent than others and that that's why social unrest hasn't erupted in their countries.

You might be right. But then your theory is "inability to start a family and non-asian -> social unrest." Clearly, not the same as "inability to start a family -> social unrest."

The theory is that an increase in X causes an increase in Y. You've pointed out cases that have a high X and low Y as counterpoints. But in order to make the point you're making you need to provide evidence that increases in X have not lead to increases in Y, so we can control for initial X.

Back to cancer, heavy smokers who did not develop cancer are an example of some X for which X->Y is false. Like you said, you would need to take a random sample of cases and see whether Y correlates with X against a control group.

You introduced the concept of Asian countries having lower rates of social unrest, and I have never said this is related to race at all. Perhaps you are reading into the genetic confound in the cancer example, but it was just intended as an example of a confound within the analogy. There are plenty of differences between life in these countries and elsewhere that can act as confounds.

A causal link might be still there, but the theory might only explain a few percent of the variance, due to many other (confounding) factors.

Not sure what variance has to do with it? Thing is, you can always "patch" an X leads to Y theory when faced with opposing evidence by blaming confounding factors. But that is not very scientific. See https://en.wikipedia.org/wiki/Russell%27s_teapot

Sure, you're right (also https://www.gwern.net/Causality#overview-the-current-situati... )

I'm just trying to say that very few things are actually that causal in big complex systems. I'm trying to say that such issues are very likely have many-many factors, and we see an aggregate effect.

And so in case of such aggregates arguing for-against them with very simple first-order reasoning is kind of useless. Because each piece of data (supporting or counter evidence) just explains a part of the big picture. So, without accounting for all data we can't really make general statements about what causes the aggregate effect that we see.

Yes, we can use "very simple first-order reasoning" even when we argue about very complex problems. If you say that "swans are white" it is enough for me to find a single black swan to prove your hypothesis wrong.

That is exactly what is being done here.