Friday 28 June 2019

A scientist's guide to accuracy

Because I lack for nothing, I am notoriously difficult for giving presents to. Being on a decluttering jag these last several year doesn't make it any easier. Last December I was in a book-shop looking for inspiration for outgoing Christmas presents, when I saw a short stack of Humanology; A Scientist's Guide to Our Amazing Existence by Professor Luke O'Neill. O'Neill is one of the few Irish scientists that has a convincingly international reputation, so it's good that the book was put together by Gill Books one of Ireland's few independent publishers. Not so good that Gill catalogued it under the Humour section because that allows the author to defffffflate too many of his explanations about the science of the human condition with a throw-away line that is occasionally humorous but also occasionally just silly. We're in the same trade Luke and me and we spent 5 years working in the same Department - where he was conspicuously more successful by all the metrics that you can apply to teaching and research in science. We both believe that Science Matters (it's The Blob's official name, after all) and that someone needs to bridge the gap between science and The People, because that matters too. I have been known to crack a joke, even on The Blob, some of which are even funny on re-reading, but I've never set out to play it for larfs as the default for making science palatable or understandable.
Q. But what do I know about publishing?
A. Me, nothing. But Amazon knows about publishing, so I mined some data:
  • Sapiens: A Brief History of Humankind by Yuval Noah Harari #16 in books 
    • (#1 in Human evolution)
  • Hello World: How to be Human in the Age of the Machine by Hannah Fry #2,146 in Books 
    • (#1 in Algorithmic programming)
  • Humanology -  of which we treat #14,632 in Books 
    • (#13 in History of the Renaissance)
  • How to Grow a Human: Adventures in Who We Are and How We Are Made by Philip Ball #115,556 in books 
    • (#97 in Human Evolution)
I tell my students to go easy on the superlatives when writing up their results. Merely saying that something is stupendous or unparallelled doesn't make it so. Copy-editor me would have shortened A Scientist's Guide to Our Amazing Existence to allow the text to speak for itself. And being told that the author of any book is 'hilarious' certainly puts a damper on my chuckle-button. Which is ironic because Chapter 8 of Humanology is an analysis of laughter which points out that you can't tickle yourself and that being told to laugh doesn't work.

Having unloaded my baggage, I was able to start reading the book. But I was brought up all standing at the bottom of p.2 by a discussion of the age of the Earth as a prelude to making the case that there has been helluva a lot of time to make primroses and primates. "Or 4.28 thousand million years? Such time spans are well beyond our comprehension. If humans had appeared at that time (and they didn't) there would have been around 55,000 generations of us since then." Ahem, no there wouldn't, not unless a human generation is [4.28 x 109 / 5.5 x 104] = 77,000 years! And even allowing for a gross failure [what's a thousand here or there?] on copy-editing, 77 is the current human life-span which is not the same as the generation time: usually taken as 25 or 30 years - between successive zygotes.

To paraphrase Lady Bracknell to get one number wrong (in two quite separate ways) is a misfortune, to get two wrong sounds like carelessness and I didn't have to read beyond p22 to land the second out-by-a-thousand figure, in a timeline of the evolution of life:
  • 1.2 BA - Multicellular Organisms
  • 600 BA - Cambrian Explosion
  • 200 KA - Humans
(where BA is billion years ago and KA is thousand years ago). 600 BA is far longer than we think the Universe has existed. 0.6 BA is meant, or 600 MA.

Does it matter beyond a pedant's excuse to snipe? Yes, in the sense of Fodor's Guide to Credibility which I have described before as an aid to crap-detecting. If there are obvious, checkable errors in one part of a book about which s/he knows something; then the reader is justified in being super skeptical about statements elsewhere in the same book about which s/he/we know nothing. But for the sunk-cost of €18.99, I should have flung the book across the room in a pet and found something else to read. It's like Trish Greenhalgh's advice about How To Read a Scientific Paper: start with the Methods section - if that's dodgy, then the results are worse than useless they are a waste of time.

I didn't get [too] cross. I did finish the book. It was okay. I may get round to reviewing it or at least filleting out some of the interesting bits and doing my own research for a future Blob. Right at the end of the book, life expectancy at birth for Ireland is referenced for men and women. That seemed redundant because I felt sure those statistics had been mentioned earlier in the book. There is no index, so I had to back-track (big advantage to having Kindle edition!) to find . . . different numbers!
Page
p.259
p.322
M
79.4
83.4
F
78.3
82.7
I don't know what the error of estimate is for these numbers but I'd probably advise my students to skip the decimal point as "spurious accuracy" and call it 79 years for men and 83 years for women. That puts us in the same ball park as the rest of Western Europe and much longer lived than people in Côte d'Ivoire, who peg out about 20 years earlier.

Stephen Hawking said of his own worthy but widely unread book A Brief History of Time. "Someone told me that each equation I included in the book would halve the sales. I therefore resolved not to have any equations at all. In the end, however, I did put in one equation, Einstein's famous equation, E=mc2. I hope that this will not scare off half of my potential readers".  Maybe Gill Books should have scrapped all the numbers in the book where they are just a hostage to fortune in an industry where copy editors typically gave up formal maths when they were 15.

No comments:

Post a Comment