Writing

For a researcher "good English" means primarily writing English well. The prospect of writing instills dread in many people, who remember endless frustration in trying to complete an essay on some artificial topic for a composition course in high school. Because of this, books on how to write will continue to be produced and to sell in abundance for much the same reason as this is the case with remedies for baldness or wrinkles. Though several such books are excellent, it is easy to overrate their importance. The great thing about a great language like English is that once you get the taste of reading good stuff, and once you have done enough of it to get you over a certain threshold, the urge to write follows naturally. I guess one wants to talk back once in a while.

Hence the best strategy to fix writing problems is to find some category of books that you really enjoy, and then to indulge. I'm not sure this works for utter pulp, but it certainly doesn't have to be literature. On the other hand, the greater the better. Make sure you have a go at great writers of other centuries: Ralph Waldo Emerson, Edward Gibbon, Adam Smith, Thucydides, Homer, Plato, ... The list is at least one hundred long. Don't feel that you can only venture in there under the guidance of a prof. Try some and you may be surprised how satisfying it can be to read one of the utter greats. Taste some right now: here are the first few pages of a book over two thousand years old.

In response to the awful writing of students there are compulsory "How to Write" courses. These often have an effect opposite to the intended one: students in engineering or computer science see it as yet another gratuitous obstacle and take revenge by forgetting everything taught in the course.

Although I am pessimistic about "how to write" books, I found the following useful: "The Writing System for Scientists and Engineers" by Edmond H. Weiss. It is the System that is valuable. Here I propose to go beyond that book and make writing even more systematic. Look at the system of software development described and recommended by D. Hoffman and P. Strooper ("Software Design, Autmated Testing and Maintenance"). Here they emphasize that in software development one should constantly aim at work products. They list a wide variety in addition to code.

Similarly, the systematic production of a paper should concentrate on standardized work products. Candidates are:

• Literature list
• Glossary
• Figures
• Tables
• Mathematical skeleton. For a computer science paper with mathematical content, one can produce an ordered list of all definitions, lemmas and theorems and nothing else . That is a mathematical skeleton.

But if you really want a how-to-write book, here are some that I have found useful:

• William Strunk, Jr : "The Elements of Style" with revisions and additions by E.B. White. MacMillan, New York. This is a small book, a rare commodity these days.
• Kenneth Roman and Joel Raphaelson: "Writing that Works," Harper and Row, 1981. This is an even smaller book. Written by people from the Ogilvy and Mather advertising agency, when it was still the greatest. Whatever you may think of the advertising industry, it must be admitted that they managed to hire more good writers than academia.
• Jacques Barzun: "Simple and Direct," Harper and Row, 1975. I selected this among many excellent books not mentioned because anything by Barzun is great reading, partly because it is beautifully written.
• William Zinsser: "Writing to Learn," Harper 1988. The problem with taking a course on writing is that there is nothing to write about. Hence assignments are made-up, and artificial. Writing is hard enough without having to contend with lack of motivation. In this book Zinsser describes a natural remedy: "Writing Across the Curriculum." This is the idea not to have any writing courses, but to accept writing as part of every discipline, and therefore to teach writing as part of most courses.

After going into such details about writing, back to the idea that writing is not only important for external communication, but also to support thinking (i.e. internal communication). How about other ways to help a computer scientist think? Here mathematics plays an important role.

In a more specialized vein, it is useful to read Parberry's paper on how to give a lecture. His guide to referees is useful to any researcher, not just referees, because it explains the entire publication process.