I can think of one thing in CS that I see gives people with little to no CS education a lot of trouble. Algorithmic analysis, to be specific big O notation. I've seen people not get algorithmic growth at all and end up implementing something that is O(n^2) when they could easily come up with something that's O(NLogN) or even O(N). Surprise surprise when they have to process even a middle amount of data they have problems. I have learned something else though. If someone tells you that their app runs in N^2, log(n), or nLog(n) time they probably know what they're talking about. If they say N! or even C^N they really know what they're talking about. If they tell you it's linear that could either mean it really is linear or that they don't know of any other running time.(Literally I saw code that was obviously N^2 but the developer said it was linear because he didn't know of any other type.)