John Stuart Mill is considered one of the most famous advocates of free speech. He was an English philosopher, political economist and civil cervant.1
In his Essay, On Liberty, he argues that one should be able to freely discuss any doctrine, no matter how immoral.
"If the arguments of the present chapter are of any validity, there ought to exist the fullest liberty of professing and discussing, as a matter of ethical conviction, any doctrine, however immoral it may be considered." 2
This is Mill's starting point—unlimited free speech. Dissident opinions are invaluable for intelligent discourse and learning, and Mill goes as far as saying that limitations on these freedoms are morally wrong.
The Harm Principle
Mill does place one limitation on freedom of speech, and that is the Harm Principle. A person's right to free speech can be overruled by a society, if it is done to "prevent harm to others" 3.
The definition of harm has been a subject of great debate. Even truthful information can sometimes be shocking, and it's difficult to say what constitutes as harm.
Sticks and stones might break bones, but what about hurt feelings?
Harm in the first instance
Mill's examples of harm are direct and in the first instance. He specifically gives an example of claiming that corn dealers starve the poor, and argues that it is acceptable to print this in a newspaper, even though it might cause financial harm indirectly.
Yelling out the same words (or passing them on as notes) in front of an angry mob, causing them to physically harm the corn dealers, would be harming in the first instance.
Internet in the first instance
Information on the internet can reach everyone. Given enough time and popularity, electronic words can reach people in a multitude of circumstances.
Let's say that a given piece of offensive or potentially harmful online information, like that of the corn dealers, will have enough popularity and time to be read in all conceivable situations. It will cause some harm, for sure. In fact, it is hard to find a single opinion that would not cause any harm in this, admittedly exaggerated, scenario.
Luckily, pretty much all harm that can be generated by the internet is not instigated in the first instance. Written word, by Mill's own example, did not cause harm directly, and was perfectly fine.
Are there internet services, where the internet might be used in a way that causes harm in the first instance?
Probably, but I can't think of any. The internet is a medium of communication, with unlimited possibilities for content and surrounding conditions, so it might be possible to set up a live-stream video in front of an angry mob and instigate a crime.
But today's blog posts and twitter updates lack the required directness, and can not by their very nature instigate such violent crimes directly.