I'm guessing you don't have this data, but we'd be particularly interested in knowing whether the situation changes when you check for age. We ask because a discussion centred around your findings raised the idea that perhaps technical users who joined BBS/Internet/Online communities earlier (80s, 90s) tend towards abstract names (less gender-identifiable), partly as a result of the culture at the time and partly due to various bizarre login/uniqueness/length restrictions on usernames.

Once those usernames came to be part of their identity, we believe this grouping tends to re-use it on new sites, possibly explaining the (odd?) bias against gendered outsiders - rather than being against gendered outsiders it may be a bias for older, more experienced developers who are more likely to be re-using an abstract identity.