Particular instances of heckling can certainly be reactionary, disproportionate, and unfair - but routinely, liberal discourse villifies heckling as problematic in general. Still, normative ethics typically rely on one of three kinds of arguments against this sort of behavior, and for the life of me I can't figure out how any of them can be maintained within the framework of modern liberalism.
A consequentialist argument would have to claim that heckling necessarily leads to bad outcomes. This is a difficult position to maintain if we are talking about first-order effects, since it's trivially easy to imagine trolling that just leads to funny outcomes, or outcomes in which the immediate good outweighs the immediate bad. For instance, when Center for American Progress President Neera Tanden recently attacked Bernie Sanders on Palestine, I pointed out that Tanden had serious conflicts of interest on this issue as revealed by recent email leaks. Practically speaking this jab probably accomplished little, but insofar as it has any moral weight whatsoever, it seems clear that the immediate value of exposing Tanden outweighs whatever discomfort the exchange made her feel.
To get around this, opponents of heckling usually starting dreaming up all kinds of second, third, and fourth order effects, and then insist that the heckler has a duty to prevent these outcomes. One common argument, for example, maintains that even if heckling Neera Tanden in this particular instance is warranted, this good act might inspire someone to commit the bad act of (say) heckling some undeserving third party.
The basic problem here is that this line of argument can obviously indict even the most benign and responsible criticism through all of the same vague mechanisms of "inspiring", "normalizing", "encouraging", and so on. The blanket condemnation of all heckling due to second order effects is too strong. You can only save legitimate criticism from its censure by introducing all kinds of considerations of probability ("is this likely to cause that?"), responsibility ("can we still blame someone for a third/fourth/fifth order effect?"), and proportion ("do all the good Nth order consequences outweigh the bad ones?") - considerations that, if we are being consistent, end up legitimizing some instances of heckling, too.
A second problem, pointed to in my consideration of proportion, is that post-first-order discourse effects are so complex, subtle, overdetermined, multivalent, prolific and unpredictable that they are basically impossible to evaluate in any kind of credible or morally compelling way. To make this sort of argument, what you would actually have to do is look at a given instance of heckling, figure out all of the Nth-order discourse effects (or at least a defensible set of them), and then weigh all of them against each other to decide whether or not it was justified. Otherwise, the argument just becomes an exercise in cherry-picking whatever remote possibilities helps make your case.
This, by the way, is the fundamental problem with a whole genre of liberal discourse theory. There are real-world situations where you can analyze second order outcomes in a way that's thorough and rigorous enough to draw compelling moral conclusions, and when we can we obviously should. But no one who has studied language, rhetoric, or sociology in any kind of serious way thinks that we can always do this with discourse. Most people intuitively get this, and center their discourse norms around the immediate consequences of what they say, only moving much beyond that in unusual circumstances. It's really only among liberal intellectual elites that one encounters these hubristic attempt to game the discourse like a butterfly trying to create hurricanes.
A virtue argument would have to maintain that heckling, in principle, reveals something bad about the heckler's character. This is what liberals are typically getting at when they suggest that heckling is an expression of personal bigotry, malice, dishonesty, and so on. Even if the heckling is somehow intellectually or pragmatically justified, one argues, it nevertheless reflects things about the heckler that we should condemn.
This, again, may very well be true in particular cases, but it's hard to see how the argument necessarily holds in principle. Once we accept argument (I) that heckling can have good outcomes, it follows trivially that heckling could simply express one's virtuous intention to do good. One can insist that this intention is necessarily misguided, but only by rejecting (I).
It's because (I) is so hard to reject (for reasons given) that the virtue argument against heckling usually leads to the assertion of ulterior motives. The heckler is accused of bad faith, of unconscious or unadmitted bigotry, of secret malevolence and sadism, etcetera. These sorts of motives are of course notoriously difficult to empirically establish even in a clinical setting, and online they usually just amount to the question-begging assertion that the heckler is evil because he has evil motives, with no attempt to ground any of this in reality.
Nevertheless, even if unvirtuous trolls exist (and they empirically do), this just means that some heckling is bad. Virtue arguments against particular instances of heckling may succeed, but once again the general argument clearly fails.
The deontological argument against heckling is paradoxically both the strongest and the weakest.
Suppose, for example, you are in a cult of personality centered around nineties pop folk singer Jewel, and take her song lyric that "In the end / only kindness matters" as some kind of divine mandate. If one accepts this, then there's probably no case that one can possibly make for heckling, particularly if we have in mind Jewel's civility-centric ideas about what it means to be kind. You can't at this point make consequentialist or virtue arguments for heckling, because the deontological prohibition rules them out absolutely.
This, it's worth adding, is the ideological basis for most in-principle opposition to heckling, even when that opposition gets rationalized with consequentialist or virtue arguments. In general, I suspect this is because most liberals are so privileged that the worst thing they typically experience is interpersonal conflict; this creates an extreme, instinctive conflict-aversion. For this reason, they're willing to sanction all kinds of injustice and suffering for others just as long as everyone around them is polite and friendly - and the easiest way to rationalize these kinds of priorities is just to insist that civility and interpersonal kindness are the most important things in the world.
This claim faces two major problems. First, no one actually buys it. Even the strongest proponents of a deontological prohibition against heckling are themselves hecklers on occasion - and when they justify it, they justify it with consequentialist or virtue arguments, not deontological arguments. (Though it would be refreshing to see a hypocritical advocate of respectability politics simply admit that it's definitionally okay for him to do it.) Outside of elite intellectual circles, most people have a more relaxed view of heckling, and see it as potentially funny and occasionally warranted; this is why (for example) an extraordinary amount of sitcom humor revolves around verbal sparring, scathing zingers, and bad / obnoxious people getting their rhetorical comeuppance.
The second major problem this faces is that in a pluralistic society people are allowed to object to the Jewel doctrine. If (say) I make a consequentialist argument for heckling Neera Tanden, and you simply reply by decreeing that heckling is Always Bad, you may very well in some real and completely legitimate theological sense be right; but it isn't something anyone should find persuasive or compelling, for obvious reasons. This is particularly true since, again, an absolute deontological prohibition of heckling is at the most a fringe view held by privileged elites, which escalates its imposition from the realm of "anti-pluralistic" into "blatantly anti-democratic".