Remember eleven months ago when the Tumblr CEO said that they would be supporting ActivityPub real soon now and then never mentioned it again? Sounds like they got cold feet. ("Wait, how does being open and interoperable make us money??") https://mastodon.social/@_jv_/110692572741573511
I think the key is right there at the end: "costing"
#ActivityPub is not a lightweight protocol or system. We've seen plenty of instance operators voicing concerns and surprise about how resource intensive it is at scale, how much it costs to operate as usage increases.
I wouldn't be surprised if Tumblr (and many other sites) looked into the costs of running giant instances and recoiled at the investment it would require to make it a reality.
And this is something we need to call out the platform over. We should vocally criticize the platform for making design choices that are so unscalable.
It's an ActivityPub design issue.
AP specifies everything from a bloated signaling protocol through behaviors that suggest exponential processing and traffic as functions of instances involved in a post.
It doesn't matter how well an implementation is written; it will be subject to that design.
@volkris @tezoatlipoca Bluesky: "Hold my beer" https://urbanists.social/@sam/110339902538138997
@volkris @jwz "Call out the platform" - Mastodon? But is this an ActivityPub *interface* design issue, or is this a *Mastodon* implementation issue? Asking, cause I don't know. Just because current APub implementing platforms like Pleroma, Mastodon, Misskey etc. aren't designed internally to scale well doesn't necessarily mean the interface the platforms implement isn't worthwhile? or that the platform dev team of a large player like Tumblr couldn't come up with a more efficient implementation?