I have a number of options with regards to how I travel to work, but most commonly I catch the train. In order to make productive use of this travel time, I often read, and my book of choice at the moment is "Science and its Limits" by Del Ratzsch — a particularly excellent book which deserves a log entry (if not an entire document) in its own right. This is my second reading of said book — it will take at least two readings to appreciate it properly.

Chapter two of this book is a brief overview of the traditional conception of science, which covers all views before Kuhn, such as Bacon, positivism, and Popper. The principle of uniformity is one of the things covered in this chapter, with particular reference to how an assumption like uniformitarianism fits into the various epistemic frameworks under discussion. There is also some discussion as to what it is that science provides, such as "predictions" and "explanations", and whether or not an explanation is just a post-fact prediction and vice versa.

Earlier this week on the train to work, as I was taking in all this philosophising, I was also keeping an eye on the stations the train was passing. I'm more likely to miss my stop because I'm totally engrossed in a book than because I'm asleep, you see, so I've learned to be a bit wary when reading. I've made the trip often enough now that I can name all the stations in order, and considering that fact in light of what I was reading in the book at the time, I had the following train of thought (if you'll excuse the pun).

I've been studying the railway line in question (the North Shore Line, for those familiar with Sydney) for some time, and we could think of this as a scientific study in the traditional sense. I've made repeated observations: the entire study is grounded in the empirical. Not only that, but my North Shore Line Theory is a very good one, and I can inerrantly predict what station a train will pass next given its current position and heading.

But even as I was basking in the cleverness of my own scientific achievement, it struck me that what I couldn't do was explain why the theory is true. Why is it that Killara is always after Gordon for a in-bound train? Why is there never another station between them? Why isn't Killara missing on some days? Well, if pressed on the matter, I could blather about uniformity, but uniformity is an assumption on my part that I'm using to make my predictions. An assumption is not a scientific explanation.

Clearly my predictions would break down under certain circumstances if uniformity were not true, but the fact that I've never observed an exception to uniformity doesn't make uniformity a scientific fact. Even if we take it from a probabilistic point of view, as Popper points out, the number of times that I've made my observations is finite, whereas the number of potential times at which the observation could have been made is infinite, so I still have a zero percent confidence in uniformity from a strictly scientific perspective. Maybe there is a station between Gordon and Killara sometimes, but not when I'm commuting, or maybe Killara only exists when I am commuting. Were someone to challenge me with these proposals, the best I could offer in return is my personal belief that the universe doesn't work that way, which is somewhat unsatisfying from a scientific perspective.

But even if uniformity were a known fact rather than an assumption, my North Shore Line Theory would still be pretty poor from an explanatory (as opposed to predictive) point of view. Think of scientific knowledge as a series of layers. "Layer zero", the base, is the real world. I don't have access to the "source code" for the real world (a complete, readable specification for how it behaves): I have to reverse engineer it by observing its behaviour to create a model which mimics that behaviour, then experimentally verify the model against the real world, lather, rinse, repeat. Think of the model as "layer one". In the case of my North Shore Line Theory, I have modeled the railway as an ordered list of stations which the train may traverse in either direction. This, so far as I have been able to tell, is a perfect model, given the scope of its factual claims: I have never detected any difference between the model and the reality.

That gives me a perfect predictor of how the North Shore Line works, but does it explain anything? To provide an explanation, I need to add another layer, "layer two", on top of my existing "layer one". Layer two would be a theory explaining why layer one is what it is. There may be more than one such theory, each of which explains some aspect of layer one. The assumption of uniformity can be seen as a layer two explanation, but it's not scientific. Similarly, appeals to such nebulous factors as the societal need for public transport, planning and decision-making on the part of railway engineers, and political pressure for the railway to service certain areas are also layer two (or higher) explanations, but they aren't scientific either. As far as North Shore Line Theory is concerned, I don't have a scientific explanation. Even if I did, what are the chances that I would then need an explanation for the explanation?

Exactly what all this implies (other than the usual "I'm mad and they will have to shoot me") is left as an exercise for the reader. And if I observe any exceptions to my North Shore Line Theory, I'll post them in the Nutter Log.