This post is a continuation of my discussion with Michael. This time, I want to discuss the power of science and what it can do to tell us a claim isn’t true. To do this, I am going to look at a few discarded scientific models of the past, and how it is we have come to know they are no longer true. But I am also going to at least pay lip service to the process and methods of science. To give any justice to explaining what science is and it’s methods, I would have to spend 2,000 words on it, minimum, and to do it full justice I would have to write a rather large book. I am not going to do that.
In my last post, I shared Sean Carroll’s description of how science works: models are created from observation and data, those models are then tested (and where there is more than one model, the models are compared in their ability to explain the data). Michael made a rather immature complaint about that, but none of his criticisms really hit the mark (see here). So, I will made some explanation on that original description, but only in the context of the discarded models I shall explore.
It is worth sharing at this point that Michael has expressed absolutely no interest in the actual data in the published journal he attempted to cite last time. My concerns about science in the media (i.e. the imperfect chain of custody that leads to sensationalised, inaccurate or outright wrong things being reported) from my last post may have warned against simply reading headlines or not researching a story further, but I didn’t even think to guard against someone who suddenly decides the data isn’t as important as what they already believed. Unlike confirmation bias (as a cognitive error), that is intentional and disingenuous.
The reason this topic remains relevant to the initial discussion (‘Does Dawkins misrepresent science?’) is because Dawkins claims that science shows God is wildly unlikely. Michael’s counterargument, then, is that there are no experiments that directly deal with that question, and so Dawkins must be misrepresenting science. I am here to argue this simply isn’t the case: scientific conclusions have reach beyond their initial and parochial experiments. If they didn’t, a scientific experiment would be nothing more than a heavily recorded event in history. Instead, science relies on induction for the experiment to be grouped with like experiments to start making predictions about the future.
This is important: induction is not simply an add-on, superfluous to the essential character of science, it is vital. Induction is the difference between ‘Steve was really hurt when that horse kicked him’ and ‘horse kicks are very dangerous’.
Going back to Carroll’s model-comparison idea of science, imagine a scientist a few generations ago trying to compare two models of disease: the Germ Theory of Disease, and the Smell Theory of Disease. Many people held to the Smell Theory of Disease, and they believed this model of reality not just because of superstitions, but because they actually observed people getting sick near unpleasant smelling things and becoming unpleasant smelly when they were ill (and then that one person could become ill from an unpleasant smelling ill person). There were no controlled experiments on the issue, but there were observations and data that informed a model.
In addition to this model, we also have a scientist who believes the modern Germ Theory of Disease. The data used in support of Smell Theory also supported Germ Theory, because germs and smells often cohabitate. It is important, then, to make sure the two models used are well-defined enough for robust differences between the models to be proclaimed. We have two such models, in this instance. Smell Theory asserts that disease is spread by unpleasant smells, and these can be masked (which should also stop the spread of disease, if the model is true). That is an experiment that can be run: potpourri would stop the spread of a disease. But, it would not stop germs. This is a real difference that could be explored.
Alternatively, germs do not exclusively live alongside unpleasant smells, so if a disease could be induced absent a bad smell, that would support Germ Theory.
Historical records suggest variations of this were tried in people’s personal hygiene, but never run as an experiment. Ever. The Smell Theory of Disease has not been falsified. But it is declared not true, by science. How does this work? It is simply the case that one model was better at accounting for the data than the other. Smell Theory was never formally falsified, it just stopped getting talked about because the Germ Theory had so much more explanatory power. This is science, correctly, declaring one thing highly unlikely to be true without directly testing it; it just got out competed by another model.
A precursor to Plate Tectonics was that land moved through the ocean bed. The model accounted for the data: that South America seems to fit into the West Coast of Africa, and that ― if they were together ― there are extinct animal fossils in contiguous locations. But, the model never took off. It never took off, despite never being empirically tested. Instead, it was thrown out of mathematical grounds: the energy required for continents to plough through the oceanic crust simply stopped it being a sensible model. The model was explored to see if it was even sensible before it was tested. Yet again, something was declared unlikely to be true by science, without experiment.
Experiment may be considered iconic part of science, and it is a critical part of creating new data. But that doesn’t mean all science has to use it in every case. For example, plate tectonics was not empirically validated until very recently. In the meantime, it was accepted for its immense power to explain the data: moving plates, contiguous fossils, volcanoes, earthquakes, and alternating magnetism in the sea floor. (Actually, that last one could be considered empirical data in support of the claim.)
Here comes in the issue: the difference between primary data collected for a particular question (empirical support) and secondary data that is relevant, which a model still has to account for. Both are important in science.
This is, indeed, a long way off being a comprehensive look at science. But with these tools alone, can science tell us God is wildly unlikely? Yes. Any well defined model that includes a God always fails in comparison to natural models trying to explain the same data. Like the Smell Theory, God-based models are consistently outcompeted. And, like Smell Theory, that means such models are discarded and not considered true.
The big contrast between Smell Theory and God-based models is that Smell Theory was well-defined. You can define a God-based model really well, and people have, but such models always fail intellectual or empirical analysis. A well-defined God-model is subject to the Problem of Suffering; but the model survives because it is not well defined and the model moulds. Suddenly God is not omnibenevolent, and so the suffering is permitted. (Or, some other horn of the Euthyphro dilemma.)
In biology, where Dawkins worked, this is overwhelmingly the case. The natural model of evolution outcompetes the God-model in accounting for the evidence. Sure, ‘tests of faith’ can be asserted to make the God-model account for more data, but that is a symptom of it being poorly defined, and not at all a strength of the idea. And this process repeats through cosmology, another domain people like to assert God-models to explain. Sean Carroll explains this very informatively to William Lane Craig in this debate.
The real misrepresentation happening here is people who refuse to claim God has an effect on cosmology or biology or chemistry. The idea that God created the Universe and Life is a claim, and the Biblical claim in particular is specific about what that looked like. And even if you want to claim that God set it all up to look exactly like a natural process, calling on just the natural process is still a better model than calling on some agent that intentionally made it look like It wasn’t there.
As a model, God makes predictions about reality. And the areas those predictions are in are better explained by naturalism. And, it doesn’t take much induction to say they always will be. God is highly unlikely. That claim is within the purview of science.