How to protect against fake ‘facts’
those superbly professional fact-gatherers sometimes have trouble verifying information. Social media can help — people can upload video from their cellphones of events as they happen. But we’re learning that social media can be tools of deception as well as truth.
So here’s an offbeat proposal: Just as the provenance of a work of art is established by art historians and auction houses, we need technological tools that will help confirm the provenance of facts.
The idea is simple: An art buyer should want to know that the painting attributed to, say, Leonardo da Vinci was in fact created by him. So specialists seek to reconstruct the chain of ownership, documenting how a work passed among collectors and galleries over the centuries. Scholars can’t always establish direct links back to the artist (and such uncertainty appears to cloud the recent purchase of Leonardo’s “Salvator Mundi” for $450 million). But the exercise is essential for the proper functioning of art markets.
There’s even a new approach known as “digital provenance.” This topic was explored at a conference last month at the Carnegie Museum of Art in Pittsburgh. One set of panels explored “visualizing object histories,” including the study of “movements of objects and people through time and space.”
The internet giants such as Google and Facebook should be tuning their systems to establish the provenance of fact. I’d like to see them using machine learning to interrogate supposed facts to establish where they’ve been — how they first surfaced, and how they were passed from user to user. If there are gaps in provenance — an unexplained missing link in the chain of evidence, or signs of misattribution — then those anomalies should be flagged automatically.
In this fact-provenance scheme, I hope we’d be able to trace information back to the source and detect evidence of manipulation. News organizations whose track record is shown to be reliable would get weighted positively by the digital system. Such measurement of source reliability would be a tricky test, but it doesn’t strike me as impossible. Google does something similar every time you type a phrase into the search box.
People could still base their decisions on dubious, unverified “facts,” just as a collector can still buy a painting whose provenance is suspect. But there would at least be fair warning.
We speak often of the “marketplace of ideas,” in the expectation that markets operate rationally and efficiently. In this framing, consumers assume that most markets aren’t rigged, and that tainted or unsafe products have been screened. That’s obviously a complicated problem when the commodity is information — the First Amendment lets us promote ideas without rating their safety. But the First Amendment doesn’t protect fraud or libel.
We live in an information ecosystem. If it becomes polluted, all the creatures that depend on that ecosystem are at risk. We say that sunlight is the best disinfectant. But that’s true only when the sun shines brightly.