Nice summary. I think it all boils down to the fact that 99% of projects just don't REALLY need what Erlang has to offer - they don't REALLY need five nines of

Nice summary. I think it all boils down to the fact that 99% of projects just don't REALLY need what Erlang has to offer - they don't REALLY need five nines of uptime, hotpatching in production (actually, the requirement is mostly the exact opposite!), or individual communicating nodes. And when the project really does need that, you can get pretty far with microservices today, which - arguably - implemented half of common LISP, khm, I mean Erlang - are supported by all the cloud providers. And anyway, if Stackoverflow can get by with a couple of .net and SQL servers, most anyone can. Maybe not Google, but are you Google? :) Erlang is just an engineering marvel, but in the real world, engineering is always a compromise.