It’s been 3 months since I first ran my scam alert bot. Since then, AO3 has implemented some additional measures to stop scammers, such as sending more data to Akismet to help the spam filter function better (AO3-7238, AO3-7262, AO3-7271), additional rate limits (AO3-7264), and patching some overlooked scam vectors (AO3-7273). I’ve turned it on again for 24 hours on March 20th to measure the effectiveness of these efforts, and we will compare the stats against the ones measured on December 24th, 2025.
The bot scraped a total of 162,228 comments. On Dec 24th, the bot scraped 164,135 comments.
Of which, 3.57% are guest comments, almost half of 6.75% in December. This could be attributed to the stricter spam filter.
The ratio of top-level to replies has stayed the same.
Only 0.468% were flagged as spam comments by the bot, significantly down from 3.51%.
The difference is much more apparant in absolute terms. We can also see a significant decline in non-scam guest comments, perhaps as collateral damage of the stricter spam filter.
There are no longer any significant patterns to analyse.
This is a very significant improvement in such a short timespan, considering this scam has proliferated on all major fanfic sites for 2 years now. I don’t know if other sites are also seeing a reduction of scammers, but this certainly marks the beginning of the end of the crisis on AO3.
Acknowledgements
Many thanks to the OTW System Committee for allowing this bot to operate.
