Which brings us back to 2001 and Apache's prefork model. Threads are pretty battle tested at this point, and allow for different reasoning about the concurrency of a problem. The problem with node is that it pretty much only allows for one type of scaling if you start running into performance problems. Go gives you a much bigger toolbox to work with if, much like the post's author discovered, your problem set does not match up with the answers node provides.
All is well until you have to deal with SEO for SPAs :) and then there is little choice other then node. https://github.com/audreyt/node-webworker-threads spawn real threads with web workers compatible API.
