Let’s presume that Nate Silver or whoever predicted every election at 80% confidence. If they got 99 out of 100 “right” that would mean that they were actually wrong. Their “80%” would need to actually be closer to “90%” to be accurate, because getting the “80% chance” 99% of the time shows some bias in your probability assignment.
I have been hearing the same thing for about 8 years. Next year is always the year that web apps kill mobile apps, just like next year is always the year of the Linux desktop. I remember reading an almost identical article about 5 years ago which also used a weather app as a core example. Funny how the simplest apps keep popping up!