Sorry to necro this conversation. Just caught up with this thread during a slow work day. Thought this picture would do well here.
hah. You can do the same chain in Sociology, Macro Economics, Micro, Finance, Statistics, Math.
As far as whether it's a "real" science? Well, the thing is, Economics, as Hodj pointed out, had a large split after the big "philosophical" years. Once Marx and the crew threw out all their ideas, and they all broke down, people began to demand actual evidence beyond simple statistical correlations. Up until that point, most economics had just been trend studies, really--with all the shitty science that goes with them.
However, more recently, especially after a few big math heavy hitters came into the field and began using economics as a base point for some large hypothesis, there began to be a divergence in Micro/Finance economics and Macro/Trend economics. The more finance oriented slant is a heavy mathematical field that relies a lot on algorithms to predict outcomes. And these guys can be very, very accurate (Sometimes not so much, considering the 08 collapse was a break down of one of these risk assessments.) Anyway, this field is LESS interested in trying to explain what economics system works for everyone, and instead tries to predict specific economic outcomes once given a data-set. It's a far more mathematical driven Economics. This kind of economics, I would say, is a lot closer to hard science.
The philosophical stuff? Like a lot of Marx? Is softer science. It has trend data and what not, but it's nothing like some of the amazing predictive work being done. And this field makes BIG money, but it's also a field that you essentially need a math degree, or a CS degree for. I actually jumped into this degree from a CS degree for that reason (When I realized the chances of me making video games were nil.)....Is it an absolute hard science? Well, depending on the scale, it can be--like they do verifiable experimentation a lot on micro economic or financial situations. But it's not a "pure" science like Chemistry or Physics, no. It's more like theoretical aspects of physics. Where you make models, and then do large experiments to see if your models hold up.
Anyway, but .02$. This is one of those cases where both sides are a little right. Old school economics is a very "soft" science. Mathematical Economics for agent prediction? Pretty hardcore, it's almost pure math, with values assigned to agents through statistical trends, rather than statistical trends fueling arguments themselves.