I’ve previously written a bit about a hypothetical new hypertext medium displaying several “cards” or “scraps” at once. I think “leaf” is maybe better terminology.
The idea is that your memex consists of some collection of “leaves” and dynamically lays out as many as will fit on the screen, something like TiddlyWiki or Ward’s Simplest Federated Wiki; but the leaves are a bit smaller, like a line of text rather than a paragraph. Each leaf is identified by a globally unique leaf ID which can be used to link to it, and those links may be activated explicitly (for example by clicking) or implicitly in order to display some other leaf. Typically these implicit links include previous and next links, which permit reading through a linear document in a more or less straightforward fashion by shifting the focus to earlier or later leaves --- if necessary by clicking or tapping them, but usually just by centering them in the display. (This implies that the potentially infinite graph of implicit links cannot simply be fully traversed to decide what to display at any given time.)
In addition to this subatomic conception of hypertext, we have parameters in the links, like #fragment identifiers on the WWW. The rendering of a leaf for display is done by some arbitrary Turing-complete code, contained in the leaf or linked from it, which is supplied some arbitrary blob of parameters from the link that was followed to display it. By interacting with the display (clicking on it, typing characters into it, etc.) you can change these parameters under the control of that code.
In this way, it becomes straightforward to write a leaf that, for example, contains data and buttons to plot the data in different ways by invoking other leaves with the data as parameters, or that invokes an “infobox template” leaf with some parameters in order to present that data visually in a consistent way.
I’d like to do some kind of infinite regress thing where the leaf contents are themselves produced by invoking some piece of code with some parameters, and the leaf ID likewise consists of the ID of that piece of code and its parameters, so that ultimately we ground ourselves in a single “primitive leaf” that provides some kind of fundamental virtual machine. But I don’t quite see how to do that yet.
In the stuff you have on the screen you can thus have a mix of stuff you wrote, stuff you’re taking notes on, and code.