evidence based

Pundits can be a pain in the fundament. I am a big fan of the BBC podcast Sideways which allows Matthew Syed to talk about people who march to a different drummer; who look at the world that standard media and Jan Modal might consider a bit out there. If we all think alike, it may be cosy and non-confrontational but it might also be bzzzt wrong. The podcast is fun because it explores some peculiarity of the human condition, develops some makes-ya-think ideas,ties everything up with a Message and signs off after 30 mins. Malcolm Gladwell, with his breezy certainties, and tidy stories, is an exemplar of the genre. As some wag had it "My take on Gladwell: his books should have been magazine articles and his magazine articles should have been tweets." 

But now I've read Matthew Syed's second book Black Box Thinking: Why Most People Never Learn from Their Mistakes (2015). It's a book in form but it's really journalism in substance. Which is fine - chacun à son goût métier - so long as you don't expect reporting on science to be, like, science. Scientists are generally really crap at explaining their ideas to ordinary folks; so there's defo a niche for Syed and Gladwell and Jared Diamond [whose breezy hypothesis about the end of civilization on Easter Island was just demolished on MeFi]. Diamond also had theories about the end of Viking Greenland [blob].

Syed contrasts the response to mistakes were made in the aero and health industries. But there are many examples from business and politics. Key message is that some corporate cultures make it very difficult to admit error so there is no learning opportunity. Nobody likes to be wrong and we tend to double-down if errors are pointed out. If management mistakes are re-framed as "shit happens" and "nothing to do with me" then the errors repeat. 

When a plane goes down, by contrast, investigators will go to extraordinary lengths to work out why. The black box for Air France 447 Rio-Paris flight was recovered from the mid-Atlantic floor 4,000m below the surface, 2 years after the plane crashed. Despite the headlines, air-travel is famously safe: and this is because pilots and engineers fess up when things go pear-shaped. Reports are filled and submitted and collated and analysed. Most importantly, international warnings, recalls and advisories go out to minimise the chance of something similar happening again [ever]. Here's another near [1.3m = 5ft] miss story analysed in the book]There has been much recent 737-max criticism of Boeing for turning a premier engineering company into a shareholder's company. 

When you tally up the number of avoidable iatrogenic deaths it is, according to Syed quoting a 2012 epidemiologist referring to a 2000 report,  equivalent to two full jumbo-jets falling out of the sky every 24 hours [USA data]. An arresting image, indeed. But more recent analysis on much larger samples specifically looking at the problem of medical error finds that two full jumbo-jets might be 10x - 20x too high.  Nevertheless even if only 5% of deaths in hospital are due to mistakes, it's still too high. And, Syed sez, nobody is comparing notes; let alone sending reports, analysis and conclusions to a central repository. Au contraire, when they killed my father in 2001, his hospital records mysteriously disappeared. We see this all the time: when babies die or are incapacitated in/by hospitals, parents usually have to lawyer up to get to the bottom of it. Mostly, they just want an explanation, an apology and some assurance that no parent will have to suffer like them in future. It's ugly and makes it all about the compo; as if a €million will get their baby back.

Afrique. Later on in the book  Syed re-tweets a story from Tim Harford [multoprevo] about evaluating interventions to better educate young Kenyans. Aid agencies have $50 billion to spend in Africa each year. Harford's useful way of evaluation such large numbers if to divide them by the relevant population. If distributed evenly, that's  a little less than $1 a week each. Anyway, the study in Kenya was effectively a controlled experiment or randomized controlled trial RCT. They found that 

• lurrying textbooks [in English] into schools made no difference
• lo-tech flip-charts with engaging brightly colour pictures ditto brrrp! 
• otoh, dosing all the kids (and the teachers?) with anthelmintics to reduce the parasitic worm burden perked everyone up, ready to learn and less likely to pull a sickie. Result: demonstrable improvement in the LOs Learning Outcomes.
  

Don't believe a word of it until it's been replicated in Cameroon! Nevertheless it's probably better if you read (some of) this book than spending the equivalent amount of time restlessly swiping left-right or slack-jawed watching youtube. As Gladwell said of his own works "The mistake is to think these books are ends in themselves. My books are gateway drugs—they lead you to the hard stuff." Which is characteristically too clever by 'arf.

!! But why waste time with secondary sources when you can get the key ideas from the horse's mouth? I have been recently quite the fanboi for Sean Carroll's Mindscapes podcast. Mindscapes#1 was with Carol Tavris the co-author of  Mistakes Were Made (But Not By Me) [2007] on Mistakes, Justification, and Cognitive Dissonance. Also on YT.
Five years and 200+ episodes later Carroll returned to the theme in Mindscapes#233 with Hugo Mercier (co-author  of The Enigma of Reason [2017] on Reasoning and Skepticism. Also on no pics YT. In my reality, I absorbed these two hour+ blasts of erudition within a couple of days of each other. So it was a bit like binge-watching House on lupus in an all-night session.