Level of de-centralization Bluesky has is somewhere between the old Twitter and Ethereum, neither of which have strong resistance against central decision making.
The problem discussed here is that Mastodon is not simply de-centralized, but its superstructure upholds a segregation policy and loves to ostracize admins based on, ahem, preferences. This in turn encourage admins to join a virtue signaling zeitgeist, and towards assuming more divisive and dismissive stances, out of fear. As a second order effect, regular non-admin users and their ability to communicate would be not only at whim of the server owner but also that of the inner group cast towards the admin.
Bluesky doesn't have this type of problem, precisely because it's not too decentralized. Either you individually get banned or not, based on levels of value alignment between you and the corpo outsourced moderators. There are also blocklist feature as well as third party voluntarily applicable moderation framework in Bluesky, but personally I can't imagine majority of users using it, or dividing the network into fragmented subgroups, and are non-factors in the grand scheme of things.
(By the way, I sometimes wonder how moderator value alignment is going to inevitably drift over time; as I understand it, social media content moderation is partially automated and exploitatively outsourced to workers from low income regions, such as sub-Saharan Africa. This phenomenon is almost exclusively discussed in context of human rights and fair worker treatment, but I think this also means a lot of people with minimal prior exposure to media, let alone the anaerobic layer of the Internet, are being trained to develop preferences on such content and especially the more flaggable yet less hateful and flaggable-but-less-flag-deserving content. i.e. stimulative but not blood and gore. If anyone is reading down to this line, you know what I mean.)
Subscribing to a labeller is as easy as following any other account. I use several 3rd party moderation services. The bar to adoption is much lower than I think you anticipate
The problem discussed here is that Mastodon is not simply de-centralized, but its superstructure upholds a segregation policy and loves to ostracize admins based on, ahem, preferences. This in turn encourage admins to join a virtue signaling zeitgeist, and towards assuming more divisive and dismissive stances, out of fear. As a second order effect, regular non-admin users and their ability to communicate would be not only at whim of the server owner but also that of the inner group cast towards the admin.
Bluesky doesn't have this type of problem, precisely because it's not too decentralized. Either you individually get banned or not, based on levels of value alignment between you and the corpo outsourced moderators. There are also blocklist feature as well as third party voluntarily applicable moderation framework in Bluesky, but personally I can't imagine majority of users using it, or dividing the network into fragmented subgroups, and are non-factors in the grand scheme of things.
(By the way, I sometimes wonder how moderator value alignment is going to inevitably drift over time; as I understand it, social media content moderation is partially automated and exploitatively outsourced to workers from low income regions, such as sub-Saharan Africa. This phenomenon is almost exclusively discussed in context of human rights and fair worker treatment, but I think this also means a lot of people with minimal prior exposure to media, let alone the anaerobic layer of the Internet, are being trained to develop preferences on such content and especially the more flaggable yet less hateful and flaggable-but-less-flag-deserving content. i.e. stimulative but not blood and gore. If anyone is reading down to this line, you know what I mean.)