Listening to our algorithms

I have a rather eclectic group of friends on Facebook. On any given day, I will see posts for Donald Trump, posts for Hilary Clinton, posts for Bernie Sanders, posts by Christians, posts by atheists, posts by conservatives, posts by liberals, even posts by a few bomb-throwing types (figuratively speaking).

I may be in the minority, and not for any reason of my own.

The algorithm made me do it.

When Gizmodo published a story claiming Facebook was biased against conservatives in how it managed Trending Topics, I thought, “And this is news?” As I read the story, though, it was more serious than I believed. Facebook could have almost singlehandedly turned Black Lives Matter into a major story when it wasn’t one to begin with.

The major newspapers and television networks know this. Facebook is the largest driver of online traffic to their news sites. And in the kind of crazy election year we’ve been experiencing, we really don’t know how much of what is happening is affected by algorithms used by Facebook and other social channels like Twitter, and how much is real.

At first glance, using an algorithm to determine what you do and don’t see in social media sounds technologically unbiased. Except for the fact that algorithms read what you read and like, and they serve you more of it, because that’s what keeps you clicking, that’s what keeps you on the site, and that’s what the social channels sell to advertisers.

So we all like cute puppy and kitten videos, and almost 140 million of us liked Candace Payne putting on a Chewbacca mask. Those are funny and fun, and we’ll keep seeing more of these show up because of that trusty algorithm.

But there’s a darker side. Do you know what kind of news and posts those algorithms really like?

Conspiracy theories.

 

Conspiracy theories abound, and thrive, on social channels. They virtually impossible to kill because they keep showing up, and they keep recycling, and if you keep seeing them (with links!) they must be true.

Vani Hari, aka the Food Babe, discovered this all on her own. She was trying to drawn attention to what she saw as the negative health effects of butter and margarine, and was getting nowhere.

Until she called it Monsanto butter. That tapped into a baker’s dozen of conspiracy theories on social media, and she went from largely ignored to food celebrity almost overnight. It doesn’t matter that nutritionists and scientists see her as misleading at best and downright dangerous at worst; she now has an audience and they will follow her over the cliff, disregarding anything contrary to their (and her) beliefs.

The problem is that we all do this. Given a choice to click on a link that supports what we already believe, and one that contradicts or disagrees with what we believe, guess which one we are more likely to choose?

And the algorithms are watching, and will serve up more of it.

Have you ever been surprised, after looking at a book on Amazon, to discover that same book showing up as a promoted post on your Facebook news feed?

You shouldn’t be. Yes, it’s that algorithm again, supplemented by a little friendly cooperation with advertisers.

It’s one thing if it’s a book. But it’s quite another when it’s an issue, a controversy, or a political belief system.

None of us are abandoning Facebook anytime soon. But we can be aware of what’s happening, seek out alternative viewpoints, and promise ourselves not to accept something as truth, even when “all of our Facebook friends” are saying so.

Related:

Worldviews, Villains, and Decisions

Wither the Media

Top photograph by Kevin Philips via Public Domain Pictures. Photograph of algorithm via Wikipedia. Both used with permission.

Twitter: No Good Deed Goes Unpunished

A mystery has been solved.

For our last poetry jam on Twitter, something odd happened – when I used the hashtag (#tsptry) with posts, my tweets didn’t show up in the hashtag stream.

At first I suspected that something had happened when the new Twitter app developed for TweetSpeak poetry has been installed and launched. Matt Priour, the developer in Texas who did the app with Marcus Goodyear, couldn’t find a problem.

Marcus discovered that my #tsptry tweets were also not showing up in Google searches. And that was the first clue.

Matt contacted Twitter and learned that, according to the current version of the Twitter algorithm, my Twitter account is considered “spammy.” Not exactly spam, but more “spam-like,” because my tweets include a lot of links and a lot of retweets. So I haven’t been kicked off for spam; my followers can still see my tweets; but forget it when it comes to using hashtags or having your tweets available to search engines.

It’s true that my tweets include a lot of links and retweets – by design. I find a lot of good things online – articles, posts, commentaries, poems – that I like to call attention to. And according to the Twitter rules: you’re spamming “if your updates consist mainly of links, and not personal updates.” (The “if” in that statement is significant – it doesn’t mean it will happen, only that it’s possible.)

The Twitter algorithm has decided that doing that is “spammy.” Actually, the people who design the algorithm have decided that. And no one actually examined by tweet stream and applied rational human judgment. Or even human understanding.

I have to keep reminding myself that Twitter’s algorithm is designed by IT people, that it’s constantly being tweaked, that they try to protect Twitter users from spam.

(And I also have to remind myself that it was the most visionary IT I knew who told me (in 1995) that the worldwide web was a flash in the technological pan, that it would soon go the way of eight-track tapes, because the future was Lotus Notes.)

Matt asked Twitter to fix it, but he said it may require me contacting them directly.

Have you ever tried to contact Twitter?

If you’re a developer or a reporter or a policeman, there are special contact email boxes. If you’re a customer, you click on customer support on the contact page, and you get an array of boxes that relate to commonly asked questions.

But no email box. That’s a hint that Twitter thinks it has answered every question possible and so you won’t need to contact them.

But they do provide a mailing address. That’s right, the king (or co-king) of social media provides one way for customers to contact them – and that’s via the post office.

Very retro. But not very cool. Unintended or not, there’s an important message there – that reporters and developers matter more to Twitter than customers.

So I will have to create a new Twitter account for the TweetSpeak poetry jams to be able to use the #tsptry hashtag and have it show up. Which I will do.

I’m also going to be radically reducing the links I use in tweets and the number of retweets – because I could get blocked completely. (One caution – this happened about the time of my first anniversary on Twitter, so be aware if something similar happens to you.)

But my days on Twitter are likely numbered – by my choice. I’m beginning to understand why people are migrating to Facebook exclusively. I beginning to see what several of the people I followed – people who have run afoul of Twitter rules because of how the algorithm was programmed at the time (and was later changed) – rarely post any tweets now. And these were people with vastly larger numbers of followers than I have.

But a social media algorithm is only as good as the human understanding that goes into it.