As someone who is professionally required to at least occasionally issue prognostications about the future, I enjoyed this passage from Cory Doctorow’s Walkaway, in which a character warns against the perils of straight-line extrapolations:
End-timers used to project our consumption levels forward, multiplying our population by our needed resources, and get to this point where we’d run out of planet in a generation and there’d be famine and war.
That kind of linear projection is the kind of thinking that gets people into trouble when they think about the future. It’s like thinking, ‘well, my kid is learning ten exciting new things every week, so by the time she’s sixty, she’ll be smarter than any human in history.’
There are lots of curves that start looking like they go up and to the right forever, but turn into bell curves, or inverted Us, or S-curves, or the fabled hockey-stick that gets steeper and steeper until it goes straight vertical.
Any assumption that we’re going to end up like now, but moreso, is so insufficiently weird it’s the only thing you can be sure won’t happen in the future.
That’s a fairly self-referential statement for a character in a science-fiction novel to make, but thankfully most of the characters in this piece of utopian fiction do not go around making speeches; Doctorow is very good at writing real, vivid people, not types or abstractions.
Perhaps the line is meant to highlight that the world Doctorow imagines in Walkaway is indeed very much like now, but moreso: it’s a straight-line extrapolation of rising inequality, increasing automation and declining labor participation. What he’s trying to imagine is the moment when that curve starts to turn into something else.
The Tangled Woof 
But isn’t it the “end-timers” who made the *non-linear* forecast? They forecast that the population curve would turn into an “inverted U”, right? Isn’t the forecast of “famine and war” the non-linear forecast?
And isn’t the linear forecast the one that turned out to be correct? Don’t we have the same things we had 100 years ago, “but moreso”? More people, more industrialization, more development and infrastructure, more urbanization, more technology, etc. etc. It seems to me that the linear forecast was quite accurate.
Over relatively short periods of time (< 250 years), the linear forecast seems reasonably prescient. Over longer periods, non-linear behavior occurs as people's ways of thinking change (classical Greece, the Enlightenment, etc.) It's those changes in ways of thinking that cause the curvature in historical development paths.
People go for linear forecasts because they’re easy, and in most cases appear more probable. This is related to Taleb’s discussions of black swans and fat tails; sometimes the improbable happens and that makes all the difference. For example, the advent of the A Bomb was a shot out of the dark that utterly transformed geopolitics, but could only be guessed at starting around 1938.
By the way, Doctorow’s “Daniel” is his wildest book: Jewish paranoia elevated to a mad spiraling phantasmagoria of suffering.