a #nostr #relay could be programmed to analyze all the publicly available data, sort it and present a curated feed, that is personalized to the npub.
such a sorting algo would NOT be centralized, because users could run their own relay. the sorting would also be done only to serve the needs of the user (not the relay). I refer to this kind of relay as a #treelay because its based on https://github.com/baumbit/peercuration?tab=readme-ov-file#peercuration and the end-product is a content tree graph.