## Sunday, July 23, 2006

### evidence, prediction, and old jokes

There is an old joke that has been reified in various ways on various TV comedies. I don't know how to tell it with a punch line so let me just describe it: a dog owner tells someone that his dog, Spot, is well-trained. The dog is lying on the ground ignoring the proceedings, and the owner says, "Spot, down!". The dog does nothing but he is already laying down so this is arguably the right thing. The owner continues, "Spot, just lay there and don't do anything!" Again, the dog complies by doing nothing.

The humor (sorry for analyzing this; I usually get annoyed when people analyze humor but I'm getting to a point eventually) comes from the fact that the commands were obeyed by default. They were obeyed, arguably, but only because the owner was able to predict what the dog would do and commanded the dog to do what it was already going to do.

Now suppose that you and I are sitting on a porch, watching a hummingbird at a feeder. It comes to the feeder, drinks for a moment, and then flies away. A few minutes later, it returns to feed again for a moment and then flies away again. It repeats this behavior five or six times. Now I tell you, "Gentle Reader, I have formulated a theory on where the humming bird is going when it flies away. It is taking the sugar water that it drinks from this feeder and going to the back of the house to fill the feeder back there, for I noticed that the feeder back there is lower than this one, and the hummingbird wants them to have equal levels."

You will probably dispute with me on the merits of my theory. You will argue that hummingbirds do not have a compulsive need to make sure that all feeders are kept at the same level, and that furthermore, it doesn't seem likely that the little creature could pump sugar water into the other feeder anyway. I argue, "But, Gentle Reader, my theory makes falsifiable predictions. The hummingbird could not have filled the other feeder yet, so I predict that it will return in just a few seconds."

When my prediction comes true, does this count as any sort of evidence in favor of my theory? Of course not. I noted a pattern in nature and I predicted that the pattern would occur again; such predictions don't require a theory. One doesn't need a theory of motion to predict how a stone will fly when you throw it. One doesn't need a theory of behavior to predict that if you sneak up on a sleeping cat and make a loud noise, it will shoot away almost faster than you can see. The only thing you need to make predictions like this is observation and the ability to recognize patterns in nature.

When a theory is designed to explain or account for a pattern that is already recognized, the fact that it predicts the pattern is no evidence for the theory at all, none, because the theory was created on the assumption that the pattern would continue and any theory created on that assumption would have made the same prediction. Therefore, this prediction cannot be used to distinguish between good theories and bad theories, only between theories that predicted that the pattern would continue and theories that predicted otherwise.

My previous post on the evidence for evolution combined several things that I should have kept separate. In particular, all three of the "predictions" that I discussed fall into this category: that they were simply continuations of a previously known pattern. Technically, I didn't have to add anything in order to show that these predictions are useless as evidence in favor of evolution.

I emphasize that this makes them entirely useless evidence, not merely poor evidence, so arguments to prop them up will not help. Twice as good as zero is still zero. In order to rescue these "predictions" from uselessness, you need to show that the patterns predicted had no part in the creation of the theory that predicts them. If they played any part at all, then they cannot be used as evidence for the theory.

So what was the point of my other objection: that evolution doesn't predict them anyway? This is a bit subtle, and I passed over it too quickly, so let me be more specific. Darwinian theory, taken as a whole, does predict those phenomena. However, the parts of Darwinian theory that predict those phenomena were specifically designed to predict those phenomena so repetition of the pattern does not provide evidence for evolution, a theory designed on the assumption that the patterns would continue to be observed.

In order o get around this, someone might argue that the fundamental core of Darwinian theory, the idea of common descent, was not created with those phenomena in mind. I doubt that this is true, but it doesn't matter anyway unless you can show how the phenomena can be predicted from the core of Darwinian theory alone. In other words, you have to show that in a world with common descent, it is impossible for those three phenomena not to have been observed. Such a proof is (it seems obvious) impossible, so you are left with a set of three predictions, none of which can logically be used to support the theory of evolution.

Now, there were other predictions in that same chapter, predictions of phenomena that probably were not observed before Darwin constructed his theory. These involve extensions of parahomology, analogy, and suboptimality to the molecular level. It isn't obvious to me that people observing these phenomena at the macroscopic level would expect the patterns to continue at the molecular level, so these are arguably better predictions.

My case against these predictions is slightly different. I don't know the actual history of when these molecular phenomena were predicted and when they were observed, but the core of Darwinian theory does not predict these things, so I predict that they were observed first, and then were explained post hoc in the same way that the macroscopic phenomena were explained. If it turns out that they were predicted first, then I'd still say that the theory itself does not predict it so this was just a perceptive extension from the macroscopic to the microscopic of the sort that could have been done just as easily based on the patterns alone, with no theory, so it still does not confirm the theory.

In other words, Chapter 3 of the FAQ contains no evidence in favor of Darwinian theory at all.