Using abstractions and models

“All models are wrong, but some are useful.” – George E. P. Box

One meaning of the word model is: A smaller or simpler version of the original item. The model could be a physical object like the small robot in this illustration. Notice that Blaze isn’t trying to move the arms of the huge robot, nor trying to move the heavy blocks himself. Instead, he is working with a model robot small enough for him to literally put his own hands around. This is simplifies the physical work he needs to do, just like a simplified model of an idea makes thinking easier.

For example, classical mechanics is a model: It’s Newton’s easily-computed approximation of the more complex reality of motion. In computer science, we make a model every time we write a program. We must choose the information and level of detail represented in our program. Some details must be left out. If we tried to include everything in a model or program, we would end up simulating the whole world!

In a complex system, we might use many different models and make them work together. We might not even care if one part of the system was switched out for something else that can accomplish the same goal. We could say we’ve abstracted that part of the system. Carefully selecting the qualities we care about and ignoring the rest of the details is the key to abstraction. When we deliberately separate our system into parts that can be individually understood, tested, reused, and substituted, then we are creating new abstractions.

See also: Modularity, map–territory relation, marionette.

No comments yet.

Leave a Reply