Recently, New Scientist published a feature on evolution: “Find The Gaps – The Hunt For Evolution’s Missing Links” by one of their longstanding favourites, the biologically unqualified Jeff Hecht. It tried to sum up the evolutionary trail by highlighting moments of transition from one kind of animal to another. Hecht went to the astonishingly inept David Hone for advice on bird-dinosaur evolution. That led to the first of the science disasters that followed…
In the historical sciences, much journalism, as with New Scientist, is a slave to any current dogma. (It may be true of experimental sciences too, but experiments tend to protect against folly.) New Scientist’s missing links piece deals with Archaeopteryx, perhaps or perhaps not the first bird, and it reminds us how much David Hone is still a joke that keeps on giving. New Scientist asks what was Archaeopteryx’ ancestor, and indeed I ask and answer this in section 5.5 of my book. Are there any transitional fossils that precede it? Oh no, New Scientist says. The nearest transitional fossils came after it. (Hmm!) But what would Archaeopteryx’ ancestor look like? we ask. Might it be something that lived, say 10 million years before, and resembled Archaeopteryx but with, perhaps, less aerodynamic wings or feathers?
Well, David Hone doesn’t think such a fossil could in any way be relevant to the search for Archaeopteryx’ ancestor. But amazingly, not only is there such a fossil – Anchiornis – perfectly poised in every way to be Archaeopteryx’ ancestor, but, Hone himself was one of the co-authors of the paper that first announced it!! Is it possible he forgot it? (Maybe, since shortly after the paper came out, he claimed it was the first feathered troodont to be found, showing he’d forgotten the highly memorable Jinfengopteryx.)
The probable explanation is worse than that though. It’s because Hone is following this pattern of behaviour:
First, come up with an initial theory, in this case to explain the evolution around Archaeopteryx. A good scientist might well interrupt here, and say: “Then look for clashing evidence, so the initial theory, the null hypothesis, can be rejected and an alternative, or its opposite, accepted?”
Unfortunately, we must remember we are dealing with dinobird palaeontologists here. So the line Hone follows is instead:
First, come up with an initial theory.
Then, find any pieces of evidence consistent with your theory, and claim, perhaps in a paper for each piece of such evidence, that it is proof of your original theory.
Next, blame the absence of evidence/fossils that your theory predicts, on the quality of the fossil record. Also, ignore evidence/fossils that are found, and were predicted by competing theories but not by yours. In fact, ignore all competing theories, as if the process of science were not to gather candidate theories and then sieve them through the evidence to see which survive.
Finally, whenever anyone points out that you’re doing bullshit science, block their papers from publication, block their comments from your blogs, and answer criticism by simply announcing to onlookers that the critic has been blocked, and hope conclusions will be drawn.
In fact what Hone, Hecht and New Scientist should have said is:
There are two theories on the ancestry of Archaeopteryx: One predicts bird-like flightless dinosaurs prior to Archaeopteryx, which we do not find. The other predicts small feathered arboreal gliders prior to Archaeopteryx, which we do find. We therefore provisionally conclude that the small feathered gliders were the ancestors, which we can report to the public and their institutions who have bought New Scientist expecting sound science.
So far, so bad. That’s an old error, but New Scientist has recently jumped on a bandwagon heading straight up a new wrong gumtree. I could be even more florid but I already have been. This new fiasco, included in Hecht’s survey, concerns the date of that most important evolutionary event: the split between chimps and humans. As that depends on the number of genetic differences between us, its date depends on the speed at which the genetic clock tends to insert those differences over the long term.
A recent paper counted the number of genetic differences between parents and their offspring in humans, chimps and gorillas*. Does this give the average speed of the genetic clock across long time spans?
Well, we already know it’s unreliable, and the reason is this: when you go on a walk, you can go straight, or you can wander around at random. If you walk at random, your feet may have taken 1,000 steps, but you may end up only 30 paces from your starting point.
Conversely, a species with a large number of individuals is a massively parallel computer, exploring dozens or perhaps thousands of new mutations every year. An individual mutation in a particular offspring is unlikely to get established or even survive, yet each year the genepool as a whole discovers and establishes many viable new mutations. These get preferentially selected for, and inherited by your later descendants but not from you. Therefore, the overall mutation rate for the species could be a lot higher than the average mutation rate from individual parents to their offspring. Indeed, the speed at which genetic algorithms find solutions is known to increase in larger populations. Unless this is entirely unconnected with mutation rates, it will be impossible for the long-range rate to depend only on the individual rate.
Combine just these two considerations, and we see that you simply cannot take the result of that average mutation rate from parent to offspring experiment, and plug it raw into your evolutionary models. The true long-term rate might be faster or slower. In fact it is already known to have a variety of rates at various scales.
But New Scientist never consulted anyone who both understood these considerations and was prepared to insist that they were taken into account. They did consult someone who perhaps might have done, John Hawks, who at about the same time, was recommending never to listen to anyone who considers the philosophy of science, in case you yourself might become associated with views not held by the majority. (This suggests you should never dare listen to or support any new idea still at the stage of being held by a minority… which in turn implies no new ideas should ever be supported, but I suppose that itself is a wisdom arising from philosophy of science…)
So New Scientist swallowed the new clock raw and whole, and decided on a chimp/human split of 10 million years (instead of say 4.15 million years, which actually fits beautifully with everything).
Now here’s where their article gets really stupid:
Orangutangs split away long before chimps: somewhere between three and four [Update 8Mar2013: actually 3-5] times as long ago. This means that New Scientist’s chimp split of ten million years would imply an orangutang split of 30-40mys.
Unfortunately, we didn’t lose our tails until about 23mya with the arrival of Proconsul. That gap from 23 to 30 or 40mys is a bit big, but it’s much worse than you might think. First of all, we’re really talking closer to a 23-40my gap, not a 23-30my gap, since orangs are probably nearer four times further away from us than chimps. And secondly, what about gibbons and siamangs? Their branch split off even earlier than orangs. New Scientist’s story hides the inescapable but impossible implication that gibbons split off nearly twenty million years before the first apes evolved, and then became apes on their own, and then orangs split off at least seven million years before the first apes evolved, and then became apes on their own as well.
How can anyone be so stupid? Here’s how. It always seems to require a correction-prevention-mechanism of some kind. The one New Scientist uses is: only consult those job title includes the name of the subject concerned. Sounds reasonable? Well, not only does Hecht have no relevant qualification, but Hone has none that relate to the information systems on which he bases his phylogenetic opinions. So they can’t even follow their own system of only asking “experts”. And of course they never accept their errors of scientific procedure are errors because John Hawks warns that experts in that area are kooks.
But the main problem with their selection strategy is not that they use it wrongly but that it gives worse results than consulting scientists at random. Any idiot down the pub can see that New Scientist’s reportage is rubbish, and you or your institution should expect more than a £3.70 chip wrapper from New Scientist.
* Langergraber et al. Generation times in wild chimpanzees and gorillas suggest earlier divergence times in great ape and human evolution. PNAS July 13, 2012