=== Asymmetric Paths ===

You can write asymmetric paths, e.g. `not . and[not x id -> id]`.
This will reduce to `and[not ⨯ id → not]`.

An asymmetric path consists of two parts:

- The functions that are used to transform the input, e.g. `not x id`
- The function that is used to transform the output, e.g. `id`

The arrow `->` is used to distinguish between the two parts.

When only `id` is used to transform the inputs, one gets function composition:

  f[id x id -> g] <=> g . f

The most difficult thing to understand about asymmetric paths is when some
function, that is not `id`, is used to transform the input.
This means that some information is removed or added to the input.

When some information is added, it happens inside the path:

  f[g(a) -> id]

When there are no arguments inside the path and the function is not `id`,
it means that some information is removed.
If the normal path exists when mapping to `id`, then it is a proof that there
was some information in the input was not needed to compute the output.

The only way one can have a normal path that uses all the input information,
is by mapping to some function that is not `id`:

  f[g0 -> g1]

This contracts `f` along the path such that:

  f[g0 -> g1] . g0 <=> g1 . f

Notice that `f[g0 -> g1]` does not need to be solved in order to be sound.
It is like complex roots of polynomials, they are always defined.
In order to convert it into a real program, one must find the solution.