This is a problem I have been thinking about for a couple of years now. A good amount of time in programming is spent between trying to switch between various “modes of expressions“: data ⇄ code ⇄ graphics ⇄ data all of which are (homo/iso)morphisms among each other. 
Each mode of expression has a distinct value from a human viewpoint.
Data allows for operations on it using code. This produces distinct transformations (let’s call them representations) which can be serialized into a format which enables transferability.
Data allows for operations on it using code. This produces distinct transformations (let’s call them representations) which can be serialized into a format which enables transferability.
Code (+ a runtime) creates process whose behaviour can morph data from one form to another (as a side effect or direct effect). It allows for wide range of expressions because as it allows for abstraction. Code can itself be treated as data.
And finally, graphics is used to build gestalts which highlights certain features and hides others. It can be interpreted to get a meta level meaning (The graph is climbing!). Graphics it is itself a form of data which is projected on to a Cartesian (pixel) grid on computers.
Code, Data, and Graphics have a symbiotic relationship between each other as code is represented as data on computers which is then rendered as glyphs in a typeface using graphics.
