Should one resist an attack?
If you are under a physical attack by another person (say, mugging or knife threat), and you cannot simply walk or run away, you have several ways of reacting:

1. appease the attacker, - offer them money, or invent a way to apologize to them, etc.
2. threaten the attacker - tell them you will call the police, or that you will hit them very hard, or pull out your handgun and threaten to use it, etc.
3. attack them yourself first - hit them in the eye and then run away, etc.

It would be interesting to know from empirical data how people statistically react in these situations, split by country, age group, and gender, and what are the actual outcomes. Anecdotal evidence (such as, available descriptions of life in prison societies, in the army, etc.) suggests that #3 is the most effective, followed by #2, while #1 leads to increasing losses.

Some arguments advanced in favor of #1 are as follows,

• the attackers will become more violent if you threaten them, and less violent if you appease them
• violence - even a threat of violence - is morally wrong, we should not "behave like them"
• an effective threat or a counterattack is difficult, gun training requires effort, and there is no reason to put in that effort because physical attacks are comparatively unimportant: they happen much less frequently than, say, car accidents or just falling and getting hurt in your bathroom
• the attackers are underprivileged victims of society, and we - the privileged - are at fault for their situation, so we should apologize and accept their behavior

Some arguments in favor of #2 and #3:

• empirically, a threat of violence or a direct counter-violence works well as a deterrent
• you can avoid counter-violence only if you are so much stronger or more powerful than the attacker that their threat is truly unimportant; for example, they will be satisfied and walk away with $100 while you are a millionaire; or you are a great martial arts expert and they can't even hit you. If you are not in this position, you cannot avoid talking to them in the language they understand.

I'd like to analyze the "statistical argument" - namely, that terrorism is a much less frequent threat than car accidents. Indeed, statistically this may be true at the moment, and perhaps will remain true in the future. It is usually said that humans are "irrational" in that they perceive certain risks as more important than others, despite the actual probabilities being lower. However, let us consider the typical human behavior in the face of a threat or a risk of any kind.

Example 1: you walk in the street and suddenly you encounter a pool of muddy water in your way. If you walk carelessly now, you risk falling into the mud. Your normal reaction is to avoid the mud. However, if we apply the "statistical argument", we need to examine the risk of actually falling into the mud. Yes, there is a possibility that there is a big hole in the pavement that you do not see because it is covered by the water. But, streets in your city are generally well maintained, and so this risk is statistically much less significant than the risk of, say, dying in a car accident. Also, the temporary loss due to getting dirt on your face and your clothes is much less significant than, say, the temporary loss of health due to a flu infection - which happens much more regularly to you than falling into the mud, and yet you are doing very little or nothing to prevent your flu infections. Therefore, by the statistical argument, you should ignore the mud and let yourself fall into it, should that actually happen.

Example 2: you are in a municipal office in a provincial town, waiting to get a certificate of death for a distant relative who recently passed away; it just happens that no other family members are available or have time to do that, so you stepped in. The municipal clerk asks for the last name of the deceased; you say it, but the clerk mishears, repeats the name back to you incorrectly, and starts typing into the computer. Your normal reaction would be to interrupt the clerk and to say the name again or to show a printed document with that name spelled correctly. However, let us apply the "statistical argument" here. The possibility of error is real (and the death certificate might need to be reissued, costing time and money), but it is highly unlikely that you will have to do this kind of thing ever again. The statistical frequency of loss due to this error is certainly lower than the statistical frequency of losing money due to incorrect billing at a grocery store, which you routinely accept. Therefore, by the statistical argument, you should ignore the error and let the clerk issue a death certificate with an incorrect spelling, should that actually happen.

The conclusions forced by the statistical argument seem absurd enough. This is so because the "normal" behavior is formed by millions of years of evolution rather than by logical or statistical arguments. Is evolution "wrong"?

Evolution certainly did not give humans an effortless intuition for solving complicated problems of probability theory. Only very few humans can learn to do that, and only after decades of concentrated effort.

Evolution optimizes for survival, - in other words, the "normal" behavior is actually the behavior optimal for survival in the real world with its real dangers, where we actually have "skin in the game" and are forced to act on the basis of incomplete information.

So, the real question is whether the sophisticated tools of probability theory are relevant to a given task or not. I would say that the "statistical argument" does not properly apply to making a decision about a single event where you have specific information. The "statistical argument" pretends that you don't actually know what is happening right now, but instead you are in a situation where the present event will be repeated a very large number of times, and you are trying to optimize the average odds of a positive outcome. This consideration is inappropriate in a lot of real-life situations - such as, falling into the mud, misspelling names of dead relatives, or trying to avoid getting mugged.