Because for the most part it’s approached completely unscientifically - especially in the corporate setting.
What code is “cleaner and more maintainable?” All vibes.
How should we write tests to ensure they’re robust and covering all expected functionality? Who cares just get the tool to 90% coverage and ship it.
A carpenter isn’t a wood scientist
That’s not “computer science”, you’re talking about programming or software engineering, which are workers building what computer scientists have figured out. There are very few computer scientists. They are basically specialized mathematicians. Think Dijkstra. Google has most of them chained up in a basement somewhere writing sharding algorithms or something. It’s confusing because many programmers get CS undergrad degrees, but they are starting to make “software engineering” degrees.
It’s true that CS doesn’t use the scientific method, but neither do library science, “scientific socialism”, etc. Popper isn’t the be-all end-all.