“Trust, but verify” is a fun phrase to think about. The general idea is that the receiver of information isn’t in the wrong to trust the source, but should still be verified to determine the accuracy of the information.

The trust aspect is important, even if it first appears redundant, because if you didn’t trust a source to even provide the most basic of verifiable and correct information, it would never be considered, or even bothered to be verified. Why waste time on something you know will very likely be false?

Here is where the thought experiment begins. Let’s first get the assumptions out of the way:

  • A random geographical subset of the population of the US is chosen.
  • A news station, or paper or site, that wants to produce stories
  • These stories are created as a best-effort to be fair to the truth

From this alone, we see trouble already. Even if the news source is fair in its version of a story, it is up to the reader to either blindly trust the source, or verify it.

If the reader trusts it without much thought, we could talk about that, but its not as interesting. Instead, let’s assume our reader does the absolute minimum necessary for verification. What would that look like? Maybe the reader will internally ask themselves these questions:

  • Does this conform to what I have heard before?
  • Does this make intuitive sense, given what I know?
  • Does this match my personal experiences?

Now the problems are really clear. If the reader outright rejects the story because of one of these most basic “tests”, then its not possible to present a “fair” narrative. Our reader has naively done a verification step that shields them from viewpoints that are against their own understanding.

Ok, but that isn’t very fair to readers as a whole. Let’s now try to picture what a more sophisticated reader might do:

  • Searches around various information sources they already trust for more information
  • Asks friends and neighbors for advice (who will inevitably follow the same paths laid out here)
  • Random searches to sources they don’t have an established trust relationship with

Now its getting really interesting. The first one is problematic on its face: the reader is likely to already trust many similar sources, and so are all likely to present the information in the same way. The second is also somewhat true, but maybe the reader gets the oddball friend that disagrees with their general worldview. That would be great, but still not likely to happen. And how likely are they to trust/verify that friend? Less so.

And for searching random sources for contrarian/reinforcing information? With no established relationship of trust, its difficult to know. Inevitably, our reader is forced to fall back to their internal mechanisms for evaluating verifiability.

And the most sophisticated reader?

  • Directly reads scientific papers
  • Goes to the location and verifies first-hand
  • Repeats the study
  • Interrogates officials

But, then what is the point of a news media if someone is at this point? And more than likely, the reader will be ill-equipped to perform any of these functions to a satisfying degree. Worse, they may do them wrong, and in the process will internally reduce their trust by a large margin of the source they originally questioned.


What does this mean in general, though? That people will get comfortable with the sources they already trust, will likely verify information they are wary of through means which reinforce their worldview, and are unable to incorporate new information without significant doubt or personal cost.

If you want to understand where the divisivness in politics comes from, its roots lay about in this area.