Wednesday, January 09, 2008

Scientific Realism and Science Fiction

C. Ewing asks a series of three interrelated questions,

Does science make statements about "the world", or is its role merely to give us the best dice roll possible? That is, when we're dealing with scientific laws and theories, are we expected to see it merely as a predictive tool or is it saying something more broad and significant, a statement of how reality is or at least seems to be?

This is perhaps related to the next:

When discussing other possible worlds and possibilities therein, ought one take science (as we have it in this world, obviously) into consideration? When our science tells us that an object cannot be accelerated to, much less beyond the speed of light, it is descriptive of this world's scientific observations, and what seems consistent within that paradigm.

However, seeing as how science is by its very nature empirical, and therefore descriptive not prescriptive (and also limited by the experimentation and data) how does is it even sensible to take it into account when discussing other worlds, save in the instances in which the thought experiment of said other world is assuming it to have a form much akin (if not in all scientifically important ways) to our own?

This is tied intimately to the third and final question:

In a given fictional realm, where it is given that a spaceship does indeed travel faster than the speed of light, and you (we'll call this person Y for sake of argument) say that this is not possible: is person Y actually saying anything reasonable/consistent? Sure, it may be reasonable to say it's not possible here, but there it seems quite evident that it is possible, so isn't claim of person Y, in fact, demonstrably false? In an otherworldly sense of the term demonstrable, obviously.
Let's take them in order.

The first question deals with what we call the realism/instrumentalism debate. Science is useful, there's little controversy there, but does that usefulness imply the literal truth of tools used to successfully predict?

The place to start is with what is called the semantic view of theories in which we think of scientific theories as sets of models. Models represent, depict, or stand for the system we are curious about. Consider Bill Cosby's classic routine on street football where the quarterback gets down on one knee to design the play by drawing it in the dirt, "Shorty, you're the piece of glass." "I don't wanna be the piece of glass." "What do you want to be?" "I want to be the Coke bottle top." "O.k., you're the Coke bottle top." Now, Shorty doesn't really look like the Coke bottle top, but in the model, certain aspects are meant to describe the real world, in this case, the lines in the dirt are meant to resemble the trajectories of the receivers running out for the pass. So models have some parts that are meant to be representative and others that are not. A model is neither true nor false, but better or worse. A good model represents the system we are interested in in such a way that we can form a mental picture of how it might work and derive results that can be reinterpreted in terms of the real system, results that can be tested. A good model's predictions come true, at least in a prescribed region and the bigger that region the better the model.

But better and worse doesn't mean real. We can have better and worse models of the space shuttle Columbia and better and worse models of the starship Enterprise, the difference, of course, is that the first one is modeling something that actually exists and the second one isn't. Which are scientific theories more like?

Instrumentalists say the Enterprise. Maybe there is something out there to be modeled, maybe there isn't, all we know is whether our models let us do the things we need them to do. Science is a toolbox, not a metaphysical microscope giving us access to some hidden underlying structure of ultimate reality. These are scientific agnostics. Bas van Fraassen, aside from being one of the best writers and having one of the coolest names in all of contemporary philosophy, is one of this view's strongest and smartest advocates.

Realists say the Columbia. The argument is that usefulness is an indicator of something. How could it be so useful if it weren't actually right, if it weren't really mirroring some aspect of the real underlying structure? It would be a miracle if it had nothing to do with reality, but just somehow happened to always get the right answer.

Now, we know that no scientific theory is finished deal. Every theory is born refuted and will be replaced at some point by a better one. But for really good theories, the replacements tend to be sharpenings, not radical overturnings. As such, we may not be wise in contending that the details must be true of the world, but don't we have reason to think at least the outlines are the likely the case? To think so is to be a realist. I tend to be more of a realist myself, I'll admit. One has to temper how much one is willing to posit of the system itself from the model, but some models are so accurate in their predictions and provide novel predictions in areas that were not even considered when the theory was formulated and in those cases, it seems that the model must be modeling something to be doing it so well.

This leads into the second question. We can talk about other possible worlds that differ from ours but follow the same laws and theories, but are different models of those theories. theories are sets of models and some of those models are parts of the world we live in and some aren't. For example, Kurt Godel showed that in Einstein's general theory of relativity, there are models with closed temporal loops, that is cases in which a fast moving electron could travel in spacetime and bounce off itself. This will not happen in our universe because it would take a very peculiar arrangement of matter and energy, but it is one of the models of the theory and we can talk about a possible world in which it is true.

Which brings us to question three. Fictional worlds could be of two kinds, (1) those that have laws like ours, but initial conditions that are different, and (2) those that follow different laws. Laws of science, even on a strictly realist understanding are not necessary truths, they describe how the world happens to be not how the world has to be. The world could have been Aristotelian, it could have been Newtonian, it just isn't. One does not risk immediate and necessary contradiction by creating a universe that runs on principles different from those that govern ours.

At the same time, it is interesting and often difficult to see where different laws clash if you want to set up a set that includes some of our laws. Trying to answer this question was a project undertaken in great detail by early analytic philosophers like Hans Reichenbach who wanted to see where Kant's version of Newton clashed with Einstein's new understanding of the universe...but that's a long and different post...actually, that's my dissertation.