After thinking about it for a while, I have some perhaps relevant experience to offer on being a Moderator.
I've worked in IEEE Standards working groups for decades, and my experience is that while debate (and beer) is necessary, suppressing fights is essential to make progress.
Agitated people tend to lose sight of the fact that it's the Working Group (WG) as a whole who must be convinced, not one's immediate opponent. And that the WG tends to interpret resort to ad hominem attacks as a public admission that one has no better argument. Then the WG votes on the issue.
In legislatures and Standards Working Groups, a common motive is delay, the hope being to prevent an undesired law or standard from being issued, most commonly when a party senses that they are losing the debate.
Typical Internet trolls intend to cause anger and disruption instead, the motive being to get attention, but are otherwise random. One defeats this kind of troll by simply not responding - Do Not Feed The Troll!
In such as PM, the motive is more often simply to silence people who hold opposing political opinions. This is usually seen as endless ad hominem attacks, often combined with mobbing. In such cases, the afflicted thread becomes polluted and thus useless, and most people silently flee.
A major characteristic of these attacks is that the original question/point is never directly addressed – it's all attack and no debate. One often wonders if the attackers had understood (or even read) the position or argument they are attacking.
However, when one talks of only allowing Truth, the issue of who shall be the Arbiter of Truth immediately surfaces, with all the usual dangers, and the Arbiter will always be accused of bias, sometimes with good reason.
One common kind of Truth claim is statements to the effect that one should "Trust the Science" in one form or another. While one may very well trust Science (a process), it does not follow that one therefore must trust any scientist or scientific organization, however impressive.
And Science is full of Truths that were eventually overturned, replaced by a better Truth. A classic example being when after 200 years, Newtonian physics gave way to Relativity and Quantum Mechanics in the first quarter of the 20th Century.
So no matter how loud the assertions, the true answer may not yet be known, and the losing side will never accept the result, and things will become ever louder and angrier.
How does one deal with such things, and get to the bottom of the matter at hand?
A battle-tested strategy for such things is to enforce Decorum, as laid out in Roberts Rules of Order. This is widely done in legislatures and standards groups and the like, but Robert's can be pretty complicated and arcane. A simplified and more suitable form for such as PM are the rules of an Oxford debating society, which rules are far less complicated and formal than Robert's.
In the PM Community, this strategy would take the form of suppressing ad hominem attacks of all sorts (including passing jabs), regardless of the subject, regardless of who did what to who before. Even-handed and sustained application of the rule against ad hominem attacks often suffices. If not, also enforce relevance.