Hail to the Victors: Exam Results.... discusses exam misgivings. I had planned to put off my discussion of this till Tuesday, but there is one point she raises that probably merits immediate attention. Here's an interesting point she makes:
It got me thinking though... in real life, how often you come across an ER diagram in isolation with no background explanation attached (such as the examples on the exam)?
There are really two issues here. I'll address the most obvious first which might be rephrased as “Did I just take a test on some meaningless logic interpretations that have no analog in real life?”. You will be lucky to receive complete documentation in many instances when dealing with a database system. A lot of times, you may just get someone's SQL code and be expected to construct the ER diagram or some equivalent. How do the tables relate? The ER diagram seems to be one of the more efficient ways of doing this.
As M Go Blue points out, there is software that will construct the ER diagram for you, but you are still stuck having to interpret ER diagrams. Further, there is no guarantee that you will know a lot about the domain (business context) in question. A lot of times you will just be parachuted in. So, basically the exercise is not meaningless because alot of times, you will feel lucky to have even an ER diagram.
That does not mean that interpreting ER diagrams is easy. In fact, I think it is very hard and have some difficulty with it myself. This leads to the second, more subtle issue I see implicitly buried in Krysta's observation. Basically, we are stuck dealing with logic machines (an observation echoed by Tranquilidad and Kornstein999). SQL, ER diagrams, and normalization are all just part of the same logic machine. They all have the same disconnect with the alogical reality we all experience. To get your projects or any real world endeavor to work right in a database, you will have to understand and deal with this fact.