Let's say you want to create a set by recursing along a million long sequence, collecting the unique values. This would be done with an operation called `reduce

Let's say you want to create a set by recursing along a million long sequence, collecting the unique values. This would be done with an operation called `reduce`. The time and storage complexity of that would be exponential if each step you had to clone a copy of the output list. With a persistent data structure, as much of the structure as possible is re-used. And a list is the simplest possible structure. What about a tree? Using `reduce` is, I would say, in the top 5 techniques used in functional programming. And it would be severely hobbled without efficient persistent immutable structures.