I work in a field of data-driven modeling of physical systems (think simulating weather patterns, particle interations, solid structures, etc.). As the name implies, producing a good model depends on access to high-quality, or "high-fidelity" data that it taken to be the truth, or a fairly accurate representation of nature or an experiment. However, the system I am trying to model is one which nobody can consistently produce simulations that exactly match experiments. This is not a huge problem for me, as I simply use the best simulations that reproduce experiments in a general, qualitative sense. I am working to develop good data-driven models which can be generalized to any variety of physical systems, not just my particularly difficult system. Hopefully there will be a point in the future where I have access to good data, but that time is not now.
However, many researchers who have been attempting to create better high-fidelity models look down on this, claiming that there is no point trying to simulate a system if the inputs don't match nature exactly. "Garbage in, garbage out" is the standard derogatory epithet. This strikes me as odd, since they are the ones tasked with producing the accurate high-fidelity models, and they cannot do it. It also strikes me as anti-intellectual, spurning an avenue of research which will have measurable benefit once they make their models better.
Am I wrong in this sentiment? Is it stupid to pursue an area of research that is essentially dependent on future discoveries and improvements in modern methods? Are there any examples out there of research that was performed under the assumption that future methods would make it useful?