Just made it in time to the next session at AoIR 2023, which starts with a paper by Yiran Duan on deleted tweets. The focus here is on deleted tweets in the #BlackLivesMatter hashtag during three key periods (including Black History Month in 2020 and 2021 and a police brutality trial in 2021). Some 37% of these have become unavailable in the last couple of years. This is comparable to the 33% of deleted Brexit tweets (but that deletion rate occurred in four years), and much more than 19% generic tweets deleted in four years that other studies have observed.
Work on this was spurred by a project that sought to code tweets in a specific dataset, which during coding found a considerable number of deleted tweets. Is the deletion rate of BLM tweets unusual, then? Of the total number of tweets in these datasets, some 21% were subsequently deleted, while some 16% showed as being from now private or suspended accounts; but we cannot know when or why a user deleted their tweets, or why accounts went private or were suspended. Previous studies have pointed to regret as a key factor, however.