By Mason Rosse
On free social media apps, there is no avoiding the fact that companies still need to make money somehow. To keep these apps free, the companies offer advertising space for businesses to buy. Sharing your ad on an app used by millions can get you some very valuable air time. But these ads won’t be shared with just anyone– the social media app’s algorithm will tailor each ad so that it goes to a viewer that will “like” it.
This is called a “black box” algorithm. We use the words “black box” because we don’t understand the formula; we know what goes in and what comes out, but not the mechanism that makes it work. The terms for what goes in and goes out are called the inputs and outputs, so that is what they will be referred to as for the rest of this article. The output is the ad (or the post, in some cases). Instagram, for example, takes your “stats,” or input, and uses it to direct ads that you will like towards you. So if you like a post with dogs in it and comment on it “awww”, Instagram will be more likely to give you ads and posts of dogs. It will do the same if you like a certain meme. The black box takes that like as an input and gives you more, similar posts as an output. The problem is that the input is your information. It will take every bit of the information you put into the app: your age, area code, name, etc. and then put it into the black-box algorithm.
This becomes problematic when you hit “accept Instagram use of the microphone,” because it will use that too. People are always scared of their phones listening to them, but now it has become a reality. Instagram has never confirmed that they have used these black-box algorithms, but there is enough anecdotal evidence to verify that they do. Once the microphone is in play, it starts to make the user unsettled. Have you ever talked about something and then instantly have gotten an ad of the same thing? You’re not crazy for thinking this. The other day I was talking about wanting S’mores, and instantly a Hershey’s ad popped up for their “perfect S’mores”. I found this weird because I had not searched anything about Hershey’s, I had not liked any pictures of S’mores, I only said the word “S’mores”. I obviously was taken aback by the fact that Instagram recorded my voice as an input.
There are different strategies that people are using to find out about these algorithms, but only one strategy is really working. The only one that works is through our own studies. This is explained in an article by Vice, they interview Dr. Peter Hannay, a cybersecurity worker. He explains why these companies are using these algorithms: “Really, there’s no reason they wouldn’t be. It makes good sense from a marketing standpoint, and their end-use agreements and the law both allow it…” This is important because it means that the only way we will learn about it is through our own studies. Because we can’t catch them in law. Now, there are people doing studies in an attempt to find what happens in these black-box algorithms, but finding what happens somewhere only by knowing what it does and what is uses is difficult.
Facebook is different than Instagram, in that it is not listening to you. It actually doesn’t have to. For one, Facebook owns Instagram, so it can get some of the information from that. Second, it has an algorithm based on all of the information that you put into it, then it guesses what you will want to see.
The thing about black-box algorithms is that no one knows what they are actually doing. That’s the whole point of them. Most of the evidence used to talk about these algorithms is based entirely on experience. That makes it hard to explain, and even harder to argue about. Some people like a personalized experience, and some people like their privacy. These algorithms won’t be retired soon, so it’s up to you what information you give it.
Sources
Vice News “Your phone is listening to you and it’s not Paranoia”
Vox News on Facebook and Instagram Ads
Wired “Facebook’s not listening through your phone, it doesn’t have to”