I’m not sure that your reply is directly related to my comment. The full sentence I quoted is “Under these definitions, Bluesky and ATProto are not meaningfully decentralized, and are not federated either.” by Christine Lemmer-Webber, but Daniel Holmgren talked more directly about “decentralised distribution of data”.

Because of what I quoted, I don’t think that “Bluesky” or “ATProto” are decentralized or federated, so it’s extremely unlikely that I’ll interact with them anytime soon. The particular reason that they are not decentralized or federated is not really interesting to me.

To get specific: it is a significant issue for me if “everyone can access the data but before it reaches the end users it goes through centralised applications”. A “centralised application” is able to restrict my ability to contact other people, whereas with a federated and/or decentralized/distributed system, it’s more likely that I will be able to contact someone that I want to communicate with. For comparison, consider how people would feel if using the United States Postal Service meant that all physical mail had to pass through the District of Columbia or if sending an email message required interacting with BBN-TENEXA just because that was the first machine to be capable of sending networked electronic mail. In the ideal case, the recipient of a message I send would not have to coordinate with me at all before they receive the message: “The first use of network email announced its own existence.”