Are collaboration bots allowed on Khan Academy?

That sounds really interesting! I'm thinking this should be fine as long as the bot isn't making an excess number of posts, or creating posts that are unusually long.

The main issue is that threads with lots of comments or posts that are very long (or short) tend to be auto-flagged by our AI filter, so these types of posts may clog up our moderation queues which isn't something we want to happen. There's also a chance these posts could end up being deleted, which may prevent your bot from working correctly.

If the bot automatically deletes its posts after a certain period of time or a certain number of posts, I think that would prevent this, so it should be okay. You may also want to add a check that makes sure the bot splits up posts above a certain length.

Thanks for asking about this! I'd love to see an example of this in action : )

That's good to know! Would it work to prepend each command with a phrase or something to give it a higher score? The commands are pretty short (based on bash), so I think they might get a low-quality score.

As for an example, I currently have it running on the Tips & Thanks of this program: https://www.khanacademy.org/computer-programming/bot-endpoint/5637429470052352?qa_expand_key=ag5zfmtoYW4tYWNhZGVteXJACxIIVXNlckRhdGEiHWthaWRfNzUwMzg4NTc4NDk2NDQzNjMwMzUzMTc3DAsSCEZlZWRiYWNrGICAnay3iY8JDA

It doesn't look like there's any comments in that thread, so I don't whether or not the commands are being auto-hidden. If you notice that it isn't working properly, adding a longer phrase to the start of the comment should work.

Oops, that's a fresh thread. I just ran a command in the thread, and it showed up even well logged out.

One additional question, in regards to other people creating inappropriate programs through the bot, what would be a reasonable enforcement strategy? For example, if a project was shadowbanned by the moderators it could be automatically deleted. However, I'm not sure if an inappropriate project would automatically trigger a complete profile ban, leading to lost work on the other programs.

Another option would be to check a flags:votes ratio being to high, and based on that notify the authors and delete the program.

I think the best thing to do is run a word filter on all newly created and updated programs, so any programs with inappropriate words are not created/saved in the first place.

No matter how extensive the list, I'm sure users will always find a way to include variations of inappropriate words or drawings, so checking the flag count on programs would be a good back-up strategy to have as well.

Thank you for the quick reply -- I'll implement that now :)

Based on your experience, would you say anything above 1:10 flags:votes would be suspicious, or should I target something lower/higher? As well as a minimum of 2 flags before activating.

I think anything above 1:10 seems pretty reasonable, as long as there's at least 2 flags.