7
The Kbin Experience. The Best Worst Way to Interact with the Fediverse.
(roundcatsblog.wordpress.com)
/kbin-related stuff. Unofficial, not moderated by devs (yet). Official ones: /m/kbinMeta /m/kbinDesign **All official /kbin magazines in one collection**
Very true. It's a lot more stable than it was even a couple months back - I used to upgrade my instance and run into constant problems, and now I can happily git pull without destroying everything.
From a user perspective, a lot of the issues you encountered early on were due to a couple factors - kbin.social was dealing with more traffic than it could handle, and the developers of lemmy added a sneaky thing that would specifically block kbin user agents from being able to federate out to lemmy instances, leading to constant error logs and issues.
I would say that, aside from platform stability, the biggest looming threat on the horizon is spam. Think about email (the original federated message system) - nobody even thought about the possibility of spam when developing email, spam exploded in the 90s, and currently spam control is managed by whichever email platform you're on rather than by the protocol itself, as well as a sort of trust system where newly registered domains are more likely to be seen as spammers. The Fediverse needs to take lessons from email and start implementing the same sort of controls before the issue becomes unmanageable.
Do you happen to know why they did it? Was it that kbin was causing them technical problems somehow and they chose to block it until they are resolved or was it just pettiness?
IIRC the official reason was that some automated anti-spam code accidentally caught the kbin user agent and mistakenly added it to a block list, and the lemmy.ml admins were busy and didn't see it for over a week - but once one of them noticed it was promptly fixed.
Also, I recall this being specific to lemmy.ml - other instances run by other admins like lemmy.world and lemmy.ca weren't affected.
Unfortunately I don't know. It could very well have been that their own instances were being flooded by traffic from kbin.social and this was their solution. However, to my understanding this was added to the source code itself so all lemmy instances upgrading suddenly become an issue for kbin instances. I had to change the user agent in my own kbin instance to stop the flood of errors as it was absolutely messing up the.coolest.zone and I basically couldn't troubleshoot any legitimate failed messages.
Whatever the case is, I'm very glad it's resolved now.