> And, for reasons we’ll see in a bit, we flip the meaning of the continuation bits: we’ll use 0 to mean “another byte follows”, and 1 to mean “this is the last

> And, for reasons we’ll see in a bit, we flip the meaning of the continuation bits: we’ll use 0 to mean “another byte follows”, and 1 to mean “this is the last byte”. EDIT: just noticed that I'd skim-read over: "imperial varint places all of the continuation bits at the beginning of the encoded value", which makes my comment completely wrong. ~Doesn't even mention possibly one of the most useful features of encoding this way, which is that you can never [1] end up with a 0 byte in the output, meaning that you can treat the output as a C string.~ ~[1] or at least never assuming you're trying to create the shortest encoding~