I sat in a meeting the other day where someone said, “well, computer scientists are obsessed with determinism and refuse to recognize non-determinism.”
And it got me thinking, again, about something I wrote about many, many years ago (2012).
What I wrote was that the history of thought was about moving from a universe where everything was understandable to a world where everything could not be understood. And that article can be found here https://www.forbes.com/sites/quora/2012/06/04/why-would-anyone-want-to-work-for-zynga/#7846a1cb658d. There are a lot of things that I wrote that are embarrassing. I was naive. I was optimistic. And yet, I was right in echoing the thoughts of much smarter people.
Later on, I synced up with an old friend, and we wrote an essay on the limitations of human understanding. That, homo sapiens are inherently limited in their ability to understand the universe. And that limitation makes revelation, the intuition of truth without the ability to prove the truth, not a failure of reason, but an indication of its limits.
And so ten years later, I found myself giving a talk to a bunch of engineers about desired state systems.
The core of the discussion was that planning algorithms that attempted to search a state-space exhaustively were inherently flawed, if the system was exposed to unknown external inputs. When trying to change the state of such a system, if you assume you know how to go from the current state to the desired state, you are wrong because the current state is invalid at the time you made the plan.
30 years ago, I remember sitting in a class learning about planning, and recent research on machine learning, POMDP, and thinking what does this have to do with anything.
It turns out, everything.
Leave a Reply