Draft talk:Gay face

Gaydar

Would this page be better as a separate section on the gaydar page? Bongolian (talk) 18:42, 28 April 2020 (UTC)

Would be inclined to agree, yea. --85.76.79.210 (talk) 18:56, 28 April 2020 (UTC)

I'm not sure it even warrants that. I've never heard of the term before and Google gives me urban dictionary and numerous links to stories a single AI program claiming it can determine sexual orientation.

Whose been using this term? Or even claiming they can tell a gay from just their face? As I have said, Ive not personally heard anything like, except the usual gaydar style claims, and I can't find much in the way of anything referring to gayface. Urban dictionary is not exactly the best source for its usage. AMassiveGay (talk) 14:07, 4 May 2020 (UTC)

AMassiveGay, Bongolian personally I think it deserves it's own place. The AI stuff blew up in the media in 2017 after Michal Kosinksi (Stanford) put thousands of gay faces through an AI bot and it could detect with 80-90% accuracy. However he was criticized for methodology, as the AI was picking up on other subtle clues like lighting/hair etc. As I pointed out in a previous edit, AI claims are disputed. However what is clear from Nick Rule's research (U of Toronto I believe), is that gay face does exist. He's been doing it for over 15 years with multiple researchers, and they've done 20+ separate research papers on it. So it isn't just style claims. His research shows people can accurately detect sexual orientation from eye and mouth regions alone, conservatives are better at guessing orientation, pregnant women can better detect sexual orientation. All of his work has involved cropping out hairstyles/piercings/lighting and ensuring bias is as minimal as possible. The article could easily be renamed to face shape and sexual orientation, or something of the sort. The point is that face shape is so distinct from stereotypical clothing/behavior that typically are referred to when we say gaydar. Perhaps let me expand the article first, and if it isn't convincing we can put it in gaydar?

• This is a 2009 article about some of Rule's early research into gay face
• Another 2014 one
• There are plenty of other papers out of this lab at UofT

Disclaimer I don't have any relation to them I just find it very interesting. (If I did I would probably have written a much more cohesive article by now lol). I have also been in contact with Michal Kosinski before, and his work was actually backed up by a follow up AI bot which minimised bias and still had success. AI research isn't even relevant. It's the fact that people can guess orientation when a deadpan face with everything else is cropped out. You can simply scroll through photos of 'ex gays' and see the gayness emanating from their faces. Some people are better at seeing it than others.

--Debunky (talk) 03:11, 5 May 2020 (UTC)

I have no objection to you making your case for a separate page by working on the page in draftspace. Bongolian (talk) 03:36, 5 May 2020 (UTC)

Thanks, I won't be shifting pages to main space in future without checking with mods first. If it's still not convincing we can probably fit it under gaydar and expand the other elements on that page. --Debunky (talk) 03:53, 5 May 2020 (UTC)

Just wanted to add that I did create this draft on the back of volume of searches for the term. Ahrefs data indicates 'Gay face' is searched for about 2,400 times per month globally and there aren't too many many results covering the topic in a rational manner. Gaydar is a more competitive search term. --Debunky (talk) 04:20, 5 May 2020 (UTC)

Mods aren't really the arbiters for that. You can ask your fellow sysops. Which also includes mods. Bongolian (talk) 06:42, 5 May 2020 (UTC)