The CS assignments I was given in University were never 'very' simple. The first assignment in JAVA was to simulate an x team football league, each team playing every other twice and randomly determining match scores, printing match results and final league table. The C and VB assignments were of similar complexity. It's very implausible that two people could write very similar programs, in face, exponentially more unlikely with each additional line of code.
In relation to collaboration as cheating, as a teacher teaching CS at A-level (16-18 year-olds), I would absolutely love to do a big real collaborative project: break the students down into teams, assign roles, distribute modules, whatever, but it's not a good idea for several reasons:
1. Assigning roles. How do you divide responsibility equally to be fair, motivate the weak and strong students, and ensure that a project will actually produce some results?
2. Marking: how do you really know who did what? Has a weak student just rode on the backs of others? They might cover for him, they might not. There's no way to tell, and examiners don't like that.
3. Different ability and demotivation: if you're working with someone who can't write code and is destined to fail, how would you feel about your project? Yes your audio module might be perfect, but if you had nothing functional to plug it into you'd still be unhappy.
There's more of course, but the idea of teaching students how to program (and at the same time how to manage a project) almost forces them to work individually. The best I can do is teach them how to write efficient, maintainable code and hope that they end up working with others who can do the same.
