Computer Science

[Eliot Moss]

On 10/18/2024 3:44 PM, matthew patton via Cygwin wrote:
> On Friday, October 18, 2024 at 02:09:31 PM EDT, Jim Garrison via Cygwin <cygwin@cygwin.com> wrote:
> 
>> Most university courses in "software engineering" don't begin to cover
>> the actual knowledge base and, more importantly, internal mental
>> processes, discipline and curiosity required to do quality software
> 
> My 200-level CS class at Carnegie Mellon (91-95) if your code submission handled ALL inputs properly (the test dataset deliberately omitted some edge cases) you got a 'C'. If you missed some edge cases the best you could hope for was a 'D'. If it didn't even compile you got an 'F' obviously. If you actually wrote comments in the code that explained non-trivial logic you got a 'B'. Only if you had robust error checking (with optional recovery as the case may allow) did you get a higher score. The instructor(s) were ruthless and forced out 2/3 of those who thought they had the chops to do CompSci. It was also all done in KR 'C' so the opportunity for failure was that much greater.

I respect CMU a lot ... but I have some concerns about this approach.  Years
ago I read an NSF report titled something like "Towards a Pump, Not a Filter."
What you describe strikes me as a drastic filter.  The reason we need "pumps"
is to improve gender, ethnic, etc., diversity in computer science and related
areas.  This is not "just" being fair, "politically correct", etc.  Diverse
teams build better products and come up with more good ideas.  The big
challenge is somehow managing both to set high learning standards *and*
support talented, but perhaps less prepared or less previously supported,
people who *can* meet the standard with enough support, and who can broaden
the pool of creative talent.

Best wishes - EM