[Originally published as Is Baraminology Bogus? Editor’s note: The following is a discussion of how some creation scientists approach the classification of living beings. It allows us to see behind the scenes into the ongoing philosophical and research-related efforts that demonstrate we are nowhere near being done understanding how God created. Part 1 HERE]
To the question of the methods themselves, particularly distance correlation, it is quite easy to use baraminology in an arbitrary fashion that yields absurd results. This was a subject I discussed in detail with colleagues twenty years ago.
The question is deceptively simple: What is the created kind, and where should we look for them?
After a detailed study of the Hebrew term min (“kind”), we became skeptical that min was ever intended to be a technical category, a biological “created kind.” But looking at the Genesis description of creation, I could see a few important features.
First, Genesis 1 mentions no species by name (except maybe human beings), so when we look at evidence of species change, the Bible didn’t seem to prohibit that per se. At the same time, Genesis 1 does mention large groups of creatures such as wild animals, beasts of the field, flying things, creeping things, and swimming things.
The picture at the conclusion of creation week is one of a fully functional creation with an intact ecosystem of multiple sorts of animals. That would preclude the evolution of all things from a common ancestor, but I think we can probably get even more specific.
For example, we might think that “creeping thing” could include rats and mice, which are rodents. But other rodents are much larger, like beavers or capybaras, and maybe they wouldn’t be creeping things?
So maybe the mammalian order Rodentia is really composed of multiple baramins? Now that’s a bit fuzzy and speculative, but you can do similar reasoning with other groups.
“Beasts of the field” would probably include cows and sheep, which are members of Artiodactyla. They would presumably be separate from other artiodactyls like the deer, which are wild animals.
Putting all that together, I thought maybe something between species and order should be where we look for the created kind. Since I know that members of different genera can cross and produce offspring (and are therefore members of the same created kind), I further narrowed my search to families.
And that’s consistent with historical claims of creationist scholars in previous generations. Not all of them of course, but there were those who believed the created kinds were approximated in today’s world by families. That gave me a context of where to look for created kinds. Are they always families? No, but it’s a place to start looking, and it seems to come up a lot in baraminology research: Canidae, Equidae, and Felidae are all mammal families that we suspect are created kinds.
Back to distance correlation, I’ve long acknowledged that any set of characters and any set of taxa might produce a clear set of clusters that may not have anything to do with created kinds. If I’m comparing a human, aardvark, and rat to a heron, robin, and cardinal, then I’m likely to find two clusters with humans in the same cluster as the other mammals.
But that is not a created kind. It’s just a cluster.
Likewise, I could very easily find a set of characteristics that would distinguish European people from Asian people, but again, that would obviously be silly. Why? Because we have independent reasons for where we think created kinds will be, and we don’t just find clusters and call them “baramins.” That’s wrong.
In the same way, merely finding significant, positive distance correlation does not automatically mean that the two taxa being compared belong to the same created kind. And if you have significant, negative distance correlation, that does not automatically mean that the two taxa are in separate created kinds. That’s an oddly common misconception that is emphatically wrong (and always has been).
Distance correlation only tells you one thing: whether or not the distances from two different taxa to the same third taxa are similar or different.
- That, in turn, might tell you something about whether they are close together or far apart in character “space.”
- And if you see certain patterns of positive and negative correlations, you could conclude that a particular group of taxa forms a cluster that is separate from other taxa.
- And then, if and only if that group conforms to what we independently think might be a created kind should we ever conclude that a cluster probably represents a created kind.
Otherwise, it’s just a cluster. (We might also conclude there is a discontinuity between groups of created kinds if we’re looking at clusters that include members of multiple created kinds.)
That said, is it appropriate to apply statistical baraminology and specifically distance correlation clustering to a group of taxa that do not conform to what we think is a created kind? Certainly, but what we might find is kind of a mystery right now.
I’ve had people repeatedly tell me that we can find “discontinuity” or “created kinds” at every taxonomic level, which at this point, is pure speculation. We don’t have enough data yet to make that conclusion. My experience has repeatedly shown that distance correlation within a created kind tends to produce poor clustering. Same deal with distance correlation with many created kinds. There are counter examples, of course, but the question here is the trend: What do we see when we look at a lot of these kinds of character sets? Right now, we don’t know yet.
Speaking of counter examples, a single counter example rarely falsifies much of anything in science. It’s possible that a particularly well-established counter example will overturn a model, but in the nebulous area of cluster analysis, where specialists can’t define what “cluster” even means, counter examples merely require a more careful examination to see why they’re counter examples.
Let’s talk about a few hypothetical examples. Let’s say I selected a good set of characteristics from a sample of taxa that included a dozen different primate species from a dozen different families. What will distance correlation show? It might show a mess with no clear clustering, or it might show clustering.
What it probably won’t show is a created kind. Why? It’s a clustering method that compares distances within a group to distances between groups. If I have only one taxon in each group that I think is a created kind, then there are no within-group distances to compare. In that case, distance correlation can only show us clustering of multiple created kinds.
Or perhaps I want to look at the characteristics of different groups of the same species. Maybe I want to look at twenty different breeds of cattle. In that case, whatever clustering I find is not a created kind, because I have good independent evidence that tells me that all cattle belong to a single created kind. Likewise, it is erroneous to focus on individual distance correlations. Distance correlation is all about finding clusters, but there are a number of different reasons that correlations might occur.
Often, when doing distance correlation on a single cluster, I find a random hodgepodge of positive and negative correlations, or I find poorly resolved clusters that share a lot of examples of significant, positive correlation. Should these results influence the way I interpret clusters that I think really are created kinds? They might, especially if we find lots of examples of putative “created kinds” within groups of taxa that we believe represent single created kinds.
So that’s all I have time for right now. In closing, none of what I’ve written here is new. It’s all explained in previous publications dating back twenty years. I’m not saying anything here that’s all that different from what I’ve put in Understanding the Pattern of Life and papers published since then. I’m merely trying to explain how I have always understood the work I’m doing.
One more question: Is baraminology bogus? The jury’s still out, but it’s not nearly as flawed as my critics believe.
Now I need to get back to the work that I’m actually paid to do. Thanks for reading. May the Lord bless you and keep you and give you peace.