The problem as described

I'm writing this down just to have it as a record in time.

Twitter (X) is changing its block function so that your tweets will still show up in the feed of a person that you've blocked.

I've seen several people openly wonder why this change is being made.

My guess for why this is happening

My person suspicion is that the change is actually about simplifying the architecture of blocking and feed construction.

Specifically, after this change, it will become possible to store a user's blocklist as a list of usernames that is stored on each device (synchronised via the cloud), with blocking happening on-device.

This frees up the algorithm that populates the content of the newsfeed, from having to consider other users blocklists when constructing the feed for a given user. At present, constructing user A's feed requires checking if any other users have blocked user A, which presumably is costly/annoying/adds complexity when constructing each user's feed.

An interesting thing to note is that old Twitter was much more willing to vacuum up data about individual users to alter the vibes/equilibrium on the Twitter feed. So DMs, metadata about DMs etc. could all be used in feed construction or in ranking tweets.

The new Elon architecture seems to be simpler, with a corresponding sloppier feed.

Comparisons to Mastodon

I would note that this new blocking method is a lot more like how Mastodon implements blocks. Note that Mastodon has to implement blocks in this way since it's not possible for someone on a given server to control whether or not an individual on another server is provided the feed of their Toots. I suppose Mastodon does allow one server to refuse to federate with another, but that is done at the server rather than personal level.

Not sure how BlueSky implements blocking, but that would also be interesting.