I came across this update early this morning (thanks to @Paul_Kinlan for resharing it):
I owed this post to Noah Slater after a twitter interchange last week. I woke up from a dream about the crystallization of flakes and couldn’t fathom precisely why the process is symmetric. In particular, how could one end of the flake know about the other. It turns out symmetric flakes are the exception, not the rule. There’s no hidden magic and the process is a function of energy minimization and some general properties of crystallization.
