When the inputs are similar to the outputs, (e.g. text with linux CLI commands) functions can be very flexible and do things they were not originally designed t

When the inputs are similar to the outputs, (e.g. text with linux CLI commands) functions can be very flexible and do things they were not originally designed to do (e.g. linux CLI commands). An assumption that's implicit in what I've said is data flow programming. Data flow programming is when programming is thought of as connecting black boxes with inputs and outputs together. This way of programming is inherently modular. This is to be contrasted with (i) iteration, or recursionn which are inherently un-modular and (ii) The von Neumann model of computation, where there is a separation between code and data, and programming is thought of to be functions acting on the data (the problem with this model is that things need to be executed in a certain order, and sometimes the order isn't explicit in the code, which leads to bugs. Contrast this with data flow programming, where time is modelled explicitly) In LISP, the common interface between functions in the list rather than text. To encourage data flow programming and discourage recusion, map, filter, and fold are provided.