Choosing the right problem to solve was always very important. The code wasn’t always the hard part, but arguably it really was the hard part more often earlier on.

For example, a personal computer operating system that runs all your existing programs at full speed but doesn’t crash all the time and the ability to run a virtual machine on an x86 PC both took as long as they did to get to market primarily due to the technical difficulty, not because it was hard to find people ready to pay for them. Those kinds of hard coding problems are less plentiful, or at least less obvious today.