Wednesday, October 18, 2017

The End of Coincidence 

“Courage, it couldn't come at a worse time”

This morning I wanted to post a “Happy Diwali” image on behalf of the advocacy group I work with as a shout-out to Toronto's large South Asian community. An appropriate photo full of candles carefully placed looked interesting and when I selected it I was directed to the site for Getty Images. Scrolling down to see if there were other options I was shown not more images for the Indian festival, but images of trending topics. That’s when I saw an image of a young Gord Downie in performance with the caption, “Tragically Hip frontman, Gord Downie has died at the age of 53.”

“Oh. How strange?” I thought. Downie had been ill for some time and though his passing wasn’t unexpected it was still unsettling. Yet I was really curious as to how Downie’s death had “trended’ on a stock imagery site? Did the announcement of the musician’s death cause journalists to rush to their laptops, search their Getty Images account for Gord Downie thus causing that search to trend on the site? That seemed the likely answer, but still a weird way to hear the news.

It was a warm evening and I decided to take advantage of the weather and go for a run. At Dundas and Broadview a truck passed with the Tragically Hip’s “Courage” emanating from the driver side window. At that point I needed some courage to keep running and I wondered about the coincidence of hearing one of the band’s signature songs on the day of their lead singer’s death. Of course, it wasn’t a coincidence at all. Every radio currently playing in Toronto at that moment was probably playing any one of two dozen popular Hip songs in memoriam for the fallen frontman. In fact, after I got home and showered, I went to the kitchen to start supper and I turned on the radio and CBC was playing a special tribute to Downie, so I heard more of his songs. When I think of it, I recall when Elvis died and we were camping in Grand Falls, every radio from every campsite was playing nothing but Elvis tunes.

All of this made me think that seeing an image of Gord Downie or hearing his music at a certain time would never be a coincidence and in fact we may have reached the end of any such coincidences. After reading The Improbability Principle by David Hand, you start to understand that in this world, what seems improbable is surprisingly likely. When I see trending images in my browser, you’re seeing different ones. When I see ads on a news site, you see different ones. The scroll of posts on my Facebook page are uniquely tailored for me. All of this is because we are in the age of the algorithm. Our devices that we use daily aren’t just the most intimate objects we own — we hold them so close and caress them so often — but they are also cheating on us.

They know who you are talking to, what you are saying and where you are saying it. They know what apps you use, what street you live on, what teams you cheer or jeer for, which films you like and which ones you don’t, what music you listen to and when you listen to it. They know what books you’ve read or are about to read. What TV shows you watch and who you would most likely vote off the island or in an election. They know when you get out of bed and when you get in it, and how many steps you took in between. They know where you’ve travelled, they’ve seen all of your holiday pics, and what crap you bought when you were there and they have the receipts. All of it, every byte, makes a data version of you which is like the radiating cloud of dust around the Charlie Brown character, Pig-Pen. And it’s not just making data about you, but data about the data about you – that’s the metadata.

Your meta-self is floating in the ether of the Internet between services and servers and it all goes into a sausage machine called an algorithm. That algorithm pushes out a sausage that might be an ad for a winter coat (you live in Canada, it is October and you used the word ‘coat’ four times this week), an ad for a hepatitis vaccine (you did just buy a ticket to Costa Rica, right?), or it might be an offer for a hotel or news alerts about an impending hurricane in Central America. The data sausage takes many forms: your Facebook news feed, your Google search results, your Amazon recommendations, the rental movie in On Demand that just became available, music and movie recommendations, recipes based on food you buy, or a charity you once thought you should give money to.

The data sausage the algorithm makes is often wrong though. The classic example is once you’ve purchased a wallet on Amazon, the company then is a fervent believer that you, yes you, have now become an avid wallet collector so every ad, recommendation and notice sent your way is a genuine leather handcrafted wallet to add to your growing collection. It’s actually kind of amazing how wrong the algorithm can be sometimes. I once read a tweet from ex-game show host, now professional jerk wad and teeth-whitener enthusiast, Chuck Woolery so now Twitter believes I’d like nothing better than seeing how American Nazis are doing? I don’t know if Mr. Woolery is aware of this, but his tweets are ganged up with racist weirdos like some kind of Third Reich Collector’s Box Set. When Twitter wants to send me this crap, I know the algorithm isn’t all it’s hyped up to be.

Stranger still, is how wrong my television is. It knows every sport I watch, every ad I mute, every show I record, and every movie I rent. It knows I like 30-minute comedies, ice hockey, American football and Canadian news. If it were only a tad smarter it could guess when I get home from work and when I go to bed. Despite all of this it still shows me ads for feminine hygiene products, John Deer lawn mowers and cars I will never buy. I don’t want to tell them how to do their job, but where are the ads for hair growth, beer that won’t make me fat and carbon fibre road bikes?

That’s the worrying part. Despite having so so so so so much information about me, the algorithms of various companies keep failing. Twitter is convinced I am devoted to the whereabouts of Rihanna and would love to know the Dodgers pennant chance, Pinterest is convinced I’m a 60-year-old woman just dying to bedazzle my jeans, Facebook does know I follow stories about cycling in Toronto but apparently thinks I’m obsessed with stories about road deaths, and Google is convinced that the only news of note is what’s going on in the Marvel Cinematic Universe (I mean, Okay, I’m a little guilty on that one). The reason these predictive algorithms are so often wrong has a lot to do with who created them. The biases of the person creating the algorithm will be baked into the formula. Think of it as a cook making you supper. If the cook is from France, your supper may have a whole lot more red wine than you bargained for. You have glasses, you have a mouth, so obviously you like red wine.

For years, people in technology wanted to make a fridge that told you when your milk was bad by making sensors on the milk carton or in the fridge but really, my debit card knows when I bought milk and probably knows how often I buy it and thus my bank should be able to guess when I’m out of milk. You don’t need "Internet of Everything" sensors. We’re already drowning in petabytes of data (that’s a million gigabytes by the way - so about 212,000 DVDs of feature length movies), we don’t really need more but we keep getting it. Why do I have to file taxes? Why doesn’t my phone that knows my employee number, my bank account and my city of Toronto tax assessment just file my taxes automatically. I’m glad it doesn’t (though Quicktax does try) because I can’t trust the magic of algorithms or Artificial Intelligence. My Gmail account knows all my secrets, but most of the time Google still can’t figure out that I live in Toronto. I’m not worried about robots running on artificial intelligence taking over the world. I’m worried when those robots do arrive they’ll be too stupid to look both ways crossing the street and we’ll just have broken robots everywhere and broken robots everywhere will be a huge mess for all of us.

Labels:

0 Comments:

Post a Comment

<< Home