Learning as Engineering

“In science, if you know what you are doing, you should not be doing it. In engineering, if you do not know what you are doing, you should not be doing it.”

The distinction between science and engineering is often framed as one between discovery and application. Science explains the world; engineering builds in it. But this framing breaks down when applied to learning.

Learning is commonly treated as a scientific domain. Cognitive scientists run experiments, publish results, and produce models of how people absorb information. But most learning, especially in the context of education systems, functions more like engineering. The goal is not to discover new truths about cognition, but to construct environments that consistently lead to better outcomes under real-world constraints.

Much of modern education technology reflects this engineering mindset. Algorithms are tuned, interfaces optimised, feedback mechanisms iterated, all in pursuit of a reliable, scalable way to move students from one level of competence to the next. The guiding question is rarely “What is true about the brain?” but rather “What works in practice, across contexts?”

This shift mirrors other domains that began as scientific inquiries and evolved into applied disciplines. Like early chemistry or computing, learning theory has matured to a point where the primary challenge is no longer understanding the system but improving it.

There are consequences to getting this framing wrong. If learning is seen purely as a science, success becomes defined by novelty and theoretical insight. If approached as engineering, progress is measured by utility, robustness, and speed of iteration. In that light, the best educational systems may not be the ones that explain the most but the ones that improve the most lives.

Science uncovers principles. Engineering turns them into progress. Learning, as practiced today, belongs more in the latter camp.