So just my personal take, I definitely benefit a lot from Mediums recommendation system, however it might work. Probably because I'm brand new to programming, and even fields like machine learning are very fresh to me even though it seems like everyone and their mom already knows the basics. So having a system that can see that I looked up a specific topic and recommend similar articles from that domain. A lot of the time I don't know what I don't know, so having this recommendation system helps illuminate the domain a bit more for me. And the general style of medium where the posts can be a more casual overview of a topic compared to the deep dive of a whitepaper, helps me grasp concepts faster. The recommendation aspect also helps me explore domains in a more focused way compared to the random nature of social networks that just bubbles up whatever's popular. I'd be lying to myself if I said I haven't benefited from Medium.
Not trying to excuse them for what they're doing, just trying to point out from my personal experience the value they've added to me and why following personal blogs or places like reddit, which I see mentioned as alternatives, isn't a one to one replacement for me. I don't know what or who I'd need to follow if the domain is new to me.
I guess continuing on this train of thought, and not wanting to complain without putting out an idea for a solution...if something were to come in and be a viable alternative to Medium it would have to
>Allow for easy publishing (editor)
>Have reliable distribution/hosting (network)
>And have some way to explore related topics or publications (discovery)
For publication I think the community would have to come around to a standard format. Like a simplified latex, or something similar to it, with an approachable interface that even my Mom or Dad could write something up without needing to read documentation.
As for distribution, the only thing I can really think of is an ipfs style network. Similar to how torrents can provide some security in the survival of a file even if the originator decides to stop seeding (hosting). And also similar to torrenting, I can see ppl willingly giving up resources for 'the cause' if they themselves benefit enough from the network existing.
The only thing left, that also contains all the things I liked about Medium, is the discovery aspect. Also seems like it'd be the most difficult to implement on a distributed network. Maybe a few designated community servers, similar to tracker servers from the torrenting analogy, carry the information of what files are in the network. Im not sure exactly what the ipfs spec implements for this aspect of file discovery. But it seems some sort of designated 'discovery nodes' would be necessary. Maybe graph network nodes that ppl can query using their own discovery algorithms or ones shared within the community? Idk how well those would scale, I've heard from the Neo4j pitch that Behance rolled over their infrastructure to neo4j from cassandra and reduced their server requirements by a factor of 10. Maybe that kind of efficiency would be enough to support the network in general with a minimal number of critical nodes?
I haven't worked on an infrastructure level with anything I've mentioned, mostly just know of the technologies, so I might be completely off base with the individual things I proposed, but I feel like the general concept is worth dissecting.