Using “will” and “should” in technical writing

I was writing a tutorial earlier today on oriented bounding boxes object detection models. Object detection models identify the location of an object in an image (i.e. the location of a solar panel in an aerial image). Oriented bounding box models identify the coordinates of objects more precisely than traditional object detection models, whose boxes are horizontally aligned.

When I was writing the tutorial, I wrote the following sentences:

Object detection models return bounding boxes. These boxes indicate where an object of interest is in an image. In many models, such as YOLOv8, bounding box coordinates are horizontally-aligned. This means that there should be spaces around angled objects.

In the last sentence, I say “should.” But, as I thought about it, the appropriate word is not “should.” The goal of the model is not to draw spaces around angled objects; this is a side effect of its behaviour. A more appropriate word is “will”, which is commital. This will happen. This is the case.

Let’s revise the last sentence:

This means that there will be spaces around angled objects.

This sentence is more accurate. By definition, bounding boxes that are horizontally aligned; there will be spaces around angled objects.

It is tempting to use words like “should” and “can”. When these words are used, it introduces ambiguity. “should” implies that something might be the case, but might not. As I think about it, there are few scenarios when something should happen in a technical product or piece of code that you are documenting.

“will”, on the other hand, implies that X leads to Y. The language is more direct. The reader knows exactly what to expect. Of course, you should only use will if X does lead to Y. For example, I used “will” in my earlier example because non-oriented bounding boxes will result in spaces around angled objects.