STET was written to explore an approach to text editing that followed the principles of Structured programming. It allows programs and documentation to be written 'top-down', with blocks of code or text kept to a limited size (usually less than a page).
This was "a first attempt to take the structure out of the domain of languages, and into the domain of editors. In addition to conventional editing facilities, STET gives the user a third dimension: a tree structure that may be traversed using Program Function Keys much as scrolling is normally implemented.
Top-line commands generally apply to the entire file being edited, and include the usual operations such as saving, searching, replacing, movement around the file, and so on (over 60 in all).
Line commands apply to a single line or a number of lines, and allow copying, moving, overlaying, merging, uppercasing and the like. An innovation in STET is the ability to mark a group of lines using the < line command to mark the first and last lines of a group. The group of lines can then be manipulated as a whole; they can be copied, merged, etc., and also make possible the 'structured editing commands' which provide the folding capabilities of the editor.
The block reference concept is very similar to the idea of links in a Wikipedia file, and indeed in STET a block can also be created by putting a new reference on a line and then 'entering' that reference to edit the block.
The structured editing commands also allow the user to see a map of the tree structure of the file, list the blocks alphabetically, and navigate the structure of the file. Navigation upwards and downwards or to a named block is possible, along with 'go to next block at this level' (that is, go up a level, find the next block reference at that level, and then go down a level). Other commands allow blocks to be renamed, expanded, etc., and also allow a flat file to be built from all or part of the structured file.
A structured file is stored on disk in a special format which allows effcient loading and navigation of the blocks in the file. This allows each file to also store useful meta-information (such as statistics of action key usage, total edit time, etc.), but has the disadvantage that a flat file has to be 'built' from the structured file before (for example) it can be processed by a compiler or word-processing utility.