Karl Popper is more controversial than I can possibly explain. To me, it is manifestly obvious that in order for an idea to be "right", there must be some way for it to be wrong (otherwise, it's a tautology, and those are not very interesting). To this end, an experiment is a technique for demonstrating that an idea may be wrong. But if an experiment fails to invalidate a hypothesis, does that mean the hypothesis is right? What if there is a defect in the experiment? And the converse can be said if it does invalidate a hypothesis. So it is not sufficient to conduct an experiment. We must conduct it by multiple independent teams.
In my mind, the "hardness" of a science is directly related to the extent to which that science can produce experimentally validated ideas, and the reproduction rate of those experiments. Of course, there is a difference between theory and practice, and even physics, the "hardest of the hard" sciences, can suffer from non-reproducible results if they are produced by bad physicists, lab technicians, or outright cranks and frauds. However, if a field produces a predominance of non-reproducible results, then it would be quite unseemly to blame it on a preponderance of charlatans (though for some fields, that may actually be the case).
While this term is applied to science in general, there is a particular set of fields for which this is especially acute, as described in Wikipedia. Unfortunately, economics is one of fields said to be "in crisis". The article mentions that in one sample, a full third of the results could not be replicated. This source says that replication failure is 40% for economics (for the sample studied). I find it curious that an economist was comforted by the "high" replication rate of 60%. If chemistry or genomics experiments could only be replicated at a 60% rate, I dare say that DowDuPoint, BASF, and Illumina would simply not exist.
If I am curious about the chemical properties of any common substance, like, say, benzene, I can find reliable results for many properties to 5 significant figures. If experiments to determine the listed properties of benzene were only reproducible 60% of the time, I dare say few people would be foolish enough to engage in large-scale production and usage of this compound (or most industrial compounds).
But nowhere is the reliance upon science more notable or critical than public policy. When the state expends the blood, sweat and tears of the population in pursuit of a common goal, said citizens generally demand some assurance that the project is worthwhile and achievable. So let's take a brief look at how the US gov't spends money on the sciences:
This is taken from the WP article on Science policy of the United States. If we focus on the basic and applied research categories, we see that HHS and energy appear to be the biggest winners. Given that medical science is directly applicable and relevant to taxpayers, it should be no surprise that HHS receives the lion's share of research dollars. And the fundamental nature of energy to economic output should also not be surprising. Representing fields, we could say that biology/medical science and physics/chemistry are all well represented by these two categories.
NASA and the DoD also receive decent allocations, and we can presume that these both predominantly represent physics and chemistry among the sciences (although there is surely a non-trivial focus on biology/medicine as well). Which means that whatever funds go to the remaining sciences most likely appear as NSF grants and whatever is covered by "Other". While the "Other" category is fairly large, the NSF appears to be the smallest allocation. This is to say that whatever economic research is funded by the federal gov't most likely funnels through the NSF, and represents one of the smallest categories of research spending.
Of course, this is not to say that the field of economics has no impact on the Federal gov't. On the contrary, it may, in fact, have the most direct impact, given the sovereignty of the US dollar, the scope and operation of the US Treasury and the Federal Reserve, not to mention the SEC, IRS, and the fact that most of Congress' power is explicitly reserved for regulating commerce. So isn't it ironic that the printing, spending, allocation, and manipulation of money is one of the largest and most impactful functions of the US Federal Government, and yet it must also be one of the smallest actual research funding targets?
I would suggest that reflects the uncertain value of economic research relative to the higher spending targets. Just consider that when a NASA funded rocket blows up on its way to space, the exact failure mode is often identified down the the seal, wire harness, or line of code. But if a tax break or incentive fails to produce a particular desired effect, the number of possible explanations is limited only by the number of economists available to offer them.
Results in physics, chemistry, and biology often reach consensus as soon as a critical mass of experimental evidence rolls in. In economics, it is not clear what would even constitute a critical mass of evidence. Furthermore, economics is so contentious that it has "schools". While one could argue that physics has "String Theory" and "Loop Quantum Gravity" schools, among other controversies, these generally relate to the very edges of theory where our ability to separate ideas by experiment reaches fundamental limitations. The economic schools, by contrast, encompass a majority of the field, rather than the fringes. It would be like physicists divided over whether the sun revolves around the earth or not.
Just to pick on one particular somewhat well-known idea in economics, I think it is fair to say that the Black-Scholes options pricing model is one of the most sophisticated and precise models in the entire field (and quite widely used to help determine options compensation for corporate employees). Unfortunately, it is wrong. Now, let us not be too harsh on Black and Scholes. Every theory requires some simplifying assumptions, and when you get to the level of messy humans trading equities, the assumptions are going to inevitably paper over some pretty convoluted realities, or quickly become intractable. Black-Scholes is "wrong" because the phenomenon it is trying to explain is simply too complex for the number of parameters in their model. If we built a model that was perfectly accurate, it would have too many parameters for us to ever use it. But this also tells us why economics can never get too close to the "hard" side of the sciences: ultimately, it depends on the behavior of fantastically complicated and absurd human beings. And frankly, we often operate in pretty unpredictable ways. It's pretty hard to make a science out of that.