What is the showS trick in Haskell?

showS uses the difference list approach to efficiently concatenate individual components of the shown value. The function takes the value to be shown, and a string to append to the result. The appended string is passed all the way down to the right-most sub-value until it reaches a leaf, where it is actually appended.

There's a description of difference lists (including showS) here http://www.haskell.org/haskellwiki/Difference_list

In the standard library, ShowS is defined as:

type ShowS = String -> String

This is a difference list. The trick is that a string xs is represented as a ShowS by the function that prepends it to any other list: (xs ++). This allows efficient concatenation, avoiding the problems of nested left-associative concatenation (i.e. ((as ++ bs) ++ cs) ++ ds). For example:

hello = ("hello" ++)
world = ("world" ++)

-- We can "concatenate" ShowS values simply by composing them:
helloworld = hello . world

-- and turn them into Strings by passing them an empty list:
helloworld' = helloworld ""

It's called ShowS because it's used in the implementation of the standard Show typeclass to allow efficient showing of large, deeply-nested structures; as well as show, you can implement showsPrec, which has the type:

showsPrec :: (Show a) => Int -> a -> ShowS

This allows handling of operator precedence, and returns a ShowS value. The standard instances implement this instead of show for efficiency; show a is then defined in terms of it, as showsPrec 0 a "". (This default definition is in the Show typeclass itself, so you can just implement showsPrec for a complete instance.)