Additionally you can't discount the fact that you might think it's just fun to try to emulate the 8-liners. Isn't that the way emulator authors usually are? They sometimes emulate things just to see if they can do it, and don't really care to play the games. Would you say this is partially true about MAME?
I would say that for the devs who have made the most progress over the years that's true. Everybody has a limited number of games that they like / enjoy, and if you limit yourself to emulating those, then you don't get very far. The devs who get their enjoyment from actually figuring things out, and completely don't care the nature of the game they're dealing with tend to be more prolific.
Of course, an intimiate knowledge of the game / system you're emulating helps a lot, subtle effects the hardware can perform are sometimes missed by people who don't know what the games are meant to be doing because they don't know the games.
Quite often that's where teamwork comes into play, and also where sites like MAMEtesters with people posting original HW videos can help. Getting the initial work done can help ensure that the roms are properly dumped, and the hardware is mostly understood, while specific references can help fine-tine the emulation and help bring it closer to a more 'pixel perfect' level, as recently happened with the titles on the Psikyo SH2 based hardware.
I was mainly just saying that people calling out the 'it's a documentation project' as BS really don't understand what's going on behind the scenes, or the nature of the project. Yes, it lets you play thousands of games, but the driving factor is more often just as much about the challenge of figuring out the hardware, and documenting it so that anybody else can use MAME as a reference piece than anything else. It's a misconception that the often quoted text is just to cover the backs of the developers.