Ok, I get it. Some settings are more focused than others, I know this, and have no problem with it. In some games, you are samurai climbing a mountain to seek a witch. It's pretty clear what the PCs do - they haven't a lot of choice there. More and more, though, I'm seeing criticism of broadly designed games for the sin of not being narrowly designed. "What do the PCs Do?" is the question that sets my teeth on edge. "Whatever the hell they want to!" is apparently not a sufficient answer. The answer must be explicit, restrictive, and simple. I see this in worded other ways as well: "What is the prototypical adventure for this game?" "How do you see the session structure?" As if there is a correct answer for each game!
The questions imply that there is - must be - a correct answer, and by not giving it, you are confirming that you don't have a clue, particularly heinous if you wrote it, you dolt! When asked of a game which is obviously broadly designed, it's a subtle variation on "Have you stopped beating your wife?" It's a debating trick and a cheap way to score points. It implies poorly thought out design when an easy answer can't be given. Sometimes a game is designed to give the PCs the option of deciding what they will do for themselves.
Thing is, I'm only seeing it lobbed at unabashedly trad games, or at games which are narrowly defined, and where it is perfectly appropriate to ask. Games like Diaspora, which I love for its pure openness, but which are wrapped in the mantle of the Forge Diaspora, are not asked this question. If you ask this of the Diaspora designers, they would have no more clue than I would if you asked me this about OHMAS. The group will define what the PCs do in either game, because they were designed to do just that. In both games, there are scads of tools at the group level that allow the group to define its own focus within a broad array of options. It's a design decision that reaps immense rewards in replay value.