I think you have the image of Galileo too firmly lodged in your mind, boldly declaring "it still moves".
Science is determined by the preponderance of evidence. And Ockham's razor plays a part too. If you found a result that suggested there was no link between smoking and cancer, we're not suddenly going to ignore the huge mass of evidence that suggests there is a link, because the greater likelihood is that you're an asswipe with an axe to grind, have made a mistake in your work, or that an artefact has weakened your results.
To pick up on your Newtonian physics point, the new science had to explain both the new areas where the old science didn't work and the old areas where the old science did work. The same would be true for a result that purported to show no link between smoking and cancer.
You're arguing against yourself, because I've never raised the notion of there not being unquestionable settled science. I've simply said that where there is strong consensus in science, people are going to be interested in pushing ahead with building on that consensus and very reluctant to reexamine the fundamentals based on a single study purporting to show they're all wrong. Things that make the latter more likely are the reputation of the authors, reviewers and publication, the credibility of any replacement theory including its ability to explain previous results, etc etc.
You've misunderstood what I meant by allocative efficiency. It doesn't involve payment. Never mind.
There was no such consensus about smoking being good for you. That's a myth. The tobacco industry drove a particular view in their self-interest, but the evidence contradicted them - the former is not science, the latter is. Here's a nuanced reading. You'll note that the epidemiological studies were only conducted in the late 50s, so it wasn't about contradictory evidence as about the first credible evidence on the subject. http://www.healio.com/hematolo...