14 C
New York
Tuesday, May 21, 2024

The Shaky Ground Truths of Wikipedia

Just outside my window in soggy Seattle, life goes on unedited. A smiley blue Amazon van nearly knocks down a man on a bike. A dog walker bags fresh poop and dumps it into a neighbor’s trash. A couple pushes a baby carriage with a Black Lives Matter sticker. The driver of the green garbage truck coming around the corner waves and smiles; he’s the only Black life I’ve seen in this neighborhood since I moved in.

The kitty on my lap focuses on the pair of pudgy robins, the chattering chickadees, the evil-eyed Tyrannosaurus crow.

Most of the time, though, I’m peering into a window that isn’t—though it wants me to think it is.

It’s a portal leading to a dizzying array of creatively curated worlds put together to cater to the interests of people deemed to be like me, worlds that entertain or disturb or inform or entrap.


No matter how hard I press my nose against this window, it can only show me visions that exist in someone else’s head, reflect someone’s else editorial/curatorial judgment or aesthetics. Those visions, in turn, often cannibalize content from portals similar to themselves—refashioned, rehashed, refreshed, sometimes airbrushed, sometimes poisoned. Sometimes, they seem to be chasing each other’s tails.

Meanwhile, I can’t see what’s coming around the corner. I can’t disambiguate signal from noise; someone has done that for me. And that gives me the creeps.

Maybe it’s because, at age 74, I’ve tripped over ground truth so many times, blindly stumbled into things that flew in the face of everything I thought I knew: that dreaded “diva” who became my best friend; that impenetrable poem that became wrenchingly clear when read by the author in person. Life experienced offers continual opportunities for reality checks, brakes on assumptions that shape our expectations, always on autopilot whether we like it or not. I did not expect the pretty ballerina in my class to be the math whiz, or the muscle-bound jock to make poetry out of physics. Unconscious bias comes in all forms. There is no autocorrect. But the truth is “out there”—if you can get out there to connect.

Now that Covid has removed me from the immediacy of “out there,” I’ve become obsessed with the problem of ground truth. I miss having literal touchstones. Discovering things you really “get” only if they get in your face. How do we know what’s true when we can’t show up in person and don’t know anyone who can? How can you stick your finger to the wind when the wind’s behind the screen? How can I know what I can’t see—and worse: How can I know what I don’t know I can’t see?

In normal times, I’m a big fan of wandering around at random, propelled by Brownian motion, bouncing off whatever, getting lost. I’m good at getting lost, though I don’t consider it that. After all, I’m somewhere, just somewhere I’d never have landed if I had a sense of direction. It’s the best way to discover what’s what and who’s who and a lot of hows and whys and whatnots and wherefores. When someone says “You had to be there!” it’s sometimes really the case.

It was just such a random encounter that made me start wondering about Wikipedia. A few weeks pre-lockdown, I saw a flyer in a window inviting the public to an “edit-a-thon,” a free workshop where anyone could learn how to build a Wiki page. The impetus for the session was the gaping hole in Wikipedia where women and people of color should be. They were missing, I was told, because they too often lacked “notability.”

I’d never thought much about Wikipedia until its pages started appearing high up in Google searches—and not just Google. If you ask Siri or Alexa a question, chances are the source of your answer will be Wikipedia too. Hundreds of AI platforms use Wikipedia data; machine learning trains on it. So if women are missing there, they will be missing elsewhere as well.

Women in science went missing long before Wiki, of course—in press coverage, top billing at meetings, appearance on panels. They weren’t much on my radar either when I first started writing about science decades ago and showed up for the “March meeting” of the American Physical Society. The March meeting is nerd mecca: Close to 10,000 physicists gather to present findings in “condensed matter,” which is everything from quantum computers to lasers to smart materials. AI and nano everything.

A friendly Black woman noticed my obvious confusion and helped guide me through the maze of talks, panels, sessions. She was Shirley Ann Jackson, who I later learned was the first Black woman to get a doctorate in theoretical physics from MIT (where, she said, she was mistaken for a maid). She took me to the reception for women in physics. I was seriously wowed. A lot more physicists were women (and vice versa) than I ever imagined. Where had they been? Where had I been?

Decades later, I had a similar wake-up call at an exclusive meeting in Aspen of top physicists in what was then known as string theory—tackling the most fundamental questions of space, time, energy, stuff. I expected that a lot of the material would be exotic and unfamiliar. What really seemed exotic and unfamiliar were the three Black men among the small elite group.

For most people, the description “theoretical physicist” doesn’t immediately conjure an image of a Black person. (Neil de Grasse Tyson is great, but one example doesn’t count, and he’s not a part of this particular tribe anyway.) After the Aspen meeting, my ground truth shifted. I could picture Black men as theoretical physicists no problem because I’d met them, interviewed them, hung out.

Then it struck me: Just about every female physicist I know, and every Black physicist I know, are people I met in person. I hadn’t even noticed their absence until their presence hit me in face.

My portals aren’t so diverse. Which is why Wikipedia in the age of corona had me worried.

A lot of people say they use Wikipedia only as a starting point, a first reference. After all, everyone knows that it’s crowdsourced. It’s proudly non-expert. A community of editors ultimately arbitrates what’s in, what’s out, what matters, what’s true. Because there are so many of them (250,000 edits a day, according to one source), the idea is truth will out.

Yet Wikipedia’s top ranking on Google gives it credibility and authority that misrepresents what it is—a community consensus. “This is a problem,” according to Atilus, a leading digital marketing company. During an audit for a client using prime SEO software, Atilus found Wikipedia at or near the top over and over again, frequently with a prominent sidebar. On its blog, Atilus posed a not-always-hypothetical question to illustrate the problem: Would you rather trust a doctor who’s undergone rigorous training or people who spend time in health-related chatrooms or an intern who blogs about heart health? That’s a big oversimplification, given that Wiki entries are supposed to be double-sourced and edited. But even when that works, it’s not even close to ground truth.

Granted, Wikipedia isn’t the only source of content that’s creeps into everything, ubiquitous and unavoidable. It could be your mom or The New York Times. What gives Wikipedia a central place in data heaven is that popular algorithms that lead us around by the nose go to the site to learn. AI reinforces whatever biases are put in front of them. “Big data processes codify the past,” writes Cathy O’Neil in her book Weapons of Math Destruction.

So if historically scientists have been a certain sort of man, that’s what makes history—literally. It puts you in the loop. If you’re not on Wikipedia, you’re out of the loop for AI, Google, and the rest. As one woman scientist put it, “the Wiki Google loop is a noose.”

The more general problem is: What happens to ground truth when so many roads lead back to the same source? Monocultures cause problems. The Irish potato famine that killed hundreds of thousands of people happened in large part because farmers came to depend on a single vulnerable strain. If bananas vanish from supermarket shelves, you can blame it on the breeding of a single variation—easier to grow and harvest but with no ability to survive disease. The SolarWinds software breach wouldn’t have caused so much damage had it not been so widely and deeply embedded in, according to some sources, almost all Fortune 500 companies, the US Treasury Department, and even Homeland Security.

All this got to me to thinking about what else I’m missing, now that I’m not getting to science meetings or much else. It’s surely a lot. Most of my favorite things (and people) came into my life through accidental encounters. Physics, for one. Also cats, tap dancing, Seattle, even (no kidding) accounting.

My portals don’t show me 74-year-old tap-dancing ladies much. Or interesting, fun accountants, for that matter.


Not seeing yourself reflected can feel as if you’ve been ghosted (ghosts, like vampires, generally don’t have reflections). Women scientists at the front lines fighting Covid-19 recently voiced their frustrations at not seeing their work in pandemic coverage. In a commentary published in Times Higher Education, several dozen clinicians and researchers from Harvard, Stanford, Yale, Johns Hopkins, and Columbia wrote: “The scientific response to Covid-19 has been characterized by an extraordinary level of sexism and racism … Women are advising policymakers, designing clinical trials, coordinating field studies, and leading data collection and analysis, but you would never know it from the media coverage of the pandemic.”

Journalists under deadline pressure reach out on autopilot to who’s most visible—most likely a man. This “distortion of reality,” the commentary warned, can be dangerous when vocal, but unqualified men get the public’s ear more often than qualified women (men who might be notable in other fields, for example, but not epidemiology or medicine).

Getting mentioned in the media matters, since one of the biggest hurdles barring women from the Wiki loop is the “notability” requirement. To be notable, your work has to be noted. Who’s doing the noticing can make a real difference, especially when content is crowdsourced. The in-crowd for Wikipedia is the editors; between 80 and 90 percent of them are men. When I was an editor at a science magazine, I sent a writer to report on a major collider project. The story he wrote had no women. Why not? There’re weren’t any, he told me. It took me about 10 minutes to find out the lead scientist on the experiment was a woman. He just hadn’t noticed.

The noticing problem might help to explain some weird and well-known Wiki anomalies, like the fact that 2018 Nobel laureate in physics, Donna Strickland, wasn’t included until she won the prize. A lot of contributors were trying to get her a page. The editors rejected them.

When they work well, my portals work wonders. I happened upon a dance studio that offered tap classes, and a portal took me in no time at all to practice routines online, a place to buy tap shoes. Portals transport me to Seattle’s fabulous parks, to physics meetings, to cat videos. I’ve been Zoom zooming around with friends to theater performances and digital dance, so we can still support the arts organizations we love. I’ve been portal-pigging-out on philosophy and old films. I’m learning to identify the birds outside my window. I’m laughing myself silly on the creative mash-ups put together by so many great TV comics and wacky YouTube videos put together by just about everyone.

But I still can’t see around the corner, and I rarely stumble upon things I didn’t know I didn’t know—despite the thousands of portals designed to promote such serendipitous encounters. AI can only feed on the diet it’s given. It can’t know what it doesn’t know either.

It’s great that the American Institute of Physics is collaborating with Black in Physics to host a Wiki-thon; this year’s March meeting features a Pi Day edit-a-thon to address Wiki’s missing women. These kinds of events might help fix the numbers problem.

More troubling cracks in Wiki’s foundation really hit home, however, when I first dug beneath the surface during the edit-a-thon I attended.

For a while, I’d been getting personal emails from readers warning me that someone was messing with my own Wiki page. I hadn’t paid attention because I didn’t have to. (Not caring what other people think is a privilege of age.)

But here I was at a Wiki edit-a-thon, so I mentioned it to our instructors. “Sounds like you have a troll,” one of them said. I was flattered. Trolls are big in Seattle.

Turned out someone had managed to get a banner put on the page that said, in essence, this page isn’t reliable because it was written by KC Cole’s friends. When we popped open the hood, we found that several people had already complained about the warning. One of them said the edit seemed malicious.

The instructor quickly looked up who’d done the edit and how it was sourced. Apparently, the troll had sourced the troll’s own blog. So much for double sourcing. And the comments objecting to the edit didn’t go anywhere. I didn’t know any of the people who put together my page (but thank you). The name of the troll, however, was familiar—someone whose work I reviewed a long time ago, not very favorably. A guy.

That’s good old ground truth for you. Not what I expected, but then it rarely is. It’s often there for the finding if you know enough to ask. Knowing there’s a question is a whole lot harder.

Related Articles

Latest Articles