I trust what I see as being a correct report of the many polls cited. I don't see that the report adds any value to the data. It doesn't turn data into information. Fr'instance the Kelly-McSally race. This article cites a VERY LARGE swing in sentiment from less than 2 weeks ago, suggesting Kelly's huge lead has evaporated. Per survey summaries at 538 and RCP, the Biden/Trump sentiment has tightened a bit in this same period. The recent polls are generally of lower quality. There certainly COULD be a large shift, but IMO the more likely issue is results being compared to one another in a way that makes it all seem a mystery. The article seems more modest-quality than untrustworthy due to bias or error.