It was a big question 20 years ago whether computer science was an actual science. It's still something of a question. I think there's been recent recognitio

It was a big question 20 years ago whether computer science was an actual science. It's still something of a question. I think there's been recent recognition that computer science is something of a metascience. It brings benefits to many other fields (for example wireless sensors feeding a database can provide sociologists hard data where they used to have to rely on surveys) and in some sense the role of computers is defined by the problems they are applied to. Pure computer science is usually pretty esoteric (e.g., making the cloud work faster and better and safer). Learning how to program, though it may be called "computer science" for undergraduates, is not really much about science. So I would differentiate between the science and learning programming. You can learn programming many ways, by doing it while working on physics for example, or company training, or a vocational school, or self teaching, or college courses. The quality of the resulting programmer may depend more on the person than the training. Doing science is a specific skill set and that probably requires special training, though you can self teach yourself that also. Engineering is more about making things work and when what you're trying to make work MUST WORK (like an airplane) then there need to be some explicit certifications and training to ensure you actually know how to make the software act safely and sanely. Other engineering applications of computers are more experimental (like devising new ways of jamming radar signals) and while they still can be life-critical (for the jet fighter pilot) they will also sometimes tolerate less formal standards as long as the application works. Other types of engineering are even less safety/life critical. So I'd divide computer engineering into degrees of criticalness which determines the level of training involved and necessary.