I had to re-read the discussion and re-examine the graphics to understand what the writer is alleging. Basically, given the that rural residents favor republicans, there should be a pattern of mail-in ballots trending more right than left as those more distant votes move randomly through the mail system. Absentee ballots from cities would be counted early and should skew left (if that many were actually cast for Biden).
In the graphs provided, the pattern should curve downward (the data is given as a ratio of Democrat / Republican votes) as more rural conservative votes arrive and are counted. The graphs for WI, PA, GA, MI even VA don’t follow the expected pattern, that downward curve is disrupted and shifted upwards to Biden in these states. This shouldn’t happen and didn’t in other states (like FL and MN). Here