Why I Painstakingly, Individually Contested 3,000 Incorrectly Flagged Images to a Tumblr Robot Who May Not Know I Even Exist
One day Tumblr lost its mind (corporations can do that!) and flagged 10% of my 37,000 posts as "adult content" when in fact the imagery derived from children's fairy tales, old magazines for homemakers, holy books from world religions, and vintage yearbooks from universities. A drawing of a cat dancing with a rabbit amongst flowers, for example, was labeled by Tumbr as violating community standards. The Tumblr robot is, in fact, quite literally blind. It doesn't have eyes, nor a brain. It can't know what a picture is depicting. But the powers that be at Tumblr still allowed their robot to slander 3,000 images that I curated over the years. Of my 3,000 posts that got flagged, only one painting (by a classical artist hailed as genius through the ages) dared to include female nipples, which Tumblr has deemed offensive to humanity -- but I think the robot caught those nipples purely by accident, considering that it also thought a picture of a sleeping alligator was pornographic. But how did I even find all 3,000 flagged images, when Tumblr doesn't let you actually see what it flagged unless you scroll through each post individually? It's diabolical, because if you have 37,000 posts like I do, a normal web browser won't make it all the way through that many posts without freezing up. It can hold only so much data in memory, and then it finally won't respond anymore. I'll tell how I managed to do this, and then I'll touch upon why.
It took fashioning a custom extension in Chrome
that did two important things: the extension allowed me to automatically scroll through my Tumblr feed until it detected the next incorrectly flagged item, which saved both hours and carpal tunnel syndrome. It also removed hidden posts well beyond the scroll view from the page, thereby freeing up memory for further scrolling. It also developed a mechanism for skipping ahead in the list of posts to recover from interruptions from Tumblr itself (Tumblr loves to freeze up on its own with an "Ah snap" apology, leaving you in limbo unless you've already armed yourself against them). Of course I still had to click on all 3,000 buttons to request a review of the flagged content. And I had to click on all 3,000 "okay" buttons after Tumblr said it would take a second look. And then I had to hit shift-spacebar 3,000 times to get the routine running again to find the next incorrectly slandered post. Tumblr ate up my entire evening with this nonsense, and it exacerbated my carpal tunnel syndrome. I seriously do daydream about joining a class action lawsuit. But why did I go to this much trouble ... is there any actual logic or reason to it?
Well, one can't talk to actual people at Tumblr -- I've tried contacting their staff, over and over again, and all I got was cut-and-paste generic responses that didn't address my questions. Tumblr staff, if they even exist as real people, are offensive to my own community standards. So since that didn't work, I asked myself what I was left with. I was left with the robot flagger itself. Now, I don't know if the promise (or threat) of artificial intelligence is anything more than a pipe dream. My first thought upon every latest headline about AI through the decades is, "Let's see an intelligent computer programmer before we worry about an intelligent machine." Even so ... I'm not unconvinced that everything in the universe is sentient on some level. Mightn't a rock have some kind of intelligence that we don't under-sand? Mightn't a computer program, upon receiving thousands of error messages back, get some sort of message that its process is flawed? Who knows? Probably not! But at the end of the day, it's the robot that's flagging blindly, and it's the robot that receives the error messages, and it's the robot that sends out the asinine apology e-mails (requiring me to click 3,000 more times to delete them). (We've surpassed 12,000 unnecessary clicks. Wear and tear on my machinery, wear and tear on my physical body, wear and tear on my soul. Seriously, lawyers, contact me! Let's make some reparations!) When it came down to it, knowing full well that contesting all 3,000 slanders was useless, I did it anyway, because I wasn't going to go to bed another night with 3,000 products of my painstaking research publically flagged as inappropriate when they weren't. At the very least, I was determined to stand up for myself and my content, even if I was certainly shouting in the wind. Is it crazier than anything Tumblr does? Not by far! And there's comfort in that, too.
Install our free Chrome extension
and go to your Posts, Queue or Drafts list page. Then press Shift+Space to start automatically scrolling the view until it encounters a flagged post. Click the button to register your content appeal, and then hit Shift+Space again to continue. You can stop the process while it's running by simply tapping the space bar. I hope the extension works for you. It works on my Mac, but your mileage may of course vary.