Sunday, June 16, 2019

Happy Cows

Originally this essay was named “People with PTSD should avoid dealing with Facebook.” When I started writing it last summer, Facebook was threatening to topple Comcast from the Bad Customer Service throne. Don’t ask.

After the effects of customer service trauma wore off, the essay’s working title became “People with PTSD and Facebook should switch to …. What exactly?” 

We choose to connect to people through Facebook’s accommodating platform. Nevertheless, everyone has a story about Facebook’s epic fails and creepy stalker recommendations. 

My biggest head scratcher came after I posted a link to my blog essay “Unfuzzy Things.” To help manage my trichotillomania, previously I'd been fiddling with oversized pipe cleaners in order to distract me from pulling out my hair. At my father’s suggestion, I started using synthetic industrial samples instead. He found the purple Unfuzzy Things while touring a mattress factory in Utah.  

“Unfuzzy Things” included a photo of one of the samples that I took myself at home, rather than downloading an image from the internet. The essay didn’t show a mattress or identify the company. Nothing on my computer or phone did. Nevertheless, the next day Facebook began targeting me with an advertisement for innovative Purple™ mattresses and pillows. Episodes like this test one’s faith in the ability of Ockham’s Razor to explain strange coincidences. Or one’s confidence in Facebook’s purported “privacy” policies.1

1You’ll be shocked to learn that five minutes after I typed that last paragraph on my laptop, Facebook helpfully let me know my friend Goran in Minneapolis likes Purple™ mattresses. Ever since, Facebook has been reminding me about Goran and Purple™ every time I open the app.

Of course, it’s not just Facebook. They’re all equally bad – Apple, Netflix, Amazon, Microsoft, AT&T, etc. For example, whenever Google Images is stumped by one of my search inquiries, it sycophantically suggests random pictures from my own website. That’s just lazy. 

The key to dealing with Facebook is to remember you’re not the target of Facebook’s business model – you’re the company’s productFacebook’s actual customers are paying advertisers. In a popular and apt metaphor, the rest of us are merely a herd of cattle. 

I’ve run the numbers, and I’m pretty happy with our little arrangement.

Facebook and its customers want to harvest two things from the herd.  

Their first goal is to change our behaviour, particularly by motivating us to buy things. However, I’m broke and I hate shopping. Although I'm sure Goran is correct and a Purple™ mattress would change my life, I'm still not going to order one. As far as I can tell, ten years of Facebook marketing efforts have been utterly wasted on me. Unless that’s what inspired me to vote for Donald Trump. 

The other way Facebook makes money is by gathering and selling data about the herd, as well as information about individual cows – I mean “members.” 

Facebook sells two basic kinds of data:  (1) who we know, and (2) what we all “do” – for example, when each of us responds to a particular Facebook post by clicking on a link, or sends a message, or leaves behind a comment and emoji.

So what does Facebook know about me? Mostly that have an interesting and diverse group of friends. According to Facebook, I currently have 359 friends on its platform, all of whom share some actual real world connection with me. (Unlike some of the exhausting extroverts I know, I don’t use Facebook to promote myself, or to set some kind of record for the number of casual contacts.)

My Facebook friends include:

·     Mormons 
·     Ex-Mormons
·     Non-Mormons (Mormons refer to them as “Gentiles”)
·     Jews (who confusingly count as “Gentiles”)
·     Canadians
·     Family members, including extended relations, ex-in-laws, and relatives of my adopted children’s birth parents
·     Lawyers from across the political and professional spectrum
·     Pro-Beard partisans
·     Anti-Facial Hair fanatics
·     People affected by mental illness
·     Musicians
·     People who read my blog
·     People who think my blog posts are too long and confusing
·     Transgender individuals
·     Cisgender men and women
·     Lesbians
·     Avowed heterosexuals
·     Gay men
·     People I’ve made out with2

2SPOILER ALERT: That last group is a small subset of the previous category. Although a Mormon girl I briefly dated at BYU friended me recently.

Many of these categories are shifting and/or overlapping. Needless to say, my Facebook friends are not a representative sample of the general population.

Whenever someone posts something on Facebook, the site’s algorithm decides who should see it. Facebook is constantly trying to guess which shiny bauble will appeal to each ruminating cow, whether it’s an advertisement for adult diapers or another picture of my daughter mugging for the camera.

Mathematicians have a formula for integrating new information into existing probability assumptions. It’s called “Bayes’s Theorem,” after the eighteenth century English clergyman who introduced the concept. As I wrote last year in “Tuning Your Gaydar,” statistics guru Nate Silver explains Bayes's Theorem with the example of “living with a partner and coming home from a business trip to discover a strange pair of underwear in your dresser drawer. You will probably ask yourself: what is the probability that your partner is cheating on you?” Bayes's Theorem tells you how to calculate the changed odds based on what you see when you open the drawer after your next trip.

Or suppose you stick your hand into a haystack and it comes out covered with needles. Then the next time you reach in and grab a handful of nails. Eventually even Facebook’s algorithm is smart enough to figure out this is not your normal haystack.

Information travels both ways, particularly if you’re paying close enough attention to subtle clues like context and body language. 

Facebook receives a little bit of signal and a lot of noise about me. Meanwhile, I’m getting all kinds of information from Facebook. Much of the data is about world events and popular culture; some of the information concerns Facebook itself. 

Mostly, however, Facebook feeds me information about other people. This data trove is particularly useful when you’re trying to figure how to interact with other human beings, but you’re kinda Asberger-y, and have severe social anxiety.

I wrote most of this essay at a park in Vancouver the afternoon before opening night of our 70s concert. I’d already been musing about Facebook data trends, and I decided to go back and summarize recent posts on my timeline. Usually I don’t post to Facebook multiple times in a single day, but I realized there had been an interesting mix of news to report:
·     Earlier that day I watched my son compete in the annual city-wide Grade 5 track meet

·     The night before I’d published pictures from my daughter Eleanor’s Grade 8 choir concert

·     Friday was also my daughter Rosalind’s birthday, so I shared cake photos from over the years

·     After the Vancouver Men’s Chorus performance that night, I flogged our fabulous 70s Show

·     And while writing in the park, I published a selfie of my fresh summer haircut.

Artificial intelligence algorithms like Facebook’s look for patterns in the flood of data they collect. Based on previous experience, they predict how the members of the herd will respond to future stimuli. For example, many of my Facebook friends are fans of pictures and anecdotes about my three adorable children. Facebook has figured out that some of you have a favourite among my kids. 

Facebook also has its own agenda. Last fall Facebook significantly reduced the number of friends who see links to my blog posts, vainly hoping that I would pay to “boost” my profile. (We'll see how Facebook punishes me for posting this essay.)

Certain memes appeal to different arrays of Facebook friends – posts involving Canada, or showtunes, or sports, or progressive politics, etc. Other subjects are obviously less appealing. Apparently only lawyers like legal stuff. 

Some people's eyes are drawn to pictures of me taken in flattering lighting. Mostly my selfies attract the attention of little old ladies. They've always been my most reliable demographic – just like the audience for long-running dramas on CBS. However, I do inspire the occasional thumbs up from members of the gay cohort. Someone should ask me out for coffee sometime. Or introduce me to a single friend. I realize you're no Facebook algorithm, but by now you all should have figured out that I'm never going to do it myself.

In describing trends in responses to particular types of Facebook posts, I don’t mean to characterize any individual friend’s age, gender, or sexual orientation. In fact, I haven’t gone back to examine who actually saw or liked last week’s Facebook photos. I'm sure they would add up to the kind of small, nonrandom sample that a real statistician would warn you not to rely on.

On the other hand, our Facebook encounters typify two of the biggest challenges that face any decision maker: incomplete information, and overdetermined connections. As humans, we are constantly presented with chaotic problems that defy the predictive power of mathematical tools.

As I’ve previously discussed in various blog posts, in Thinking, Fast and Slow, Daniel Kahneman offers a useful model of how our brains rely on two contrasting mental processors, which I've referred to as Thing 1 and Thing 2. Thing 1 is fast and automatic, constantly multi-tasking as it retrieves memories and generates intuitions. In contrast, Thing 2 allocates our brains’ limited conscious attention to effortful mental tasks. Among other things, Thing 1 is responsible for maintaining our brain’s working model of the world around us. Our brains necessarily rely on numerous mental shortcuts. This allows us to smoothly interact with everything from gravity to peer pressure without our heads exploding.

Because we’re profoundly social animals, our unconscious mind is particularly interested in evaluating each of the people we encounter. Are they members of our tribe or outsiders? Potential threats or lovers? As a result, my brain automatically generates all kinds of social expectations – including assumptions about the types of people Facebook will select to see the next picture of my family or the beach, and how particular individuals might respond to specific stimuli.

Meanwhile, everyone else on social media is unconsciously generating their own working models of the world.

In addition to being social animals, humans are uniquely intelligent. After millions of years as just another primate, humans became the dominant species on the planet a few thousand years ago. According to evolutionary psychologist Steven Pinker, this happened when we developed the skill of making effective but “fallible guesses from fragmentary information.” Our brains solve each “unsolvable problem by a leap of faith about how the world works, by making assumptions that are indispensable but indefensible – the only defense being that the assumptions worked well enough in the world of our ancestors.”

Pinker’s brain model is consistent with Bayesian probability, as well as with old-fashioned American pragmatism, which teaches you to keep an open mind and apply reason to the best evidence you can find. My favourite literary detective Nero Wolfe sums up this approach when he sends his gofer Archie Goodwin out into the world: “Act in the light of experience as guided by intelligence.”   

Life unfolds as a series of decisions based on overwhelming yet incomplete data. Facebook spends billions of dollars trying to anticipate and duplicate human decision-making. But human brains still are much better at this process than any of the artificial intelligence algorithms generated by Facebook or the other rapacious monopolists. In fact, humans are smarter than any other species we’re aware of in the 14 billion-year history of the universe. Including cows.

No comments:

Post a Comment