One of the hardest parts of this research — and something that we’ve struggled with for as long as I’ve been doing research on tweets (~8 years) — is coming up with a coding scheme and “rules” for distinguishing between “mainstream” media and “alternative” media. The lines are REALLY blurry and any line you draw to distinguish one from another is easily contested. For this research, participation in the alternative narrative became the tool that I used (for the most part) to identify the domains that I focused on. If you look at the ICWSM paper, at the figure that colors the nodes by “affirm” vs “deny” or“evidence”, you can get a sense of how the “alternative media” in my graph featured here overlap with those domains that participated in the propagation of alternative narratives. Many of those domains described themselves as “alternative” or “anti-media”, and so I adopted the language they used to define themselves as opposed to applying my own (a technique of my research field). There were also sites that identified that way that did not participate in the alternative narrative and I wanted to honor that distinction in the paper— I wanted to show that there were alternative media sites in this graph that did not propagate the conspiracy theories related to mass shooting events. This is why I kept the two dimensions (media type and affirm/deny/as evidence). But I think there’s a second critique here that holds… I can agree that “alternative media” is much too broad a term to perfectly equate with this subsection of that ecosystem. What I identify and describe here is likely more of a “conspiratorial alternative media” set of domains.