Okay: so let's look at the other options:

#2 is a critical mass problem - "if a site has a critical mass of women, it will be used by women". Looking at the way that some social networks and other websites' communities work, I can see why this might be the case. Sexism against women on the web is endemic, and this makes many online communities unappealing to women (men, on the other hand, will sometimes overlook it; so men continue to join and women don't). If you look at the online communities where this is not the case, they are either communities in which a critical mass of women already joined (e.g. Pinterest), and thus are self-regulating, or where community membership is "opt-in", rather than "opt-out": Facebook and Twitter, for example, lean towards a model whereby you socialise first and foremost with people you already know (or, at least, people you select) - compare to Reddit, for example, where your very first experience is to be bombarded with the most popular stream of the firehose.

How about #3: there's something about the format itself that makes women more-likely to use it? Maybe, and I've certainly had female friends say this to me (of Pinterest), but I'm not certain that there isn't some confusion between this and the previous point: a society is built by its members, and - again - the first thing you see on the front page of Pinterest is the popular shares, which as we all know are dominated by women. This isn't a reflection on the format of the site, but of its makeup: i.e. option #2. Sure, there are aspects of Pinterest's design that meet aspects of stereotypical "girly" design (the floral logo, for example), but I don't think that these are significant in themselves.

In short: I suspect that your second suggestion, that a core critical mass of women is the fundamental force that shifts a website to having an ongoing female presence.