Aug 22, 2011

Creationist apologetics: Ignoring what's right in front of them.

It's presidential hopeful time in the US again. What does that mean? It means that otherwise intelligent-sounding people will fall over each other to proclaim their lack of belief in basic sciences like stratigraphy, nuclear decay, and evolutionary biology.

I could wax philosophical about why I think that happens, but instead I'll comment on how scientific measurements often get twisted by people with an agenda (and in particular for this case, a religious one).

Before I begin, let me discuss the fact that there is a fundamental assumption in science that "what we observe, reflects what is". Incidentally, this ultimately allows one of the most important hallmarks of modern science: reproducibility.

If what I observe reflects what is, then what someone else observes should be consistent with what I observe. Similarly, if you repeat your measurement over and over, it will agree with previous measurements if you've done things correctly. Furthermore, this allows something in statistics called "the central limit theorem":

If you repeat your measurements over and over, they will fall within a bell-shaped distribution called the "Normal" or "Gaussian" distribution, with the average representing the measurement of the "true" value and
the "spread" representing the accuracy of your measurements.

The way that this thinking is applied to measure the age of the earth is called "radiometric dating". In particular, the most robust modern method is called "isochron dating".

One example of a creationist argument to refute this is shown here:

I think the people that wrote this have just enough information to be misleading, but not enough to get the whole picture.

Isochron dating is a statistical technique, because the samples 
involved are so widely varied in spatial extent, placement, and time of acquisition. There can be situations where natural fluctuations occur throughout the earth's crust, and can affect the results locally 
to various levels. This is *naturally* included in the statistical 
 uncertainty of the measurement. It's not a question of one single measurement, but the ensemble of measurements. To affect the mean, 
there needs to be a systematic bias in the underlying sample itself.

In addition, all of the measurements must come with an uncertainty attached to it.

You'll notice that this piece of information is not shown on the creation wiki. Each of the measurements they show contains an uncertainty, and the statistical procedure to interpret what the "best fit" is has to account for those uncertainties.

The question is *not* of "cherry picking results", but in looking at the entirety of the distribution, and not only the tails. In fact, given enough data, it is statistically CERTAIN that a given fraction of them will deviate from the mean (with a very prescribed distribution, typically with the Gaussian distribution described above).

When scientists talk about uncertainties in radiometric dating, we're not talking about 100% uncertainties (or worse). We're talking about percent-level uncertainties (certainly less than 10%). So even ACCOUNTING for these assumptions, there is no reasonable way to move from "4.55 billion" to "6 thousand" whatsoever.

You MIGHT get from 4.55 billion to 4.1 billion :), but even that's pretty generous.

Science is a quantitative field. It is the only field in which assumptions are dealt with as objectively as possible. Quantitatively, all observations are accounted for. The conclusion is still that the earth is *billions* of years old.

Scientists are not cherry-picking the tails of the distributions they want to agree with preconceptions. They are looking at the whole picture and properly accounting for uncertainties.

On the other hand, the people in the creation wiki ARE cherry picking specific samples. They are not showing how "often" a given result is found to disagree with the mean age, they just quote one or two that vary by a factor of two or so, and then conclude that it's possible that the earth is 6000 years old. This is just shoddy reasoning altogether.

Furthermore, NONE of the samples they came up with were in ANY way consistent with a 6k year old earth. If they look at one given sample, it might get them to 3 billion, or even 2 billion years (plus or minus
a billion or so). If you do this repeated times, over and over, you come up with the mean (via the central limit theorem, see above) which settles down to a value of 4.55 billion years, plus or minus a few percent.

There is literally zero chance that the current ensemble of observational evidence is consistent with an earth of 6000 years old.

So as I see it, you've got three options:

  1. Reject the core hypothesis that "what we observe, is reflective of reality", but I don't see how this is a tenable stance. You couldn't logically use materialistic items for any purpose (including this computer, for the most part). 
  2. Claim that even with all of these observations, the earth is 6000 years old anyway because God is merely testing our faith by blatantly misleading us with observational evidence. This is a subset of Option 1. 
  3. Accept the old earth model as "what is", because that is what we observe. 

I'm going with Option 3 myself!

1 comment:

Democurus said...

Another thing creationists often do, either unwittingly or maliciously, is to refer to inaccurate data as a result of using these tools outside of their intended scope.

It's like using a straightedge ruler to (inaccurately) measure the circumference of circle and then proclaiming that rulers are wrong.

Furthermore, different dating methods that rely on different assumptions, where their respective intended ranges overlap, they agree.

That means creationists aren't just saying that some dating methods are wrong, but that they're all wrong, but yet wrong in such a way as they still agree with each other!