Lots of people look at a chart of the past and use their imagination to extend the trend out into the future. You ought to be able to imagine multiple futures. Let's take this @MPRnews chart of Minnesota hospitalizations from @dhmontgomery as an example. (Thread.) 
Recently hospitalizations have been plateaued, rather like the plateau in mid May. That was a temporary leveling out in the ascending part of the curve. So maybe the downward part continues as well, with some bumps along the way, of which this is one.
On the other hand, the recent plateau is also rather like the one in late May. That marked the turning point—a temporary flattening as the direction changed from ascending to descending. If the recent plateau is like that, then the extrapolation is a turn back upward.
Then again, the recent plateau could be the start of a new phase—after a couple months of bumpy ascent and a month of bumpy descent, maybe we're starting into a month or two of bumpy plateau, with little wiggles up and down but no overall trend.
Which of these is right? That depends on human behavior, weather, and much else. But no one knows the precise relationship between the trajectory and those variables. All I'm sure of is that if your eye only sees one continuation of the curve, your imagination is too limited.
