criticism of bluesky as I come to understand it more deeply
The primary reason #bluesky will never be decentralized is because the resource requirements are way too high. Yes, you can host a #PDS (personal data server) if you like, and that will provide you control of your own personal data. However, you're still dependant on someone else to host a relay, in order for your PDS to become part of the network. Hosting a relay seems to need upwards of 3 TB of SSD storage, and a ton of memory. And once you get *that* done, you need to host an app view, in order to make use of all of the data on the relay. While that's not quite ready for primetime yet, when it is, it ain't gonna be cheap. Based on what I can understand, an app view needs to index *all* data on the bluesky network, in realtime. This is the kind of architecture that only our VC funded #capitalist big tech masters could actually afford to deploy. So if, when you say "decentralized", you mean that Google, Microsoft, and Apple could each afford to run a shard of the network, sure it's decentralized. But if you mean that Joe and her homelab could run any useful portion of the network, it is in no way decentralized, and never will be. Even if compute prices come down to the point where small organizations could afford to deploy this sort of thing, you've still got the logistical nightmare that the relay hosts *all* of the data. From *everyone*. NAZI's, pedophiles, if you run a relay, all of that data is going to be passing through your server. Content labels and blocklists mean you might never see it, but if you run a bluesky relay, it still *exists*.
