Distortions
Signs we're getting better at spotting tech's warping effects
One of the truly fascinating things about being a tech journalist over the last dozen years or so is living through rounds of what I like to think of as learnings, or the revisiting of how the micro-industry thinks about its beat. Of course, it’s not unique to tech. Certainly political journalists have spent the last handful of years rethinking assumptions, but it’s arguably the most true in technology, where we’ve left behind an era of cheerleading to one where we’re grappling with making sense of our subject.
And because the field is so new — to pick one obvious data point, Facebook didn’t exist before 2004 — the really fun aspect is that journalists and the public and the industry are all going through it at the same time. One of those parallel learnings seems to be happening in real-time right now: we’re getting better at spotting tech’s subtle distorting effects.
“Technology is neither good nor bad, nor is it neutral.” — Kranzberg’s First Law, 1986
We’re all now fairly well familiar with the idea that we might call ‘algorithm bad,’ or that the machine’s working in the background of popular technologies can be hugely problematic; we’ve lived through the mess that was Twitter and Facebook’s role in the 2016 U.S. presidential election, watched “The Social Dilemma,” let Anderson Cooper walk us through a “60 Minutes” segment on the use of facial recognition by police.
But of late we seem to be really picking up on the good stuff — the even more nuanced ways that technology is warping the world as we daily experience it.
Let’s take a quick tour through a few current examples.
First up: the auto-cropping of photographs on Twitter. There’s been public pressure on the company over the idea that its algorithm for automatically resizing images to fit more easily in our timelines was quite literally centering men and people with lighter skin.
And so Twitter conducted a bias assessment, the recently released results of which find that its algorithm is distorting what users would see without it in play.
But the results of Twitter’s assessment weren’t straightforward.
Twitter’s auto-cropping feature disfavored, for example, black people at a 4% difference from what would be demographic parity, but it also slightly favored women at an 8% difference. When the pool is limited to black women and white women, white women are favored at an 7% difference from parity. It’d be easy to conclude from that that the algorithm is badly done and needs adjusting.
But Twitter’s takeaway, per Rumman Chowhury — Twitter’s director of machine-learning ethics, transparency, and accountability — writes in a blog post, seems to be that a “‘fairness” analysis isn’t even the right approach. And that’s because trading away the ability of users making their own determination about how they’d choose to be represented in favor of speed — as in, letting machines crop user photos to get those tweets on the timeline quicker — was a bad decision here.
Here’s Chowdhury:
“One of our conclusions is that not everything on Twitter is a good candidate for an algorithm, and in this case, how to crop an image is a decision best made by people.”
Another example: Amazon, and its effect on the U.S. economy. D.C. attorney general, Karl Racine, has filed a antitrust lawsuit against Amazon, arguing that the platform’s agreements with third-party sellers “impose an artificially high price floor across the online retail marketplace and allow Amazon to build and maintain monopoly power.”
Over on his Big newsletter, Matt Stoller is making the case that the D.C. suit is less conventional than it is on its face — looking as it does as an attempt by Racine to fit it into the traditional antitrust think of harm tied to raising prices for consumers — because it really gets at how third-party deals power Amazon’s big money maker.
Writes Stoller, “the Amazon Prime program, the keystone that holds Amazon’s dominance over retail together, is effectively being subsidized by the scheme Racine laid out,” in that Amazon relies on higher prices with outside sellers to subsidize the free shipping that makes Prime nearly impossible for would-be customers to resist.
You don’t have to sign on to all the particulars of Stoller’s analysis to see it as an attempt to unlock our understanding of not just how Amazon operates but its impact on retail in the United States — a grounding that federal regulators will need to have if they opt to grapple with Amazon. (That is, should President Biden ever appoint antitrust leaders, including the chair of the FTC and the head of the Department of Justice’s antitrust division; as Kim Hart reports for Axios, historically speaking, slow rolling his appointments in that arena.)
And last example has to do with Instagram. That platform is far less scrutinized and less understood than Facebook’s News Feed — a true blessing for Instagram during the 2016 election — but Instagram’s still extremely big business that shapes the U.S. economy.
And for Vox, Allie Jones looks at one slice of that business: the prize giveaways where, for example, a Kris Jenner or a Kourtney Kardashian offers fans a chance at “Saint Laurent handbags, luxury baby strollers, and credit cards ‘preloaded’ with thousands of dollars.” These so-called loop giveaways, it turns out, often serve to funnel followers to users who pay to borrow the platforms of bigger Instagram names. Jones:
“The lesser influencers and brands pay a marketing firm like Social Stance to be on the must-follow list, and Social Stance pays the featured influencer, like [“Vanderpump Rules” star Stassi] Schroeder, to post about the giveaway.”
While in this case it’s still unclear to Jones and the rest of us whether anyone actually ever wins anything, all this picking apart is encouraging, frankly.
We’re starting to see a media-public-industry feedback loop that’s ever more mindful of how these things are designed, and how as a result our engagement in the world gets shifted away from what it otherwise would be.
It becomes an intriguing thing to keep an eye out for: what are the more subtle ways that the technology you use are distorting your daily life?
Speaking of mindfulness, I’ve been practicing gratitude lately — in the form of listing five things I’m thankful for, ideally before I get out of bed — and one thing I’m appreciating today: YouTube.
The TEDx talk above is recirculating, on why it’s important to pronounce people’s name correctly. This comes up a lot for me, as I’m often, in the course of reporting, calling up experts who I don’t know and asking them to take time to explain stuff to me. Getting their name wrong in that situation is unconscionable, and so a trick I use involves YouTube; I search their name and cross my fingers that they’ve introduced themselves at an event.
Maybe we should have a day where we all record and post clips of us saying our names?
(You might think there’d be no way to mispronounce “Scola,” but you’d be wrong.)