Tuesday, October 24, 2006

A challenge to creationists

I'm being very lazy about the Protein Challenge at the moment, due to a combination of general lassitude and complete lack of any idea as to where to start. I will get onto it, especially since the issues raised seem to be flaring up again over on Paul's blog (he's doing a series on Dembski's design inference and specification).

In the meantime, here's another money-where-mouth-is challenge to those of you out there who have very different beliefs to me in the area of origins.

Science and predictivity

A scientific model (a hypothesis or group of hypotheses) is said to be predictive if it tells us the results of experiments that we haven't performed yet. For example, the model of quantum mechanics tells us how electrons will behave in various potentials without our actually having to generate those potentials and throw electrons at them.

It's fair to say that predictive models are the holy grail of science. These perfect crystal balls, these insights into the future, justify the entire enterprise - they are what makes science so unbelievably useful.

Evolution and predictivity

Evolution has been the only model that accurately fitted all the data for over a century now. However, it's theoretically possible to create any number of models to fit a given set of data (although currently the only alternatives are very heavy on the Goddidits), so that in itself is not conclusive evidence for evolution.

What is conclusive evidence that evolution is at least on the right lines is the large number of confirmed predictions that evolution has made. For example:

  • When the genomes of the great apes (inc. humans) were first studied, it was observed that humans had one chromosome per gamete less than any of their closest relatives. As humans were unique in this respect, it was hypothesised that having one more chromosome was the ancestral condition. It was known at the time that it is nigh-on impossible to just lose a chromosome, hence it was hypothesised that at some point in the human lineage two chromosomes must have fused. It was therefore predicted that one human chromosome would look exactly like two fused chimp chromosomes. This prediction was later confirmed.


  • It's been known for a long time that primates are unable to produce vitamin C (chemical name: L-ascorbic acid). Since most mammals can, it was hypothesised that this was the ancestral condition, and that primates have subsequently lost that ability. With the emergence of population genetics, it became apparent that, in this case, remnants of the vitC gene should exist in primates - these pseudogenes wouldn't have had time to fade away. Hence it was predicted that primate genomes would contain DNA strings very similar to those used to produce vitC in other mammals. This prediction was later confirmed

Neither of these facts were predicted by any other models that existed at the time.

The Challenge

The challenge is aimed at anyone who claims that creationism, or any other alternative origins conjecture, has scientific merit. Simply provide me with one valid prediction that your model has produced. If you can do this, I'll publicly declare that your model is scientific, and devote the next month or so of my life to figuring out

The conditions

Some of the following criteria for what constitutes a "valid prediction" will border on the insulting to the more honest of you, but I've seen people argue each of these points so it's best to get things out in the open. The criteria I'm concerned with are:

  • It must be new knowledge - it can't be something we already knew to be true at the time the prediction was made. "Predicting" the existence of the universe doesn't qualify.
  • It must be concrete, i.e. relating to the physical universe - sociological predictions derived from physical models will not generally be accepted. Predicting that people will be picky about your model doesn't qualify.
  • It must be testable, i.e. both verifiable and falsifiable. "Predicting" that evolution won't be able to explain something doesn't qualify, as that conjecture is falsifiable but not verifiable (it's actually an hypothesis). Note: "predictions" that are verifiable but not falsifiable may be accepted as a weaker evidence for your model.
  • It must be unique to your model - so no predictions that would also be made by current mainstream scientific models. Predicting that the sun will come up tomorrow doesn't qualify.
  • It must be confirmed. Predicting something that hasn't actually been tested yet doesn't qualify.
  • It must follow inevitably from the model. Conjecturing that God exists and hence predicting that Mars will have subterranean water will not qualify unless you provide a damn good rationale. Throwing out thousands of mutually-conflicting conjectures and hoping one will stick is also not acceptable.
  • It must be checkable - I must be able to confirm that all the other criteria hold. For this reason, predictions older than 50 years will not in general be accepted.

Disclaimer: I reserve the right to modify these criteria if someone comes up with some really really daft workaround that I hadn't thought of, but I solemnly swear that this will only be done in absolutely exceptional circumstances.

These criteria are not excessive - both of the evolutionary predictions I discussed above pass with flying colours. These criteria are not ad-hoc - each and every point is essential to proper scientific hypothesis testing. I'm currently fairly sure that these criteria, rigorously enforced, are sufficient to confirm that a model is at least broadly accurate.

Before anyone asks, emotional arguments, or logical arguments not relating to hypothesis testing, will not be accepted for the purposes of this challenge.

No, I would accept the evidence

One response that I've heard in a variety of contexts is "you're an atheist, you wouldn't accept the evidence even if we gave it to you". This is factually inaccurate.

I self-define primarily as a member of the reality-based community. Although, at present, most RBCers are atheists, this is for strictly pragmatic reasons rather than any sort of underlying resistance to any given brand of theism. In short, if you provide me with the evidence, I will accept your hypothesis as being more accurate in this area than the mainstream scientific consensus (if one exists in that area).

2 comments:

John said...

The trouble with "creationists" is that they have already assumed the essentially reductionist asana of separation from the all pervasive Love-Bliss-Radiance of Real God.

And to ask the question "does god exist?" and to try to prove his/hers
existence is in effect to affirm the negative proposition--god does not exist.
Also to try to "prove" that the "creaor" god exists by appeals to processes in biology is absurd because all biological forms (including of course humans) disintegrate and die. Where is the "hopefulness" in that immutable fact? It is also a rather childish search for the protective father "god".

These 3 related essays provide an Illuminated understanding of the science vs exoteric religion shouting match.
1. www.dabase.net/creamyth.htm
2. www.dabase.net/noface.htm
3. www.dabase.net/rgcbpobk.htm

Lifewish said...

Hi John, good to have you posting.

I'm undoubtedly in agreement with at least some of what you say, and will definitely spend some time reading your site. However, my overall response is the same as my response to creationists. You have a model of the physical universe - consensual reality. If that model is objectively real, what predictions does it make? If it doesn't make predictions then does that mean it's "only" real for individuals who find it personally helpful?

Science is fundamentally pragmatic, and I'd tend to agree with that stance. If you have an alternative to the current mainstream approach, you'll need to be able to tell us: where's the beef? What are you achieving with your new approach that we couldn't already achieve with ours? Or is it merely more philosophically satisfying for you?

If that's the case then there's nothing fundamentally wrong with that - it's on a par with me setting my watch ten minutes fast to improve my punctuality - but I feel this is something that should be acknowledged.

And to ask the question "does god exist?" and to try to prove his/hers
existence is in effect to affirm the negative proposition--god does not exist.


I'm honestly not sure what you mean by this - are you bothered by the premise that "no" should be the default answer to the question? If so, I'd ask: why should our default response to the question "does God exist" be any different from our default response to the question "do unicorns exist"?