It is easier, but it's still possible to solve the UX issues with a decentralized service, in my opinion. What I think is the main issue is that these decentralized services are made by programmers with little regard or intuition for UX, and there is also a lack of funding to work on UX problems.

I agree the UX challenge is much more challenging for decentralised services. I don't know enough about Bluesky to really comment on whether it is centralised or not.

Regardless, I think there's another thing that helped Bluesky: VC capital. In particular, to hire people to work on UX. It's a bit of a pet-peeve of mine, but I find it strange that designer don't contribute more to projects like Mastodon, which definitely need it. Even from the selfish angle of building a portfolio, helping solve Mastodon's UX challenges is much more impressive and realistic, than doing the millionth redesign of Gmail that will never get implemented.

People are running every element of the Bluesky / AtProtocol stack independently. Bluesky could disappear and it would continue to function as is (albeit with lots of Bluesky data lost).

PDS's to hold users data, relays/firehoses to aggregate & forward traffic, AppViews to create composite views of likes, replies, etc, resolvers to lookup DIDs, clients to access the network. Each of these has independent implementations. BlueSky is already decentralized & already has viable credible exit. It's not decentralized, and indeed the scalability & accessibility of having firehose consumers has the greatest scale out decentralization characteristics we've seen anywhere short of BitTorrent.

Arguable indeed: plenty of email clients have great UI. Both RSS readers I used are better than Facebook.