I think the best response to this "direct/indirect" distinction is that it doesn't matter, regardless of how you choose to define these terms, or what sorts of examples you pick to represent each type.
What matters is the accuracy of predictions that result from the combination of the evidence that you have (whatever type it might be) and the particular hypotheses and theories you induce to account for the evidence.
To take your magnetism example, it doesn't matter whether you refer to the observations of iron filings on a piece of paper just above a magnet to be direct or indirect. What matters is whether you can form a hypothesis about magnetism based on this evidence, and then use that to make accurate predictions about some other situation that you haven't observed yet, but could observe by controlled means (i.e. by experiment, which could either be direct manipulation of objects and forces, or focused inspection of natural phenomena).
If the prediction turns out to be correct, then the evidence is good enough, no matter how you got it, and your hypothesis was sufficient to account for the phenomena. Of course, an essential caveat here is that the evidence and hypotheses must stand up to repeated reassessment, to check for possible biases, unexpected conditioning factors, etc.
The more evidence you find to support a given hypothesis or theory, no matter what type of evidence it is, the more confidence you have that the hypothesis or theory is valid and useful.
To rub it in, you can point out to the creationist that the bible has been used to predict the date of the second coming of Christ (i.e. "the end times", "the rapture", the end of the world as we know it). In fact, there is currently a widely circulated prediction that this will happen on May 21, 2011. It was previously predicted to happen at several previous dates, which have long since gone by. Indeed, Jesus Christ himself, according to scripture, stated that this event would happen during the lifetime of the people who were listening to him, and apologists have spent a lot of effort to navigate around the apparent failure of his prediction.
All previous predictions about the second coming (whose end dates have all passed by) were obviously wrong, and we'll know within a few weeks whether the May 21 prediction has any merit. Given the track record and the nature of the evidence, would your creationist friend care to stake anything of importance on its potential accuracy?
Ask whether he would posit any other predictions, based on purely scriptural explanations, without appeal to the scientific theories that have thrived and proved their worth over the last few hundred years. It's important to be clear that when the prediction fails, attributing it to God's will is irrelevant. A failed prediction is a failure, period.
In order to hold a candle to what can be done (and is done on a regular basis) by the combination of geology, physics and evolutionary biology, a creationist's prediction would have to handle a collection of facts equal in diversity and detail to the following:
Suppose we're poking around at a small cliff near a stream or river and find a mammalian skull, and we decide to hand it off for study. One researcher takes careful measurements (cranial capacity, jaw length, brows, ocular and nasal openings, teeth, etc). Another researcher independently takes a small sample of the skull itself and of materials that were adjacent to the skull when it was found, and applies a variety of dating methods to estimate how long ago this being was alive. A third researcher looks at the materials in the cliff above and below where the skull was found.
Evolutionary theory predicts that their respective findings, which are all established independently and are based on techniques that have already been proven useful/accurate by means of other independent predictions, will converge:
(1) the skull will be found to be intermediate between other known skulls of similar (but slightly different) shape and size;
(2) the dates determined from skull and adjacent material will likewise be intermediate between the dates that were estimated for those other similar skulls;
(3) the dates of deposition of the strata below and above the position of the skull in the cliff will also be found to be older and newer, respectively, than the age of the skull and the material it was embedded in.
Fossils that trace the development of cetacean, equine, canine, feline and primate/hominid species have all demonstrated that sort of convergence. And we haven't even mentioned yet all the DNA and other biological/morphological evidence drawn from the species living today, which provides yet another (and even stronger) confirmation for the relationships inferable from the fossils and their comparison to modern skeletons.
Of course, creationists love to fantasize about the inaccuracy of dating techniques, but there's a whole section at EvC devoted to this (and plenty more, presented quite efficiently, at
TalkOrigins Archive: Exploring the Creation/Evolution Controversy), which should suffice to show that these fantasies fail, and one must be willfully ignorant to accept them as a refutation of dating methods. To nail that down, just point out that oil companies spend many millions of dollars annually to make the dating methods as accurate as possible, because when those methods work accurately, the companies make billions. (And guess what: they do make billions.)
You can call all the evidence indirect if you like. The important thing is: it works.
autotelic adj. (of an entity or event) having within itself the purpose of its existence or happening.
™ Version 4.2
(1)
