Sound Writing

The purpose of CS papers, like in other sciences, is to contribute to academic knowledge in the field and share your work. The audience is other computer scientists, though not necessarily in your specific subsection of the field. A paper shares new ideas or contributions, so that others can learn from them and potentially build on your work. While student papers may or may not actually be presented at a conference, they are practice for this type of academic work and should be written with the same goals in mind.

“If you’re on the monkey bars, you have to have one hand here and then you reach out and grab here and then you let that go and you reach out and grab here. At no point do you let go with both hands. In the same way you are carrying your reader from one sentence to the next to the next and you can’t let them go.”

In academic papers, computer scientists must be able to convey complex ideas with enough detail and clarity that their colleagues can understand them. Therefore, it’s vital for the writer be able to break down complicated ideas into precise, logical steps. This is sometimes referred to as algorithmic thinking. For example, think about the level of clarity that would be required to teach a child to perform a task like tying their shoes... over the phone. Your writing should walk the reader through every step, never making logical jumps that might leave them behind. It should start at the beginning and introduce all the important terms and notation before you use them.

Concise writing is also valued in CS. While you should include anything the reader will need to understand, avoid unnecessarily fancy words or sentence structures. The ideas are complex and difficult to understand, but your writing should not be. “Untwist” your sentences, clearly linking nouns and verbs and avoiding dangling phrases. Likewise, precision in language (see Subsection 7.3.1) is very important. Avoid vague phrases or sweeping claims. Instead, use language that conveys your ideas as exactly as possible without going beyond the scope of your work. The main challenge is including all the necessary information, but nothing extra.

In CS papers, evidence takes the form of data sets or models from your own work, as well as previous related information or conclusions from other peer-reviewed academic papers. This often includes equations or tables mixed with text. Be sure to meticulously interpret these data for your reader, as they are often confusing. When using evidence from others, also be sure to explain how it relates to your topic and to cite it properly. For your data to be meaningful, think about how you can use experiments to measure and convey the success of your project. This gives your work meaning and context for a larger audience.

The Science of Scientific Writing⁶ by Drs. George Gopen and Judith Swan includes excellent advice on how to convey complex ideas clearly. BUGS in Writing: A Guide to Debugging Your Prose by Lyn Dupre is a useful book about writing for people who work with computers.