Three Phases of Engineering

There are three basic steps to any engineering task.

First, you have to understand the constraints that the product will operate under, and also to understand as much as you can--to the limit of your research budget--about the environment that the project will operate in.

Second, you have to understand the capabilities of the existing materials and solutions to the problems you'll face in this project. Sometimes this means researching and finding obscure solutions or materials that fit the peculiar constraints of your project.

Third, where there are impossible constraints (and there usually will be), you have to invent new techniques or materials until you can satisfy the constraints of the project.

Most of the problems come from the first part: not getting the requirements right, or not sufficiently understanding the environment that the product will operate in. There's also the risk that you'll mis-estimate the odds of some low-probability, high-impact failure case. You may estimate something as having a return period of a million years, but if it's more like one hundred, you might be in big trouble.

This is why audacious moonshots that require spending a lot of time doing the third part end up pushing society forward, where technology products that just combine existing tools don't have the same effect.