Transducers are composable algorithmic transformations; they provide a way to compose your operators to create a new one. They are independent of the input source and the output, and for this reason they can be applied not only to observables, but also to arrays, streams, and collections.
The transducers let you describe the operations you want to apply to an input source, this way you can create a pipeline of operations to be applied to them, without ever creating any intermediate aggregation.
You probably remember from previous chapters that, every time we apply an operator to an observable we create a new observable, so if we apply two operators to an observable we will create two extra observables, if we apply three operators to an observable we will create three extra observables, and so on. This does not happen with transducers, because you first describe all the transformations you want...
