Wait, am I just completely misreading this or is this contradicting itself? > the research team found that Tor users made similar quality edits to those of IP

Wait, am I just completely misreading this or is this contradicting itself? > the research team found that Tor users made similar quality edits to those of IP editors [...] and first-time editors. The paper notes that Tor users, on average, contributed higher-quality changes to articles than non-logged-in IP editors. Is it similar quality or higher-quality now? The text also appears (word for word) on the linked website at nyu.edu. Reading the original paper, guess what I found? > Using hand-coded data and a machine-learning classifier, we estimated that edits from Tor users are of similar quality to those by IP editors and First-time editors. We estimated that Tor users make more higher quality contributions than other IP editors, on average, as measured by PTRs. Almost the same wording and contradiction again. There is a subtle change, namely that the "of similar quality" judgement is a result of hand- and machine-classifying, and "more higher quality contributions" is the judgement of a metric called PTR* . There might also be a difference between "similar quality edits" and "more high quality edits" (e.g. Tor users might do more crap edits and more great edits by one metric but simply be about average by another metric), but I'm not sure if that's just random variation in phrasing or intentional. * PTRs are "persistent token revisions". I don't find it very succinctly/adequately explained at first use, but probably if you read the whole paper it makes more sense. To my understanding, it's basically just how much of the contribution was later changed (within a fixed number of subsequent edits), presuming that if it was largely left unchanged, it was probably a welcome and correct edit. While the article and paper are all positive, I'm not sure whether this might just be because of course we'd all love to hear how great Tor users are (many of us are also Tor users: we like to think of ourselves as freedom fighters, privacy advocates, etc.), but I'm not sure that's what this unambiguously shows. Perhaps it's worth the moderation effort to unban them, perhaps not. The paper does acknowledge this to an extent: "We simply cannot know if our sample of Tor edits is representative of the edits that would occur if Wikipedia did not block anonymity-seeking users." Perhaps, instead of ban vs unban, we just need another system to anonymously contribute changes, like a moderation queue, which would make it less attractive for vandalism. (On StackOverflow/StackExchange, anyone can edit without logging in and not even your IP address is shown. While moderating it, I very very rarely see trolls or spambots there. I'm not sure if that's because of some magic system I don't know about or if it's simply because a manual review filters all the garbage and there is no point in trying.)