Wikipedia’s Cleopatra mystery, AI and the future of truth on the internet
Future News 134
At first, they thought it was a film, and then they figured it out. Why exactly had Cleopatra, the Queen of Egypt from 51 to 30 BC, been so prominent on Wikipedia for so long, consistently placing near the top of the 100 most popular pages on the world’s favourite free encyclopaedia?
The answer lies in artificial intelligence. To be exact, Google’s own AI and its virtual assistant, Google Assistant. It first deployed on Android-enabled smartphones back in 2017.
Samsung owners, who outnumber iPhone owners globally, started being prompted in 2021 about Cleopatra and her associated Wikipedia page as part of the voice assistant’s set-up procedures.
The tie-up allowed the page to overtake the latest and greatest news stories and streaming and movie trends, which typically dominate Wikipedia’s rankings. ‘Deaths in 2022’, ‘Stranger Things’ and ‘The Grey Man’ were some of the most popular pages in July, for example.
Funny enough, there is another Alphabet connection on Wikipedia's leading pages: YouTube. Content creators on the video and streaming platform often link to Wikipedia articles in a video’s description section, while the YouTube Wikipedia page features prominently on Google search.
With most Wikipedia pages placing high on Alphabet’s flagship website, there seems to be an unofficial nexus between the two internet and media entities.
As Wired’s Noam Cohen has observed: “Google and Wikipedia have been in a kind of unspoken partnership: Wikipedia produces the information Google serves up in response to user queries, and Google builds up Wikipedia’s reputation as a source of trustworthy information…”
Well, in 2021, the same year ‘Cleopatra’ started to consistently rank as one of Wikipedia’s most popular pages, Wikimedia Enterprise launched. A year later the unofficial relationship between Alphabet and Wikipedia became formal.
A by-product of Wikimedia Foundation’s 2030 Movement Strategy, Wikimedia Enterprise is led by Lane Becker and has its own LLC status. It is wholly owned by the Wikimedia Foundation.
What exactly does the business arm do? We are told it enables companies to “retrieve data from Wikimedia projects in any language, access metadata packaged exclusively for Wikimedia Enterprise, and detect vandalism or important updates at the article level.”
The launch announcement of Wikimedia Enterprise gives further details:
“In most cases, commercial entities that reuse Wikimedia content at a high volume have product, service, and system requirements that go beyond what Wikimedia freely provides through publicly available APIs and data dumps.
The information panels shown in search engine results and the information served by virtual home assistants are examples of how Wikimedia content is frequently used by other websites.”
Naturally, this offering is very attractive to Big Tech companies because of the scalability aspects it brings by helpfully and easily linking into the massive information resource that is Wikipedia.
For the Wikimedia Foundation, Enterprise brings in revenues to help fund Wikipedia and other projects, whilst bolstering the organisation’s $240m worth of total assets.
But the project does raise serious questions and concerns about information gatekeeping and how Wikipedia editors and contributors, such as PericlesofAthens, one of the prominent members behind the Cleopatra page, are rewarded.
As it stands, editors and contributors volunteer because they are presumably motivated by the idea of spreading and sharing good and otherwise hard-to-access information to the world. The amazing resource they have built over more than two decades, Wikipedia, can be used by individuals, small business, scholars and major corporations alike.
All is fair, all is equal. But that changes with the more sophisticated API and meta-data access, which can be used to improve voice assistants, search engines and AI training models, amongst other products, for a chosen paying few.
Why shouldn’t the core information providers — the editors and contributors — get a cut of the revenue streams, and why don’t they get access to this improved service that they have helped built?
There is also an inequality in information access and speed. Everyone has access to the Wikipedia API, but not the new Wikipedia API 2.0 (let’s dub it).
So what? It helps the encyclopaedia going at the end of the day.
With Wikimedia Enterprise only launching in 2021, it’s unclear where this now leads to. If contributions to the Wikimedia Foundation grow, will direct and indirect pressures mount on the organisation to accommodate Wikimedia Enterprise customers down the line?
Consider another part of the publishing industry: the newspaper sector. It has always boasted of so-called Chinese Walls between sales and editorial teams, ensuring that journalists are not harassed or strong-armed by the advertising wings of their outlets.
But reporters at some commercial-minded outlets have long-known about the potential and actual editorial impacts of big paying advertisers and benefactors.
Weekly UK satirical outlet Private Eye even has a dedicated column, Street of Shame, which occasionally highlights alleged conflicts of interest between publishers and advertisers, including tales of how stories are killed at inception and later down the publishing process because of commercial and/or political sensitivities.
A cynic would say this is the reality of day-to-day politics and business, and there is nothing much to see or complain about here. Fine, but now the Wikimedia Foundation – and by extension Wikipedia itself – has its toe officially dipped in these sometimes treacherous waters for the first time. It’s therefore reasonable to ask: Does it truly know what it’s getting itself into?
FN is read by hundreds of readers. I would like to make it thousands. Subscribe and share.
📺 Media news of interest