Everyone seems to be talking about using threads to speed up and parallelize our code. I just published a blog post about a few ways that MRI uses threads internally to organize code, rather than to optimize code.