I’m nearly finished Bryson’s book, A Short History of Everything, and things don’t improve: the more we get about the history of human beings and their supposed forebears the more speculative it gets. This isn’t Bryson’s fault: he’s merely reporting what goes on. But it doesn’t say much for the so-called science that’s involved – or, perhaps rather, it should be called the speculation. Just as with the dinosaurs, where incredible beasts have been conjured up out of insufficient bones, so with the various versions of bipeds extraordinary suppositions have been made – and it gets worse when these scientists start taking DNA and claiming to work out who came from who and how ancestors we actually had, and even, in one case, in the book, The Seven Daughters of Eve, the writer (Bryan Sykes) imagines detailed personal histories, things that he can’t in any conceivable fashion know. As Bryson says to Rosalind Harding, a population geneticist in Oxford, ‘So genetic studies aren’t to be trusted?’ She replies: ‘Oh you can trust the studies well enough, generally speaking. What you can’t trust are the sweeping conclusions that people often attach to them.’ When she says, ‘people’ she doesn’t mean Joe Bloggs in the street, she means scientists.

And this is the same for what we know or don’t know about the earth’s history: huge amounts of it are speculation. Just as whole areas of evolution theory are speculation. We don’t know, because we can’t know; we don’t have anywhere near enough evidence. Science is supposed to be a game where you test your hypothesis by practical experimentation. When you can’t do this – as in the age of the earth or how humans got here – then you’re stuck, and it seems that an awful lot of scientists invent stuff to fill in the gaps. Which really makes them no different to writers of fiction. It’s all very well popularising science, as long as it remains science. It’s when it’s just popular that you have a problem.

Found an interesting quote from Rosalind Harding:
On getting a poor answer faster:"In population genetics there is usually little reason for confidence that an estimate is correct even to within an order of magnitude, but reaching it faster is definitely progress."
Reference: J R Statist Soc B (2000) Vol 62, Part 4, page 638.