IMO, MrHands' most important point was that engaging with other's opinions is important for changing people's minds.
As MrHands said himself in the op, this style of impersonal internet debate is unlikely to change people's minds. Often, therefore, it's not important because it has about the same chance to work as Wales beating England by four goals today. (Praying it happens though.) Does that mean trying to change people's minds this way is pointless? Well, it depends.
Different people have different persuasion odds. To help understand that, I want to set up "deep wrong" versus "surface wrong". Everyone has room for improvement in their beliefs, factual understandings, opinions, and values.
(If your response is "But opinions are subjective, how could one be better?" My response: "Is 'murder is generally bad' a better opinion than 'murder is generally good'"?) However, not all beliefs are equally imperfect. "I love fascism" is deeply wrong, and it tells more about a person, bad things, than "I think 15*15 equals 255," which like, oh well, no big deal.
The 15*15=255 person isn't that deeply wrong, and they probably share a lot of correct beliefs with me, giving us existent and good common ground. They probably think mathematics follows certain rules, trust calculators, you get it. I can enter 15*15 into Google, they will see 225, and they'll change their beliefs to be correct. In general, people who are more surface wrong are easier to persuade than people who are more deeply wrong.
From a lot of experience talking and watching politics on the internet (and real life some too), vocal Trump supporters tend to be very deeply wrong. They frequently prioritize their feelings of "winning" over actually becoming more correct, frequently box people into "evil villains" or "irrelevant other who deserves no attention" for very shallow reasons, and are frequently hostile to the processes of logic and science, among other things. The odds to persuade them, and benefit to talking with them in good faith, are very low on average.
I like disagreement and debate within a "less wrong range" more, as it's more likely to have people learn stuff, and it doesn't expose us all (especially the minorities threatened by Trump and his party) to deeply wrong and embarrassingly bad "argumentation." For example, steeldragon's reference to "brainwashing of the youth" is a frequent anti-LGBT dogwhistle,
though, almost by definition, dogwhistles are not quite always used to hide the evil beliefs they frequently cover, so there's your disclaimer.
I would not necessarily support steeldragon's posts being deleted. However, I also see no need to actively draw more awful "discussion" from them with a good-faith invite like MrHands did. I
especially don't see the need to draw them in out of desire to avoid a laughable "far-left circlejerk" bogeyman.
Edit:
This video, while the tail end of a six part series, talks about challenges and disincentives to try and persuade people who are deeply wrong. I don't agree with all of it (at least with reference to the smogon dot com politics threads), but it has lots of good stuff, and maybe you see it differently. (just, like, replace feminism with not-republicanism and gamergate with republicanism) (great series btw)
Edit2: I HAD written about some important takeaways from the video, but it didnt save apparently. And Wales didn't win :(
Here's what I remember. When MrHands mentioned/implied that talking with deeply wrong people can benefit onlookers, he was totally right. However, as the video says, that benefit doesn't come from giving the deeply wrong arguments or people respect that is undeserved. It doesn't come from assuming good faith and giving friendly invites to spew more junk. It comes from systematically deconstructing the deeply wrong statements for the explicit benefits of onlookers.