Blue button pressers are more into incest

I seem to have accidentally founded an active social media site. It’s still small and early, but has around 600 daily users. I designed it ultimately for data analysis; I wanna gamify data collection so that I - and other users - can test hypotheses! It requires a lot of investment - you take quizzes, rate your friends, match with people, get personalized results, and most recently get to graffiti other people’s avatars - it’s chaos and it’s amazing.

(If you want access, it’s invite-only and paid subscribers can access an invite code at the bottom of the post here)

So when the button question started trending on twitter, I couldn’t help myself. I found myself the deity of a world that people had begun to care about. I could test how people would behave when something was actually on the line. And thus:

When logging onto Glosso, users got this page. In order to access Glosso, you had to answer. I threatened people with bans if they simply closed the page. I gave this only to users who’d already signed up for Glosso, so nobody could make another account just to interfere with results.

I shoulda thought it through, because it hurled Glosso into a ton of drama. Some people got mad at me, people got mad at each other, people mourned, families were torn apart.

After about 36 hours, the big flashing timer on the page finally ran out, and users were delighted(?) to find that blue had won. Everyone was saved. No account bans. We all cheered in the little impromptu chat room I pulled up for us.

Despite many people in the debates over the last day claiming they had voted differently from the twitter poll now that they had actual accounts on the line, the results were…

Exactly the same as Tim Urban’s poll. A perfect 42/58 split. I have no idea why. Tim’s poll went extremely viral across all of twitter; my Glosso test was a narrow set of users selected for being active on a weird experiment quiz site. ?????

One benefit of my site is that I can correlate everything across everything. Glosso is loaded with questions, and roughly half of people who answered my button question had answered a decent set of questions elsewhere. So let’s investigate!

(you should know that David Shor did some much bigger polling about this and got some pretty cool, likely more robust correlations than me here and you should check it out)

Here’s the top correlations with the button, that are probably robust-ish despite the lots-of-testing I did (we have 1300+ questions with n>300, this can lead to false positives).

Okay first - obvious things stick out. Stuff associated with my MTG color test (the glosso version) white are most strongly correlated with blue pushers - communal, weak-protectors, etc. Interestingly, economic liberalism was slightly more strongly associated than social liberalism, though this might be noise.

There’s also a nature theme that surprised me; not only spending time outside as a kid, but also finding nature erotic?

I did control by gender, otherwise I’d assume it’s a gender confound.

Gender breakdown, by the way, was this:

Okay so wtf is going on with the incest questions?? I can’t think of any obvious confounder here. Gender isn’t it, and liberalness as a confounder would primarily bring up fetishes besides incest (genderplay is more strongly associated, for example).

And… gigolos?? I don’t know. It might be noise, but upon light checking I think these associations are reasonably strong enough to make it unlikely to be noise, and two incest questions is a lot to show up by accident.

Probably this is confounded by something. But I don’t know what. If the sample sizes increase, I’ll let you know.

In general, I think people parse the blue/red button question without doing much game theory about it, as evidenced how the correlations trend towards ‘nice group thinker’ vs ‘evil iq believer’. I myself am a blue button pusher, which is an unpopular opinion in much of my closer culture.

How did you vote? Are you into incest? Fight in the comments.