You can still take the 2010 approach and just run everything on Rails and Postgres. It’ll be a long time before any startup outgrows that stack, and by the time

You can still take the 2010 approach and just run everything on Rails and Postgres. It’ll be a long time before any startup outgrows that stack, and by the time they do they’ll have the ability to rewrite like Twitter did. Anyone who starts off with anything more than that (or Django or your favorite language’s equivalent) is just wasting time and effort trying to be trendy. Over-engineering a startup only serves to keep you from testing market fit, which is only a good plan if you’re trying to milk more investment money because you know you’re going to fail as soon as you launch.