You're missing the point: The graph is the same for the period of time which the editor selected. But he carefully selected the left (past) cutoff point to be on the same level as the right cutoff point. But in the original data, there's a steep incline at the beginning of the graph.

Now I can't vouch for the data or that the original data doesn't have a steep drop-off just a little before the snippet we're shown, but this is a prime example how choosing the sample properly changes the result dramatically.