However personally speaking I would (and do) in fact discount a reading where even two sentences are highly suspect; it makes it not worth spending the time to read the rest of it.
As another example, I work primarily in data analytics. If I produce a report where even a single number is wrong, it almost immediately calls into question all of the other reporting I produced (did they use the same unreliable source? what transformations did they apply? was any sanity checking performed?). And, as it should.
Accuracy is incredibly important to making things appear legitimate.
I guess my main point -- phrased a bit more aggressively than needed -- was that if you have a huge community-sourced pile of data from multiple people in multiple parts of the country, relating to complex and chaotic situations that are unfolding as we speak, and all you need to dismiss it out of hand is finding a couple of things that you find suspect... well then you're always going to dismiss it.
Identifying, discussing and removing data points that don't belong is absolutely useful and fair. Taking a glance at a mountain of data, pointing out a couple pieces that you don't like and implying that the entire pile is rubbish is neither useful nor fair to me.
And I'm sorry for calling you dishonest, or at least heavily implying it. That was stupid and rash of me.