SOME BASIC PRINCIPLES

Research is a highly nontrivial human activity and there
is a lot to say on it. But little is typically said.
I wish to deviate from that tradition

1. Why do we write papers?

Almost all human activities have a purpose,
and understanding the purpose is instrumental to
discussing how to conduct that activity.

The purpose of writing papers is to communicate a set of ideas
to people who can use them and/or develop them farther.
(The cynical answer by which the purpose is to promote
an academic career will not do; it does not identify what
promotes such a career -- serving the scientific process does.)

2. Identifying they key idea(s) in the paper.

3. Identifying the relevant research community and its needs.

N.B.: These are not only the experts but rather also the novice.

4. The "economics" of writing-vs-reading.

* The writer is better positioned to clarify the ideas. 
  It will take her/him less energy to clarify.
* The writer serves many readers. So his time saves the time of many.

5. Serving the readers.

* Focus on the needs of the readers, not the desires of the writer.
* Elaborate on the novel conceptual steps, 
  not on the standard technical derivation.

Novice feel insecure about the standard techniques and tend to diminish
their own innovations.  Thus, they tend to focus on the standard derivations, 
and hide their innovative conceptual steps. Both things are very wrong.


PITFALLS

* checklists (of things the author wishes to say).

* obscure generality (rather the right level of generality).

* idiosyncrasies: 
  using terms and/or symbols that bear a meaning only to the writer.

* lack of hierarchy and structure.

* Talmudism: discussing ramifications and critiques prematurely.

* labyrinth of implicit pointers (rather than explicit references).

* sentences with complex logical structure

* mixure of Math and text that is hard to parse

* cumbersome notations and terms


THE PAPER ITSELF

No canonical model. Fit the structure to contents.

* Title (informative) and author list (alphabetical).

* Abstract (informative, self-contained in high level)

* Introduction

Note that some readers may decide to read based on the title,
some may only read the abstract, and more just the introduction.
Serve them all...

* Body

- discuss definitional choices: 
  are they immaterial, for simplification, or essential?
- emphasize novel conceptual steps
- use a single numbering system for technical items
- use informative citations (e.g., include Thm number).

* Conclusions (justified only in 5% of papers).

Should discuss issues that cannot be discussed in the introduction,
since they can be well-understood only when reading the body.
Only such issues should be discussed in the ending section,
rather than in the introduction (e.g., a subsection of it).

* References and Acknowledgements