Sorry, I should have been clearer; I think it is disingenuous to downvote for that reason if there is no FAQ or guideline expressing the current instantiation of the community's preferences for what is on- or off-topic.

I am all for using a numerical mechanism like voting to aggregate information, but it does come with a reputation cost for people who make the effort to create a post or pass on a link. So it does more than just signal what is liked or disliked; it also has a personal element that may discourage people from trying. If the expectations are clearly spelled out, then voting/downvoting is fine and can be interpreted against that informative backdrop.

Also, this sort of voting allows us to aggregate a coarse "yes" or "no" kind of preference about a post, but I think it would be pretty difficult to impute nuanced preferences, such as classifying topics and sub-topics as on-topic or off-topic, just by aggregating these votes. There's no clear delineation of the "why" behind the vote, and that metadata is more important for understanding the squiggly, discontinuous boundary between "on-topic" and "off-topic". Without the "why" metadata, we're getting a workable, but very coarse, low-resolution, smoothed boundary between the two. I advocate that a FAQ or guidelines for downvoting is a low-cost method to raise the resolution of that boundary.