Alright, so I had this crazy idea the other day. It was about figuring out, programmatically, who’s, you know, gay in an interview setting. Sounds wild, right? I know, I know, it’s ethically super questionable, and honestly, probably impossible, but I was curious to see what I could even try.
First, I spent some time thinking about what data I could even use. Obviously, you can’t just ask someone that directly, and even if you could, this whole thing is about trying to figure it out indirectly. So, I figured I’d focus on the interview itself.
I started by watching a bunch of random interview videos online. Nothing specific, just people talking about their jobs, their hobbies, whatever. Just to get a feel for how people present themselves. I wasn’t looking for anything in particular, just trying to soak it all in.
Then I started to get a little more… focused, I guess. I searched for interviews with people who are openly gay, and interviews with people who are, as far as I could tell, straight. This was all based on public information, of course – what they’ve said about themselves, what they’ve written, etc. I wasn’t digging into anyone’s private life.
My Messy Process
- Watched tons of interview videos, both general and with people of known orientations.
- Tried to identify any patterns in speech, body language, or topic choice. Totally subjective, I know.
- Read some articles and studies (superficial ones, mind you) about nonverbal communication and “tells.”
- Considered using some kind of voice analysis software, but quickly realized that was way beyond my skills (and probably super inaccurate anyway).
- Tried create a basic “gay” detection model.
Then came the “coding” part, which was honestly pretty pathetic. I tried to put together some kind of super basic checklist of things I thought might be indicators. Things like, I don’t know, vocal inflection, use of certain words, maybe even posture? It was all completely subjective and based on my own limited observations and biases, which I fully acknowledge.
I basically ended up with a really crude scoring system. Like, “+1 point for using the word ‘fabulous’,” “-1 point for talking about sports.” Seriously, it was that ridiculous. I knew it was garbage, but I was just messing around.
I ran a few of the interview videos through my “system,” and unsurprisingly, the results were completely random. Sometimes it would “flag” someone who was openly gay, sometimes it would “flag” someone who was straight, and most of the time it just gave nonsense results.
Ultimately, I concluded that this whole thing was a fool’s errand. There’s no way to reliably determine someone’s sexual orientation from an interview, and even trying to do so is problematic. It was a fun (and slightly embarrassing) thought experiment, but definitely not something I’d ever try to apply in real life. It just reinforces that you can’t judge a book by its cover, and you shouldn’t try to pigeonhole people based on superficial observations.
I quickly abandoned the project. I * most important takeaway? Don’t try this at home, folks. It’s not only impossible, but also pretty disrespectful.