Facebook's survey to rank trust of publications asks two questions: “Do you recognize the following websites” and “How much do you trust each of these domains?”
Last week, Facebook said its News Feed would prioritize links from publications its users deemed “trustworthy” in an upcoming survey.
BuzzFeedAlex Kantrowitz
Context & Ripple Effects
Facebook is filling in the details of the publisher-trust survey it announced days earlier, which will feed directly into News Feed rankings as news shrinks from 5% to 4% of feed posts. The mechanics are now on the table: two questions per domain — whether users recognize the site at all, and how much they trust it.
The reveal lands amid open criticism: The Atlantic had pressed Facebook on exactly these design choices, including whether media can contest rankings and whether user ideology skews results (those unanswered design questions frame why the question wording matters).
First-order effects
Publishers' News Feed visibility now hinges partly on brand recognition among surveyed users, so smaller outlets with low name awareness can be down-ranked regardless of content quality.
Facebook gets a scalable trust signal without employing editors, shifting editorial judgment onto its user base while keeping the resulting scores opaque.
Second-order effects
Publishers have an incentive to invest in broad brand-recognition campaigns rather than reporting quality alone, since familiarity is a gating question before trust is even measured.
Trust becomes platform infrastructure — a proprietary, uncontestable ranking input — and the later controversy over Facebook News counting Breitbart among 'trusted' sources shows how hard such judgments are to defend once codified.
If the pattern holds, platforms rather than press-industry bodies become the de facto arbiters of publication credibility, with the survey methodology itself as the contested terrain.
The trend: Platforms are converting subjective user sentiment into algorithmic trust scores that quietly govern distribution, making credibility a computed property owned by the platform rather than the press.
There is an entire discipline of experts who specialize in survey design. There's even a peer-reviewed annual conference called @AAPOR but Facebook has ENGINEERZ and UX and the DATAS. ¯\_()_/¯ http://www.buzzfeed.com/...
I'm seeing a lot of criticism of this Facebook survey, but not from experts in survey design. Do any of those folks care to weigh in? http://www.buzzfeed.com/...
this is kind of like a brand awareness survey, like an advertiser would run. Trust in news is much more complicated. How well-sourced is the article? Are other sites verifying it? Is it news or analysis? Don't trust any one site “entirely.” http://www.buzzfeed.com/...
Why Facebook's survey about trust won't make or break the media, featuring some new thoughts from @mosseri http://www.theverge.com/... http://twitter.com/...
Second thought: The fact that the survey itself is simplistic doesn't rule out the possibility that Facebook is handling the results with more sophistication than people are assuming. See this thread with @mosseri, head of news feed: https://twitter.com/...
Facebook's news survey is two questions long. How many publishers do they ask each user about? Are the users randomly selected? And so on. This is all clear as mud. https://www.buzzfeed.com/...
How trustworthy people feel a publisher is a actually a big signal for publishers for which we have data, and not a signal at all for publishers for which we don't. It's an important signal with limited coverage.
The other important thing to understand is this isn't a simple vote. We are not just valuing more publishers that a lot of people trust, but rather valuing more publishers that a lot of different types of people (based on reading habits) trust.
I'm curious to know how it compares to other signals — informative, local, how long people spend reading, whether they comment or share, etc. I assume given preponderance of signals, this one data point will have a limited impact in many cases