Surprised Economists

I've been struck repeatedly over the last few years by how many stories about the economy contain phrases along the lines of "economists were surprised by..."
It seems like they are surprised by any economic news that comes out. Something is always more or less than predictions, and not by just a little, but by orders of magnitude. For example, economists were surprised again today by job numbers and factory orders. (Incidentally, I wish I'd got to this earlier, the wording earlier today was different, the "news people" softened their description of how surprising the numbers were).

a government report showed the nation’s employers cut far fewer jobs than expected last month, stirring optimism about the buoyancy of the economy. . .
The Labor Department’s report that employers cut 20,000 jobs in April was a relief to Wall Street, which had been expecting payrolls to fall by 75,000 jobs. . .
The Commerce Department said U.S. manufacturers saw orders increase 1.4 percent in March. Economists expected a 0.2 percent increase after declines in January and February.

They were expecting 75000 fewer jobs. They were only off by 275% This one isn't sooooo bad, after all there are a lot of jobs in the nation, and the usual gains are 200-300 thousand. Maybe they had a bad day. But, 0.2 versus 1.4? That's off by 600% and the viable range on those numbers is never more than a few percentage points.
I can only think of 3 explanations for the constant stream of these stories:

• First, the stupid economists are the only ones giving interviews.
• Second, these are the smart ones, and even they don't know what they are talking about.
• Third, and this is my theory: You cannot reduce the countless individual choices of a quarter billion people to a formula with any degree of accuracy.

This may disappoint some of my economist friends, but economics is NOT a hard science, it is a descriptive science. It is valuable when looking retrospectively, but absolutely useless as a tool for precise prediction.
Economics is not the only 'science' with this problem. Political science, sociology, psychology, and even, to some extent biology have the same shortcoming. This is because they are trying to predict the actions of beings with souls and free will. In such cases, the best you can do is averages and general trends.
That doesn't mean they are not worthwhile tools, but their practitioners need to recognize the limits of their science and not advocate more detailed policies than the science can support.