Hacker News new | past | comments | ask | show | jobs | submit login

Well... how do you think Google Analytics guesses whether or not a person is male or female? It uses a machine learning algorithm based on browsing habits, and "views Youtube videos about technology/open source/Raspberry Pi" is a feature that is strongly weighted towards maleness.



I'm a woman in tech. Google used to have a page somewhere where it told you what it assumed about you from your habits. It was 100% confident that I was male until I started searching for pregnancy stuff all the time.


Also a woman in tech. A while back I checked my profile. It thought I we was a man.


Sure, but in many cases it didn't have to guess. It knows. Moreover, even when it had to guess, the guess can sway both ways based on a person's watch history too: Wouldn't you think a person who watches a raspberry pi video after they watched 10 videos about handbags and jewelry and mascara can be guessed as a female? After all, Google IDs unique people and devices.


There are millions of women out there that don’t watch videos of handbags and jewelry and mascara, as there are millions of men who don’t watch sports or cars or whatever.


I'm currently running a forum where people has to select it's gender. I'd say, by the samples I've done, that most of the people don't lie.

And the stats are not surprising, I have about half females, and they participate way more in "girly" things.

The same happened in another forum I had.


Do you know that's how it works, or is it a guess? That's not really the standard way to do audience segmentation.


Valid point. Although tons of people have both GMAil and Youtube accounts and Google forced you to connect them a few years ago when they were pushing Google+, so they should have actual personal information on many people these days.


If you're logged into your Google account when you watch a Youtube video, I'm guessing they don't have that much uncertainty in their assumption of your gender.




Join us for AI Startup School this June 16-17 in San Francisco!

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: