There are two issues here. The first is determining what is the worst thing that a person should express. The second is the issue of determining the worst thing that a person should be allowed to express. While these might seem to be the same issue, they are not. There is a distinction between what a person should do and what is morally permissible to prevent a person from doing. The focus will be on using the coercive power of the state in this role.
As an illustration of the distinction, consider the example of a person lying to her boyfriend about playing video games when she was supposed to be doing yard work. She should not lie to him (although there are exceptions). However, the police should not be sent to coerce her into telling the truth about this. So, she should not lie about playing video games, but the state should allow her to do this.
This view can be disputed and there are those who argue in favor of complete freedom from the state (such as anarchists) and those who argue that the state should control almost every aspect of life (totalitarians). However, the idea that there are some matters that are not the business of the state is an intuitively plausible position. What follows will rest on this assumption and the challenge will be to sort out these two issues.
A plausible and appealing approach is to take a utilitarian stance and use Mill’s principle of harm to determine the worst thing a person should express and should be allowed to express. The right of free expression is limited by the right of others not to be harmed in their life, liberty and property without adequate justification.
In the case of the worst thing that a person should express, I am speaking in the context of morality. There are, of course, non-moral meanings of “should.” To use the most obvious example, there is the pragmatic “should”: what a person should do in serving their practical self-interest. For example, a person should not tell her boss what she really thinks of him if doing so would cost her a job she desperately needs. To use another example, there is also the “should” of etiquette: what a person should do or not do to follow the social norms. For example, a person should not go without pants at a formal wedding, even to express his opposition to the tyranny of pants.
Returning to morality, it is reasonable to weigh the harm the expression generates against the right of free expression. Obviously enough, there is not an exact formula for calculating the worst thing a person should express, and this will vary according to the circumstances. For example, the worst thing one should express to a young child is different from the worst thing one should express to a jaded adult. In terms of the harms, these would include such harm as offending the person, scaring them, insulting them, and so on for harm that can be inflicted by mere expression.
While people do not have a right not to be offended, people do have a right not to be unjustly harmed by other people. To use an obvious example, men should not catcall women who do not want to be subject to this verbal harassment. This sort of behavior certainly offends, upsets and even scares many women and the right to free expression does not give men a moral pass that exempts them from what they should or should not do.
To use another example, people should not intentionally and willfully insult deeply held beliefs simply for the sake of insulting or provoking the person. While the person does have the right to mock the belief of another, the right of expression is not a moral free pass to be abusive.
As a final example, people should not engage in trolling. While a person does have the right to express his views to troll others, this is wrong. Trolling is, by definition, done with malice and contributes nothing of value to the conversation. As such, it should not be done.
While I have claimed that people should not unjustly harm others by expressing themselves, I have not made any claims about whether people should be allowed to express themselves in these ways. It is to this that I now turn.
If the principle of harm is a reasonable principle (which can be debated), then a plausible approach would be to use it to sketch out some boundaries. The first rough boundary was just discussed: this is the boundary between what people should express and what people should (morally) not. The second rough boundary begins at the point where other people should be allowed to prevent a person from expressing themselves and ends just before the point at which the state has the moral right to use its coercive power to prevent such expression.
This area is the domain of interactions between people that does not fall under the authority of the state yet still allows for preventing people from expressing their views. For example, people can be justly prevented from expressing their views in the workplace without the state being involved. To use a specific example, the administrators of my university have the right to prevent me from expressing certain things in my classes, although the specific restrictions can be debated. To use another example, a group of friends would have the right, among themselves, to ban someone from their group for saying racist, mean and spiteful things. As a final example, a blog administrator would have the right to ban a troll from her site.
The third boundary is the point at which the state can justly use its coercive power to prevent a person from engaging in expression. As with the other boundaries, this should be set (roughly) by the degree of harm that the expression would cause others. There are many easy and obvious examples where the state would be right to limit expression: threats of murder, damaging slander, incitements to violence against the innocent, and similar such unquestionably harmful expressions.
Matters do, of course, can get complicated quickly. Consider, for example, a person who does not call for the murder of Democrats but tweets his approval when they are assassinated. While this seems to be something a person should not do, it is not clear that it crosses the boundary that would allow the state to justly prevent the person from expressing this view. If the approval does not create sufficient harm, then it would seem to not warrant coercive action by the state.
As another example, consider the expression of racist views via social media. While people should not say such things, as long as they do not engage in actual threats, then it would seem that the state does not have the right to silence the person. This is because the expression of racist views (without threats) would not seem to generate enough harm to warrant state coercion. Naturally, it could justify action on the part of the person’s employer, friends and associates: they might be fired and shunned. Or might have been in the before time.
As a third example, consider a person who mocks the dominant or even official religion of a state. While the rulers of such states usually think they have the right to silence such an infidel, it is not clear that this would create enough unjust harm to warrant silencing the person. Being an American, I think that it would not, but I believe in both freedom of religion and the freedom to mock religion. There is, of course, the matter of the concern that such mockery could provoke others to harm the mocker, thus warranting the state to stop the person for their own protection. However, the fact that people will act wrongly in response to expressions would not seem to warrant coercing the person into silence.
In general, I favor erring on the side of freedom: unless the state can show that silencing expression is needed to prevent real and unjust harm, the state does not have the moral right to silence expression.
I have merely sketched out a general outline of this matter and have presented three rough boundaries about what people should say and what they should be allowed to say. Much more work would be needed to develop a full and proper account.