‘DB: Is there a difference between how control theorists and machine learning researchers think about robustness and error? BR: In machine learning, we almost always model our errors as being random rather than worst-case. In some sense, random errors are actually much more benign than worst-case errors. […] In machine learning, by assuming average-case performance, rather than worst-case, we can design predictive algorithms by averaging out the errors over large data sets. We want to be robust to fluctuations in the data, but only on average. This is much less restrictive than the worst-case restrictions in controls. DB: So control theory is model-based and concerned with worst case. Machine learning is data based and concerned with average case. Is there a middle ground? BR: I think there is! And I think there’s an exciting opportunity here to understand how to combine robust control and reinforcement learning. Being able to build systems from data alone simplifies the engineering process, and has had several recent promising results. Guaranteeing that these systems won’t behave catastrophically will enable us to actually deploy machine learning systems in a variety of applications with major impacts on our lives. It might enable safe autonomous vehicles that can navigate complex terrains. Or could assist us in diagnostics and treatments in health care. There are a lot of exciting possibilities, and that’s why I’m excited about how to find a bridge between these two viewpoints.’
This is my bet: the age of dynamic languages is over. There will be no new successful ones. Indeed we have learned a lot from them. We’ve learned that library code should be extendable by the programmer (mixins and meta-programming), that we want to control the structure (macros), that we disdain verbosity. And above all, we’ve learned that we want our languages to be enjoyable. But it’s time to move on. We will see a flourishing of languages that feel like you’re writing in a Clojure, but typed. Included will be a suite of powerful tools that we’ve never seen before, tools so convincing that only ascetics will ignore.
‘IRC without netsplits’ using Raft consensus
On a humid afternoon, Zhou went shopping for some of those very parts at a Bao An market. As he pulled his maroon minivan into a crowded parking lot, the full scale of Depu Electronics came into view: a three-story concrete behemoth roughly bigger than a Costco and roughly smaller than the Pentagon. Inside, it looked like the world’s largest Radio Shack going out of business sale: an endless series of booths with cables and circuit boards and plugs and ports and buttons and machines piled so high on tables that the faces of the clerks who were selling them were hidden from view. Each booth seemed to argue: We have exactly what you want and we have enough of it for all of your customers. Short of motorized wheels and molding, the market offered nearly everything an ambitious factory owner would need to build a hoverboard, just waiting to be bought, assembled, and shipped.