13 ms·
> Is it similar quality or higher-quality now? I also read the paper. Depending on what metric you choose, it is either similar quality (based on revert number
by nl 6y ago
> Is it similar quality or higher-quality now?
I also read the paper. Depending on what metric you choose, it is either similar quality (based on revert numbers) or higher-quality (based on PTR - basically the characters that survive subsequent edits).
> While the article and paper are all positive... I'm not sure that's what this unambiguously shows.
You are absolutely correct!
The paper actually shows that Tor edits were much worse than any other group ("Overall, 41.12% of Tor edits on article pages are reverted, while only30.3% of IP edits, 35.2% of First-time edits, and 5.5% of Registered edits are reverted.") before the ban was implemented.
After the 2013 ban they "consider the small number of edits made in this later period. Using the methods described above, we computed the reversion rate and the PTRs for the population of 536 edits made after 2013 along with the same number of time-matched edits from other groups as described above." - and THIS dataset is what the headline comes from!
So this positive conclusion is based on these 536 edits, not the much bigger dataset! Additionally, they filter out edits to non-article pages where "this analysis provides evidence that Tor was used to engage in edit warring in violation of Wikipedia policy."
It's a pity, but this paper doesn't really support the idea that Tor users are good Wikipedia editors.