Changing languages is not how progress happens. It's only progressive when the long-term benefits of the language outweigh the inefficiency of training all your devs to use it. Assuming that is true for Go (debatable), at the end of the day, SoundCloud still gimps their hiring pool far more than if they choose something like Node. Learning doesn't bother me at all - I like learning - but I can't advocate for battle-testing Go in a mainstream environment when their are plenty of other fast and tested languages. If Go evolves into a language that is more desirable in the everyday stack, that process should be organic, just as it was when people decided to switch to Ruby.