I'm souring on reading papers. 

For example, "Keeping CALM" is a really well written paper that introduces a new concept (monotonicity). But what can you actually do with this in terms of designing or evaluating a system like Farcaster or Ethereum?

Not much - it's just a well articulated restatement of some core distributed systems consistency principles that we've known for a while. 

Most CS papers don't really seem to do "Science" anymore - there are a few important ones, but the vast majority are publications for the sake of publications.

I'm souring on reading papers. 

For example, "Keeping CALM" is a really well written paper that introduces a new concept (monotonicity). But what can you actually do with this in terms of designing or evaluating a system like Farcaster or Ethereum?

Not much - it's just a well articulated restatement of some core distributed systems consistency principles that we've known for a while. 

Most CS papers don't really seem to do "Science" anymore - there are a few important ones, but the vast majority are publications for the sake of publications. 

https://arxiv.org/abs/1901.01930

I agree, many papers I have read were theoretical improvements or ideas and may be some mathematical formulae to address well known problem who nobody asked for.

Ethereum and Farcaster are at the far end of P2P systems where speed of synchronization is often sacrificed for eventual consistency, exact problem of Farcaster. https://iohk.io/about/ explored this in big way.

Technowatermelon. Elder Millenial. Building Farcaster. 

nf.td/varun