Hello, tumblr! I saw something on here the other day that worried me, so I decided to Do Science about it. But I can't do it alone: I need your help to build the dataset!
Here's what I need you to do:
If you see a post with a "mature content" label, and it's 2026, DM me a link to the post.
Yes, that's really it.
I am hoping to collect several thousand such posts, so that I have a decent sized dataset. I do not care what the post is about; if it's labeled as "mature content", I want to add it to my dataset.
If I get 10,000 posts in my dataset before August 31st 2026, I will post my preliminary findings then. I won't feel comfortable calling my findings "settled" before 2027, unless I get over 50,000 posts.
Tumblr Science FAQ, round 1
What's your hypothesis, OP?
I am not talking about that unless I have results to share. That would bias the results.
I did write them down and I did share them with a trusted contact who can prove that I wrote them down the same day I made this post. (While I did so before I made this post, I am not sure they will be able to provide proof of that, because I did so on the same day.)
OP are you interested in...?
Do you have to click through to see the post? Does the clickthrough contain the words "mature content"? Then yes.
OP are you interested if the post is about...?
I am interested in the mature content labels, not the content of the post. Is there a clickthrough that contains the words "mature content"? Then yes.
Tumblr Science FAQ, round 2
Should I reblog this for reach?
Yes, please. I felt really silly when I noticed I forgot to include that in the original post.
Is it okay if I send you my own posts?
Yes, those are perfect for what I'm looking for. I actually need to do some processing on a post to make it useful for testing my hypothesis, and this makes it easy.
Are you looking for "potentially mature content" also, or just "mature content"?
I want both, please. Anything that throws up a blocking screen that you have to click through. The distinctions between them are one of the things I am hoping to study.
Does it matter when the post was made originally?
Technically no. There's no way to respond to this without introducing some bias in the results, and I don't want that. However, I do collect some data on a post as part of making it useful to me, and that data is easier to collect if the post is recent.
What if I request content label review on a post after I send it to you?
I need to see the mature content label to be able to use the post. Because the mature content label hides the content of the post, it is very hard to use a post that no longer has a mature content label. You could send me a screenshot, but people could use that to lie to me.
Basically, it's more work for you to make it usable to me.
OP are you a transphobe? It would ruin the experiment for me if you're a transphobe.
I promise I'm not a transphobe and not doing this for transphobic reasons. You should still double-check that I'm not a transphobe for yourself, though. I am not sure that this study will have the useful effects you're hoping for; I am studying something specific, and it may not be what you hope.
hey OP, I've been reblogging the stuff that's mislabelled. it's not all 2026 original posts, but all have appeared on my dash recently, I'm not digging for these they are in circulation. of course these don't include actual mature posts, it's just incorrect ones. if that would skew your data, don't look!
also anyone seeing this through the post notes: please feel free to submit posts to this blog when they're mislabelled as mature content. I'm sort of building a collection to see how frequently (if ever) they get fixed, and also as a way of noting the scope of things the assumed bot mislabels
