First and foremost, there's just not a lot of good research on this specific question. The idea that QTs are a sluice of harassment is kind of an intuitive canard at this point--and with good reason. Getting a hostile QT *is* an express maglev to Paintown.

But in terms of establishing how frequently quote tweets lead there, there's only a bit of data on the ground. It's often a small part of research into other, larger issues.

Let's first consider "Echo chambers revisited: The (overwhelming) sharing of in-group politicians, pundits and media on Twitter" by Magdalena Wojcieszak and colleagues--Andreu Casas, Xudong Yu, Jonathan Nagler, and Joshua A. Tucker.

In short, it's what it says on the tin. They studied how Twitter users interact with posts by political elites on the platform to try and put some empirical handles on whether there are indeed "echo chambers" on the platform (spoiler: yes.)

Here's the relevant bit on quote tweeting, drawn from a large sample. Classifying how ordinary users responded to liberal and conservative elites' posts (journos, politicians, etc.).

If you suspected conservatives live in a tighter echo chamber, congrats, here's some data for you. But liberals' isn't that much larger.

And here we see that most QTs when directed at *out-group* members are negative. Overhwhelmingly so.

This may confirm our intuitions that QTs are often negative; Wojcieszak also argues that the ingroup sharing helps reinforce a certain amount of ideological conformity, even when positive.

But bear in mind the limitations. This is how us ordinary slobs interact with the powerful. Some might argue it's *good* that we give them a reality check, say.

This is the problem with this kind of data. Determining what it means is hard.

Taking a broader view, the evidence we have is mixed. Alice Marwick's "Morally Motivated Networked Harassment as Normative Reinforcement" interviews numerous harassment victims and moderators; one cites quote-tweeting as an abuse vector.

In "Platformed Antisemitism on Twitter," Martin Reidl, Katie Joseff, and Samuel Wolley find evidence that QTs both amplified antisemitic content *and* recontextualised and neutered it.

But the truth is that this is always going to be hard to quantify. In doing this sort of content analysis on large data sets we're always in "is it a particle or is it a wave?" territory as we try to operationalise our variables--making them into things that can be counted.

The question of "what does this mean?" or "what should we value?" invariably gets into philosophical territory.

Which brings me to my next point--

There is every possibility that if we were to count up every QT and were able to objectively label them as harassment or neutral or positive in tone, we might find harassing QTs outnumbered. But then we get into quality over quantity.

What if the abusive QTs just matter more? What if they loom larger in the public consciousness? What if they generate more engagement? What if that engagement is itself largely abusive? (And this is the heart of the problem, by the way.)

The QT itself matters. But what *really* gets us is the second-order effects, which can be hard to accurately measure. This is the networked harassment Marwick describes--and she brilliantly covers how moral motivations often drive abuse--or the harassment campaigns *I've* studied. A quote tweet may flood with abuse in its comments, and it may also *inspire* more monadic abuse being hurled from all sides.

That's tough to measure, but it often overwhelms positive use cases in scale.

That doesn't mean the positive use cases don't exist, mind you. Nor that they're not important. *I* miss quote-tweeting a lot because of the way you could easily 'yes-and' someone worth amplifying and bring a constructive conversation to your followers.

Or as Reidl and colleagues observed, the ability to recontextualise hostile content and provide a robust counternarrative.

The question to answer is: is this worth all the downsides? And numbers can't answer that for you. Only you can.

Addendum: another qualitative point to consider is the psychological impact of certain kinds of QTs.

So, as a trans woman I find myself following a lot of other community members who frequently make a habit of QT-dunking on some bigot (indeed, I was just kvetching about this a little while ago on here). Recontextualisation is happening, an epistemic battle is being waged--and perhaps even won.

But constantly seeing the words of people who hate you? Not exactly ideal.

And that very issue kind of explodes the negative/positive QT binary. Because even a constructive or resistive use can have negative second-order effects. The panopticon of doomscrolling has many sources, after all.

So that's something else to consider, and something else that's harder to measure (a methodology using interviews would be a better approach here, for example).