The distinction between writing code and writing prose is not as great as some might imagine. I recently read an article that compared novelists’ metaphors concerning writing. The distinctions included the “architect” who meticulously plans out the structure of the novel. Plots, characters, sets, chapter structure—everything—are diagrammed and refined prior to beginning writing. All that remains is word choice, dialogue, and the gritty details of putting it all on a page. Compare this to the “farmer” approach where a seed is planted in the form of a key idea or plot development. The writer begins with that seed and nurtures it in a continuous process of development. When the tree grows lopsided, there is pruning. When a branch withers, there is watering and attention. The balanced whole builds organically and the architecture is an emergent property.

Coding is similar. We generally know the architecture in advance, though there are exceptions in the green fields. Full stack development involves decoupled database back ends, front end load balancers and servers, and middleware/coding of some stripe. Machine learning involves data acquisition, cleaning, training, and evaluation. User experience components then rely on “patterns” or mini-architectures like Model-View-Controller and similar ideas pop up in the depths of the model and controller, like “factory” patterns that produce objects, or flyweights, adaptors, iterators, and so forth. In the modern world of agile methodologies, the day-to-day development of code is driven by “stories” that are short descriptions of the goals and outcomes of the coding, drawing back to the analogy with prose development. The patterns are little different from choosing to use dialogue or epistolary approaches to convey parts of a tale.

I do all of the above when it comes to writing code or novels.… Read the rest