For the purposes of this debate, “truth” = “models purported to be actual conditions of reality”.
Science is a method of collecting independently reproducible, empirical data, interpreting the data, categorizing it and developing useful and predictive models from that data – useful in the sense that it predicts future data of the same sort, hereafter referred to as IRdata. One can consider “science” to be, entirely, the collection and development of a kind of IRdatabase, the purpose of which is to take IRdata and develop it into useful IRmodels.
For this argument’s sake, Let’s call all other data and models non-scientific, or NSdata and NSmodels.
Am I obligated to accept as true (in the sense of “what reality is”) any model developed by science, regardless of how useful and independently repeatable it is? Certainly not. In the first place, science doesn’t claim that “what reality is” (truth) is the same as that which is independently repeatable (scientific data & scientific theories/models). Science only claims its models function successfully and are independently repeatable. Science claims its models are provisional and subject to future revision or even abandonment; this is contrary and irreconcilable with any claims that a scientific model = truth claim about reality. Scientists that make truth-claims about reality are necessarily abandoning the fundamental principle of the provisional and fallible nature of scientific model-building.
Secondly, there is no natural or man-made law that says I should consider science at all when developing my views on what reality is. I’m simply not obligated to consider science anything more than a tool that happens to be useful in my experience – and I’m not even obligated to hold science as that much if I wish otherwise. Science itself makes no metaphysical demands upon my worldview whatsoever.
All scientific models are provisional and subject to change. There is no added scientific value in considering any scientific model “true” and, in fact, such a view might be a problem if new data indicates a different model. Once invested in as “true”, scientific models become psychologically entrenched as a paradigm that might thwart the development and acceptance of better models. Science eschews the assertion that scientific models represent truth (the way reality is). That is a metaphysical or religious worldview.
What does science say about NSdata (non-scientific) and NSmodels? All it has to say is that such data doesn’t meet the criteria for being admitted into or used by the scientific database. It doesn’t claim such data is true or false, only that it is not appropriate for scientific modeling.
What does science claim about models that contradict its models, such as the model that the Earth is only 6000 years old? Science would only be able to claim that the old-earth model is more useful than the 6000-year model; science doesn’t claim either model is true or false, because all scientific models are provisional and only models, not claims of reality. Science doesn’t claim its models are true, only useful (as described above), and so cannot be used (as far as science is concerned) to claim the 6000-year model of Earth is false. In science, the term “prove” is not a true-false claim about reality, but only about a provisional model’s capacity to produce independently reproducible (IR)data.
Am I obligated to dismiss NS data and NS models when it comes to beliefs about reality? Certainly not. Science has nothing to say about what the true nature of reality is; it only talks about IR models developed from IRdata. Nothing more.
So, how is it that some people claim that science has disproven, say, a Young Earth Creationist model of reality? Science can only say that the YEC model is not as effective at producing IR data as some other model – say, the old-earth model. Science itself can make no claim about the “reality” value of either model. Science cannot claim the Earth “is”, in reality, more than 6000 years old; it can only claim that its current, most effective IRmodel is that of a much older Earth. Some new evidence may show that our previous means of determining the age of earth were mistaken and that the Earth is more likely young – just as, in the past, scientists changed their view of the age of the Earth from thousands, to millions, to hundreds of millions, to billions of years.
Whatever the age of the Earth actually is may be entirely irrelevant to the usefulness of the IRmodel. If a false IRmodel produces more useful IRdata than a true NSmodel, then in science the true NSmodel will be discarded in favor of the false IRmodel.
One might wonder, how could a false IRmodel produce more useful IRdata than a true NSmodel? The answer is that reality itself may not mostly or entirely conform to the consensus, independently repeatable paradigm. Thus, in some cases, the true model may not be independently, universally repeatable (NS). As such, science would not even be able to accept the NSdata that would lead to such an NSmodel; it could only find whatever approximation could be supported by IRdata translated into an IRmodel.
Also, it might be that humans have only a limited access to true data, have a limited capacity to interpret it, and have a limited capacity to model it towards the ends of generating IR predictions. IOW, it may take something other than a human to be able to create scientifically functional true models; the best we may be able to do with limited capacity is to generate scientifically useful false models. IOW, the 6000 year-old earth model may be true in reality, but reality may be such that there is no IRdata that would support that model, or at least no IRdata that a human mind can apprehend, translate and model as such.
When one claims that science has either proved or even just indicates that some other model is not true (meaning, not a factual condition of reality), they are no longer speaking from a scientific viewpoint, but rather from a metaphysical viewpoint that has incorporated the IRdata and model system as part of its ontology – also known as “Scientific Realism”.
Science cannot disprove anything in any ontological sense; it cannot ascertain if the YEC model is true or false. Science cannot even say if it one model is more likely true than the other without circularity. Science can only say that one model is currently more scientifically effective at making independently reproducible, empirical predictions about data than the other, given the nature and limitations of the human condition & experience.