Making the Internet a Better, Safer Space

So anyway, I have this live storytelling show in Portland, and our primary paid advertising platform for selling tickets is Facebook.

Making ads for Facebook can be frustrating, I have learned, because they have a very long list of things they will and will not allow you to do and say in your ads — but that list is not available to advertisers. You basically make your ad and cross your fingers that it isn’t pulled for inappropriate content.

It turns out I am very bad at this.

In June we had a series of ads for an all-woman show pulled because they contained the phrase “five kick-ass women.” In July we had an ad pulled because it used the old Mad-magazine trick of using “%&@?” to stand in for a cuss word. Last month we had an ad that had pictures of that show’s five artists in it, and one of the pictures was this picture of a local radio icon dressed as Catwoman

and this photo was deemed to overtly sexual. Earlier this month we had an ad pulled because we we quoted a storyteller as saying “hamsters be funking” and noting that the word used wasn’t really “funking.” Just this morning we had an ad pulled for “inappropriate content.” Here is that ad:

 

We are still awaiting an explanation for what exactly about this ad Facebook finds inappropriate.

So imagine my surprise to learn this evening that all we ever needed to do to have our ads approved was to focus less on our show and its artists and more on hating and burning Jews.

Please do be so kind as to share this post.
TwitterFacebookRedditEmailPrintFriendlyMore options

24 thoughts on “Making the Internet a Better, Safer Space

  1. I have to imagine that a lot of this filtering is being done by machines. And the important thing to remember about machines is that they’re really stupid.

    I wish Valve, Google and Facebook would also remember this.

    Report

    • I wish Valve, Google and Facebook would also remember this.

      And then what? The reason they use machine learning to do this stuff is that it would be prohibitively expensive to pay humans to filter the vast oceans of user-generated content that these companies deal with. I’m sure they know that their algorithms are imperfect, which is why they’re constantly working on improving them.

      It seems to me that the least-cost avoiders here are the people who get offended at this stuff and treat every unintended consequence of a computer algorithm as some kind of scandal. Filtering user-generated content is an extremely difficult problem that isn’t fully solved yet, and may never be. People just need to chill out and cut them some slack on this.

      Report

      • You’re not wrong, the cost of human-filtered systems is likely prohibitive in a lot of cases. I’m not sure what thew answer is (in Valve’s case at least I think there are some automatic things they could do to improve their game selection) but we need to keep the limitations of the systems we use in mind in case opportunities come up to improve them.

        Report

  2. So imagine my surprise to learn this evening that all we ever needed to do to have our ads approved was to focus less on our show and its artists and more on hating and burning Jews.

    I’m sure you don’t mean this literally, but I’m not sure that you really appreciate how inapt this comparison is. Here’s what I suspect is actually happening. The ads that you place are likely screened by humans who have a checklist of things they’re supposed to look for.

    In contrast, the linked story is about is an algorithmically generated list of terms you can pay to associate ads with. It’s likely that no human at Facebook ever actually screened this list. Note that Pro Publica did not actually attempt to place ads about burning Jews. If they had, it would have been rejected by the people who screen the ads.

    It’s entirely possible that the people who screen the ads don’t even see the terms the ads are to be associated with. That is, they might have seen the innocuous ad Pro Publica was attempting to place, but not that they were attempting to target antisemites. And if they did, is that even a problem? What’s wrong with targeting an ad to antisemites, if the ad itself is not offensive? Maybe someone wants to try to change their minds.

    Report

  3. Well, Tod,
    If you’re advertising medium is giving you trouble with vague guidelines and pulling your adds for apparently no reason, maybe you should consider dropping them.

    I’m no on FB much and never have seen much ads, but I do get regular spam from Capital Fringe in DC about their work (There kinda edgy theatre guys). They got my email when I used it to buy some tickets. Why don’t you try that method

    Report

    • — Facebook is pretty hard to avoid cuz all kinds of network effects. The thing is, no matter how bad they are, you will lose money if you ignore them, and losing money “on principle” is seldom a good idea for a business. So, we put up with Facebook.

      Report

        • — But you have to get everyone to jump over at once, or at least in sufficient numbers. Believe me, we’ve tried a few times to get transbook to relocate en mass. However, it just doesn’t work, because not enough people go and too many remain on FB, which renders FB still the best place to meet cool trans folks — even tho FB treats us like garbage. It sucks, but large groups follow a logic quite different from what any member of the group actually wants.

          Report

            • — Probably, but the point is, it’s not about one person’s rational choice. It’s about an unpredictable change in mass behavior. Plus, Facebook was presumably different from Myspace in just the right ways. So anything that is going to beat Facebook will be unpredictably different. It’s not simply a matter of recognizing that FB management sucks, nor that we can do “the same thing but better.” That’s not how it works. (Unless it is, of course. Being unpredictable means precisely that I don’t know what will happen.)

              Also note that, while FB replaced Myspace, you can certainly find people who miss Myspace. It might be the case that FB falls to some new thing, but that new thing is actually worse for individuals, cuz everything is terrible and entropy always wins.

              Report

              • If you ask me, the whole social media enterprise is kind of deeply suspect.

                There’s this dynamic where it’s not valuable because nobody is on it, then it’s good, then it’s worthless because too many people are on it, and there’s context collapse, and you can’t say anything useful or entertaining at all.

                Report

  4. I’ve seen a bunch of these stories about Facebook’s tools and policies and have been thoroughly unimpressed. Seems like a lot of click bait. There was that story with the headline, “Facebook’s Secret Censorship Rules Protect White Men From Hate Speech But Not Black Children.” Which, despite implying something much more lurid, really just lets us know that FB’s policies allow you to make pronouncements about sub groups of people buy not overly broad categories. The headline could have read, “Facebook’s Secret Censorship Rules Protect Black Men From Hate Speech But Not White Children.” … or white flamenco dancers or Mexican haberdashers or any other combination of people not part of a legally protected category.

    With the advertising stuff, it just seems that Facebook has a great big algorithm driven by machine learning that can be hacked to produce off results of you try enough odd combinations. Essentially what’s happening is ProPublica and various other outlets are debugging Facebook’s algorithms for no pay from Facebook. Instead, these outlets are taking what they find and using it to drive traffic to their stories. As a symbiotic economic model, maybe it makes some sense, but I can’t quite see the journalistic value.

    Report

  5. A lot of the filtering definitely raises my eyebrows. As in “you filtered for ‘%&@’?” Really?

    It would be nice for the whole thing to be a bit more predictable.

    Report

  6. My g/f got suspended last month for posting tiddies (hers), even though FB’s current “official” position is that female breasts are okay. (Perhaps trans gal breasts are not. I dunno. Facebook is a shitshow.) Then they got upset with me cuz I posted this link (perhaps a bit NSFW), and their algorithm auto-included the picture from that article. So, I guess Wikipedia is now a hotbed of queer porn, or something.

    In any case, Facebook seems to do a terrible job filtering stuff.

    That catwoman pic is not remotely sexual, like wtf? This pic, by contrast, is just wonderful.

    Report

    • Good lord, that link. It’s like all the cliches at the same time.

      And, it’s not an advertisement, so it probably goes through an entirely different set of filters. Just an observation, not an endorsement.

      Report

    • Then they got upset with me cuz I posted this link (perhaps a bit NSFW), and their algorithm auto-included the picture from that article.

      Whenever I paste a link into Facebook, I remind myself to wait several seconds while they go get it so I can see what dumb description and image they will grab, so if I don’t like it I can disable the link preview.

      It is extremely dumb that, if you are fast enough and beat Facebook pulling it in, you can make a post with a link in it, and then end up with both an image _and a bunch of text_ that you didn’t type and didn’t realize were going to be there.

      That catwoman pic is not remotely sexual, like wtf?

      I am finding myself completely confused by this concept, because I am getting a lot of Dragon Con cosplay image posts in my feed currently (It ended two weeks ago.), and that Catwoman outfit is like a 4 if 10 is full frontal nudity…and I’m getting a lot of 7s and some 8s, no problem at all.

      Like that Tank Girl outfit, but in real life. (Sadly, there is no actual Tank Girl cosplay I have seen.) Really, some past that.

      I actually thought images had to be reported for Facebook to look at them?

      Report

    • That catwoman pic is not remotely sexual,

      Nothing sexy about it at all.

      Tod, can you send me all the other versions of that photo, so I can, you know, not find them of any interest?

      Report

  7. I think James and Brandon have it right. Machines are very dumb but they are a lot cheaper than humans doing it and making value judgments via time and discussion. I’m Jewish and subscribe to some Jewish groups on Facebook. There was a time when facebook would show me an article from Tablet or some other Jewish source and down below were suggestions that were blatantly out of the Neo-Nazi homepage.

    Semi-OT but am I the only person who really dislikes the term kick-ass.

    Report

  8. our primary paid advertising platform for selling tickets is Facebook.

    Curious to know what comparisons were done to other paid advertising platforms. How do you know that Facebook is worth this trouble in addition to the $$$?

    Report

Comments are closed.