Translated into chess databases, I believe that I could take all the games of Morozevich from the ChessBase database and package them up and sell them as "Joey's collection of the games of Morozevich" and I would be okay because I didn't copy chessbase's entire collection. Now if chessbase had added something to Morozevich's games, like a fake game, I might be in trouble. That's why mapmakers sometimes add bogus roads to their maps (bicyclists who rely heavily on maps sometimes get messed by this - "We were supposed to pass a road called "Memory Lane" - did anyone see "Memory Lane"? )
In the Feist v. Rural case, proving that you had copied all the raw information did not seem to do much good. From my understanding of this thread, selecting games of Morozevich would be enough of a "creative" act to enable you to copyright your collection of games.
This is a fun thread
What about the copyright laws in countries besides the US?
I don't know about other nations. But the question of adding "something" to the database is an interesting one.
If someone makes up a game and includes that game, then that is a purely creative work, and not a fact -- since it did not transpire in reality. In such a case, copying that "something" (be it a game that did not happen, or a coninuation that didn't happen but was rather analysis that wasn't played after a resignation for example), is clearly a violation of copyright because it is precisely not factual data.
So there's an inherent obligation on someone publishing a new database to ensure that the games being provided are indeed factual games and not creative imagined games designed to "watermark" a database.
Of course you would have to prove that they knew it was a creative work in order to sue them for stealing fiction. Variations in analysis are in the public domain, even though the verbal language accompanying the analysis could be copyrighted.
Something I don't really get, although the answer is probably staring me in the face and I'm just being oblivious: What's the point of having a huge database of games along with software like CB or SCID to manage it? There's obviously a demand for them, but it doesn't seem like having a few million unannotated games on hand would be that useful -- for study, it's trivial to find free, online, annotated games, and that doesn't involve taking up how many gigabytes of hard disk drives.