Here are some of my ideas regarding preferences in style, which I believe are discordant with some of the current Wolfram practice. Style is a matter of taste so many may differ from my suggestions but I do try to give reasons. Some suggestions may seem obvious but not to everyone.
Mathematica notebooks written in the Wolfram Language are potentially a means of technical communication that is orders of magnitude superior to other mediums. Poor style and lack of simplicity and stability has hindered this goal.
The following are what I consider desirable features for a style sheet:
(1) A not too large gradation in font size between Title cells and Text cells.
(2) Input, Output and Text cells should have the same font size. That is what published papers use.
(3) Input cells can use a very light background color to distinguish them from Output cells. Otherwise, Input, Output and Text cells should have no extra adornment. Then you can close the Input cell and the Text and Output cells will read like a published paper.
(4) Section and the various subsection cells should have GroupOpeners on them. Input/Output groups should not.
(5) There should be sparse or no use of color, except for the background of Input cells or Title and various Subtitle cells. And graphics of course. Color in itself has no intrinsic meaning so if your trying to encode meaning with lots of color you better describe the coding to your reader.
(6) Black is the best choice for font colors on any text that is carrying important information.
There is a certain graying out of the Wolfram Language that I find undesirable. For example many Wolfram tutorials have the main text in GrayLevel[0.3]. This is the principal information in the tutorials! For what possible reason is it changed from a Black font? Just to be different?
Also labels on Frames or expressions are in gray. Are the labels less important? They might instead be distinguished with a slightly smaller or a different font instead of graying them.
Then units are also displayed in an even lighter gray font. Again, units are an important piece of information and should display in the same color as the rest of the output. In a notebook they have Tooltips; that should be sufficient to identify they as unit quantities. Does the Physical Review use a gray font for units?
The Inactive, Inactivate and Activate commands are extremely nice for the didactic display of calculation steps. But again the font has now been changed to a light red. In my mind this degrades the purpose of showing steps in a derivation or calculation. I have been able to change both of these displays to a black font by finding their styles in the Core stylesheet.
Wolfram Research is often making changes to fonts and styles in the Core styles. This often seems to be for no good reason other than playing around. It is especially bad when the new font intrinsic size is different than the old font. Then I have to change my stylesheets to keep my text and output the same size. This lack of stability is a major problems for those who are trying to use a consistent style!
My paradigm for general usage is to regard a notebook as a blank sheet of paper on which I'm writing my material. Wasn't this the classic style of writing documents? So I don't like Wolfram writing anything into my notebook accept in response to what I have entered or requested.
The great advantage of Wolfram documents are their active and dynamic features. Aside from that I think one should follow classic publication style for technical material. Except for one important matter. The physical medium of classical publications have been books and journals in codex form and this imposed the horrible burden of pagination. If one wants to add or remove material from the document pagination has be redone. If there are footnotes at the bottom of the page then they must be coordinated with the point of reference. If there are graphics that go along with the discussion then there is a problem in getting them on the same (or facing) page. If the discussion is long this becomes more difficult. (Scrolls were better at this.)
Then there is the horrible practice of equation numbers that usually refer to sequence in some chapter or section. Otherwise the numbers are meaningless. It's not always easy to find the equation. There may be many pages without an equation to use as a guide. Then when you've found the equation you're not at the place you wanted to use it and when you're at the place you want to use it you're not at the equation. Write it down or memorize it.
A basic maxim is DON'T JERK THE READER AROUND! Bring any material that the reader needs to the point of use. Wolfram documents are not bound by any requirement for pagination. So throw pagination away. Notebooks are basically scrolls controlled by scrollbars. However one long scroll is bad style because then the reader has to search for the point of interest. We have Sections and Subsections to divide the material into chunks based on the logic of the material and not the size of a page. I believe a good style is to present a notebook in Sections (and subsections) with openers all closed.- outline form. Otherwise you're presenting the reader with a long scroll. (I really dislike the tutorials that are presented as one long scroll without openers.) References and footnotes can be entered as Openers, Tooltips or popups at the place where they're used. Equations (and other material) could have descriptive names instead of meaningless numbers. The descriptive name could output the equation at any points of use. Or perhaps displayed with an Opener that also contained a button to print it in a new cell. If there is extended discussion associated with some graphic then the graphic can be displayed in a window next to the notebook. Scrolling the text will not scroll the graphic out of sight.
Then there is the user interface style of some of some of the calculational facilities. I think WRI would do better to adhere to a Newtonian Reduction/Synthesis approach rather than top down routines with a bevy of options. Roughly speaking, WRI should do the reduction and the writer does the synthesis. The trouble with top down with options is that WRI can't anticipate all the variation writers might want. If the writer can have access to lower level objects of possible interest in a calculation then she can use them in new innovative ways. It's not that high level routines that performed common tasks should not be provided, it's that a writer should be able to repair to lower level objects when she sees an innovation or rearrangement that's not accommodated by the options..
For example it would be nice if significant pieces of graphical objects, such as curves or surfaces, could be provided as graphical primitives, basically combinations of points and lines. Then these could be transformed or rearranged and combined with other elements in a natural way without having to jump back and forth between the Graphics .level and the primitives level. It used to be easy to do this but now more and more elements are hidden or the graphical structure is often modified by WRI. Objects such as Gauges are really graphical objects but they cannot easily be combined with other graphical primitives. One might want to align a HorizontalGauge precisely with an edge. This is extremely difficult to do with Inset because the size and spacings in the gauge are not available to the writer.
Another example is the RowReduce routine. I have a facility in my Presentation application for manipulating matrices step by step in a didactic manner. I would like to use RowReduce but confine the pivot elements to a given block in the matrix. For example I might want to attach an identity matrix to the right of another matrix.But I only want to pivot over the left side columns. More generally I would like to specify a block of rows to be reduced and a contained block of rows and columns to pivot on. But I can't do that with the given options. A pity since WRI is much better at row reduction than I am.
Wolfram documents are a powerful new media. Don't waste time trying to make them conform to old media. But it still lacks stability, clarity, simplicity and has too many cutesy elements. But it has many nice things also, which is why I still bother to use it.
