Bryan Caplan makes a very compelling point that if you think first-hand accounts are not credible, you should find the news even *less* credible: https://www.econlib.org/first-hand-experience-is-less-biased-than-news/

I think the second two points are doing most of the work there.

I tend to agree with his assessment, but I think he's discounting (or missing) several things working in the other direction:

1. Institutional and reputational forces. If news consumers care about truth or if the culture among journalists is to value truth, news organizations have incentives to work against these biases.

2. "Reliable" is ill-defined here. If most people's threat model is "someone might lie to me, have been deceived or be wrong", in theory news organizations may be more trustworthy - they are much more likely to simply be wrong than to lie, and they can and often do hire fact checkers to follow up more deeply into a story than any random person would.

3. People may be aware that "shows up frequently in the news" is only weakly correlated with "happens a lot", and may already strongly discount this factor. Reading an accurate account of a very unlikely thing is not really *wrong*, and it's unclear how it compares to a dubious account of a relatively common thing in terms of pragmatic value.