Mathsenvy.
The first generation of computer scientists included eminent mathematicians, but computer science was concerned both with writing software (on highly constrained machines) and constructing the machines. Over time three things happened.
1. Building the machines became both more specialized and less important to general computer scientists. We use commodity hardware even in top end applications now.
2. The machines became less constrained.
3. Maths departments (which often hosted Computer Science departments) became more pressured due to falling student rolls and at least in comparison to Computer Science rolls, and highly defensive of academic promotions and kudos.
So perversely the incentive for academic Computer Scientists was to focus on “hard” mathematics at exactly the time when the concerns of usability, team scale and maintenance became significant – about 1970. At the same time the Electronic Engineers that could have tempered this tendency disappeared from the scene.
The fundamentals of modern software development all emerged at about this time; Smalltalk in 76, relational databases at IBM in 1970, Arpanet in 68, GUI’s at Xerox in 73. I would definitely point to the web as another critical innovation. but it very definitely blindsided