An interesting thing happened to me this week, I thought it would be worth recording as it demonstrates the current state of fact-checking ecosystem on the internet - and its remaining vulnerabilities.
Part of my process of compiling the Memia newsletter is to scan a *lot* of stuff during the week and then the articles / memes which I think are worth sharing I’ll post out using Buffer to Memia’s social media accounts on Twitter, LinkedIn and (way less engagement) Facebook.
This week I shared the following article from The Guardian and received the following notification from Facebook:
(I assume this also meant that the article wasn’t shared any further.)
Fair enough, so I clicked through to the recommended link. which took me to an article on climatefeedback.org (who?) containing very detailed reviews from named scientists questioning the scientific credibility of the article:
The scientist reviewers themselves were identified:
Three out of four of the reviewers’ profiles went through to a profile page with further onward links…: (one was a broken link)
All very transparent and I could click through to the Nature and Wiley journal websites to clarify that the qualifying publications were authentic.
At the bottom of the main review article is this credential:
Clicking on the green circle device took me to:
…and there pretty much the trail ended, except for the links to ICFN and Facebook’s announcement of ICFN/Poynter (who?). About 15-30mins down a rabbit hole ending with a website and a logo.
Only Facebook’s initial endorsement - by linking to the Climate Feedback website - gave it any authenticity.
So… on the one hand it’s illuminating that this fact checking infrastructure is out there - but as it stands it is so open to a DDOS attack. Given what we now know of GPT-3’s capabilities, it would be *really easy* for a malicious actor to create 20(,000) of these “Fact Checking” websites and provide 50 fake reviews for every news story… sowing doubt on pretty much every news story out there.
This “Fact Checking” chain needs to be more securely verifiable and usable. This means baked into the web browser itself - I think for every website visited next to the Padlock icon there could be an “Authenticity accredited by” device as well… and this daisy chained back to selected authenticity providers themselves (likely to be “Those Endorsed By Facebook, Google and Twitter”. (eg NOT endorsed by QAnon).
This was covered by The Verge a couple of years ago: Browser Plug-ins that spot Fake News show the difficulty of teckling the information apocalypse.
As I’ve previously written, clearly a lifetime of work still to do.