An apology from Ganesh

Update 2: Hello, I want to address three points.

  1. “There’s no way it was just the Drive links.” It wasn’t. There were around 15 million books and 80,000 Drive links, so clearly not. The bulk of the data (and, as I found after investigating, the drafts) came from two places: nyuuzyou’s AO3 dataset on Hugging Face, and FicHub’s 1 TB storage box dump from their Discord server. You can find links to both independently. Together these sources held about 29 million stories across several sites (Wattpad, Fimfiction and FanFiction.net, mostly from the FicHub dump, which you can verify by looking for https://u521841-sub1.your-storagebox.de/ on the FicHub Discord), but I only downloaded the AO3 content.
  2. “They can’t be Indian, they place their commas the Western way” or “They can’t be Indian and 17, when I was 17 I had to write so many exams!” Thankfully, I’m done with all my exams, and I was never forced into the mismanaged rat race that is India’s public exams: neither NEET nor JEE, both of which have a death toll. I have only my parents to thank for that.
  3. About the OTW. Their legal team has filed a URS complaint against this domain to protect their trademark, and the domain will be taken down in 14 days if I don’t respond. I concede, and I’ll let it be taken down without any fuss. Though, a URS is based on trademark, so it can’t be enforced against other domains that, say, don’t have “ao3” in their name.

Update: A list of all the links and sources are available here, there are over 70,000 of them, and many of them include drafts - which I was not cognizant of.

Additionally, I’m unaffiliated to other sites like ao3.ac, blk-archive, fanfic.party, ficatlas, etc. I’m done with this site - however if I can be of help to the community, please shoot me an email at legal@ao3.life.

In the time this site has been up, I’ve facilitated over 2,000 takedown requests. It was never my intention to hurt so many authors - and I’m truly sorry to you all. There’s no way to show my sorry better than taking this site down.

Why did I charge for AI Slop audiobooks? I doubt anyone knows how expensive it is to run a server like this, especially with the amount of usage this site got. It had 126,592 unique users within 24 hours of going up, and had terabytes of bandwidth in usage. In an attempt to keep the site up, I thought audiobooks were a good option. I sold like 11 audiobooks, if it’s any consolation.

It was never and is not my intention to hurt authors, and I hope every subequent action I’ve taken is representative of that - From an attempt at revenue sharing, to responding to takedown requests as fast as I could, to taking down the site as a whole.

The data powering this site was all from public scrapes, I have no affiliation to AO3, OTW, or anyone that breached AO3.

There was a lot of love, and a lot of hate. On Reddit and Twitter people were really mean, and there was a lot of love on other channels. I’m grateful to everybody. I had a survey on the site for whether this site made you happy or not - It was 49% happy (3,184 yes) and 47% sad (3,057 no), which also made me realize that I wasn’t doing the right thing.

Yours truly,
a 17 year old from India that thought they were doing something good,
Ganesh

https://www.reddit.com/r/AO3/comments/1xm2tpp/situation_ao3life/?sort=new <- I’m monitoring comments on this subreddit and can answer any questions, and via email at legal@ao3.life

FAQ

  1. How did my drafts end up here? A major source of work was sourced from public drives being shared around twitter, and google docs links - that’s likely how.
  2. How do you have my fic from 10 years ago? There’s this arcane dataset from years ago that was loaded into ao3.life, that’s where like all the older stories came from

In response to reddit comments

Technical breakdown

  1. A swarm of AI Agents controlling 40 virtual machines were scraping the internet for all drives, scrapes, and sources.
  2. These various formats were normalized into a standard EPUB format where possible and packed into archives of 200-300 EPUBs at once and uploaded to an S3 server (like spritesheets!) based on their ID - this allows for downloads using HTTP Range requests, and was due to me hitting upload ratelimits on S3.
  3. Work names, tags, summaries, etc. were extracted from the EPUBs and stored on a large Meilisearch instance to facilitate downloads
  4. To respond to takedown requests quickly, I allocated the task to a small part of the swarm to identifying and removing the offending rows from Meilisearch, S3, and the site. Usernames, IDs, and links were extracted heuristically (regex) for the most part, and any ambiguous requests were sent to the swarm.
  5. The site itself runs on SvelteKit and Komodo for deployments, with a blue-green setup that the reverse proxy (nginx) flipped between. When usage skyrocketed, people started seeing 'the site is updating' which was nginx flipping between blue/green when one went down.

From a technical standpoint, this was really fun, I did it all by myself and the time/urgency made me feel really cool. It's not everyday you have everyone talking about the site you made, but I'll make sure it's for something praiseworthy in the future. I'm sorry for hurting the authors, and I'm attempting to make amends to the best of my ability.