Under this interactionist definition of harm, could we have some kind of coherent and unified way of explaining what all types of harmful actions have in common? — TheHedoMinimalist

The fact that the harm was interactionally derived guaranteed previsibility which is necessary for a principle to work. The harm must not be phenomenologically open in order for the principle to work, it must be phenomenologically closed.

I think some communities, especially more religious ones, may conceptualize harm as something that could be impersonal and victimless. — TheHedoMinimalist

Your problem is with the harm, or more precisely, with actions that truly concern only the self and more specifically, whether or not actions really can be private because of the social embeddedness of human actions. Two problems emerge with assessing the significance of harms : which harms are relevant and the privacy of the action that caused the harm.

The relevance of harm is hardly a settled matter. Are harms instigated by false beliefs real harms? Is a harm a mental state or are there other relevant matters for assessing harm? Furthermore, does the way in which the harm was brought about relevant? Should we think that merely allowing a bad thing to happen is sufficient in order to consider constraining it? Or should we in fact think that only doing harm is relevant? Maybe we should criminalize willful doing of harm. But does willfully allowing harm essentially equate to willfully doing harm?

For example, they might say that smoking weed causes harm because it upsets God or because it violates the sanctity of the community even if done completely privately. — TheHedoMinimalist

If privacy matters, even if it hurts your values, you can be against criminalization of that act but be morally against the commission of the particular act. See for example the prime minister of Canada, P-E Trudeau in 1969 when he decriminalized homosexual intercourse :

We don’t allow homosexuality, we’re just saying we’re not going to punish. We will not send the police to the bedrooms to see what is going on between consenting adults in private. […] We are taking the idea of ​​sin out of the Criminal Code. — Pierre E. Trudeau

Another commentator, Thérèse Casgrain, commented on a case where people were imprisoned :

It is inconceivable that we so arbitrarily lock up an individual for acts which do not violate public decency since they are committed privately or against the integrity of young people, since they were committed between consenting adults. — Thérèse Casgrain

Maybe Trudeau and Casgrain were against homosexual intercourse, but privacy did matter when assessing the significance of the harm committed. It seemed clear, in that case, that actions between two consenting individual were not a public affair, because homosexual intercourse is conceptualized as a singular event which did not concern public law, even though it shocked more than the two persons involved in the action.

Thus said, I hardly see how these matters actually affect our agreement over the the harm principle. We can agree with the harm principle without agreeing to a common theory of the significance of harm. We can agree that 1) an action which only hurts me 2) jand does not impact someone else 3) in a significant manner in order to constitute it as harm, should not be criminalized. We can agree on the principle (albeit a very reinterpreted one) without agreeing with what precisely is a significant harm comprised in condition 3. Therefore, the harm principle isn’t in any way vague about what conditions are relevant for prohibition, but it is quite harder for it the be specific about what it prohibits.

If the definition of a harm was prealably deliberated as involving no phenomenological assessment ffrom the person being stolen, there is no difficulty in assessing anaction using the harm principle.

If the the mere fact that a person loses possession of a good is considered as a harm, without need to consider the individual's phenomenology, it can be considered as a harm whether or not she cares. If I break my arm doing skateboard and consider it no big deal, I am harmed whether or not I feel pain. If I get stolen without caring about it, I am harmed whether or not I care. If our definition of stealing implies that you take possession of something without permission, and consider this as a harm, the personn has less than she would have if she had not been stolen from, and thefore it is hurtful for her whether or not she cares. We don't really need to assess her phenomenology farther.

Of course most psychological harms are phenomenological (like suffering is much harder to live by than simply experiencing pain), but if we assert that certain kinds of phenomenologies are harms, it is compatible with the harm principle. For example, if you make me angry by smoking weed, and that you consider the phenomenological emotion of angriness as being a harm, it is a harm whether or not I think angriness is harmful or not. I may agree that since you are smoking weed, I am angry, but I may maintain that it is not harmful for me to be angry, since honestly, the feeling of shouting your lungs after someone kind of feels good after all.

If we have an interactionist definition of harm, preferably, one which was prealably deliberated with the community, we have no trouble applying the harm principle.

The harm principle (or my interpretation) simply says ''If X does not imply Y, decriminalize X''. The rationale being : ''if Y is not harmful, the reason we have for criminalizing X vanishes''.

The problem is when someone proposes that the definition of ''harm'' is purely and simply phenomenological (as you seem to be suggesting is the correct definition of harm) : '' i say that it is a harm, so therefore it is harm''. We cannot predict what kind of event would generate this kind of assessment, and therefore this would really make the harm principle unpredictable and vague.

The action of smoking weed does not necessarily imply harm towards others, even if some who use weed may hurt their relationships because of it, it isn’t a necessary consequence of the action of smoking weed.

However, imagine a powerful warrior’s drug that gave an urge to assault someone : the consumption of the drug would necessarily imply harm towards other, and I think the harm principle would justify banning this drug and its use.

I think that this critera of ''innateness of harm'' makes the harm principle quite clear for decriminalizing kinds of actions.

I think the question is not whether or not the harm principle is clear (according to my interpretation), I think the question is whether or not we want to consider all merely probable phenomenological harms instead of harms planned by the law, and the latter would require deliberation. The former would make the harm principle quite vague. So maybe it's the notion of harm that is vague.