How does one have the gall to make this kind of claim with such a flawed methodology? > Out of the top 3,479 profiles, the visual recognition API classified 1,

How does one have the gall to make this kind of claim with such a flawed methodology? > Out of the top 3,479 profiles, the visual recognition API classified 1,515 profile images with a discernible gender. 3,479 seems to be an arbitrary number and no explanation is given as to why that was number was chosen. But, more importantly, more than half of these top users, who could be of either gender, were eliminated from the results, which already makes this study inconclusive. The second major flaw is that the author determined a user's gender by their profile image, using a "masculine" or "feminine" classification. If already less than half of user profile images had a discernible gender, who's to say that, of those images that did have a discernible gender, the profile image accurately represents the user's gender? What if I decided to set a celebrity's image as my profile image? If my profile had Melissa Joan Hart as the image, does that make me female? If my profile image was of Bryan Cranston, does that make me male? Also, how is "feminine" and "masculine" determined by this API? Does having long hair qualify as a feminine trait? John Romero has long, flowing hair and painted fingernails; does that make him female? Conversely, if a female has short hair does that means she's categorized as part of the "masculine" group? The title to the article is absolutely misleading and wildly inaccurate. Even after the author admits: > Determining gender on Stack Overflow turned out to be trickier than we initially anticipated. She still has the nerve to make the claim that there is only one woman in the top 100 SO accounts in the title and throughout the article.