In medicine," says Atul Gawande, "we've been obsessed with getting the best technologies, the best specialists, the best drugs. But that's a terrible way to design anything. Imagine putting together a car with Porsche brakes, a BMW chassis, a Volvo body, a Ferrari engine - you just get an expensive pile of junk that doesn't work.
"In many ways, that's what medicine feels like. We gravitate towards the best components rather than trying to put things together most effectively."
Traditionally, as Gawande puts it in his new book, The Checklist Manifesto: How to Get Things Right, "surgery has been regarded as an individual performance - the surgeon as virtuoso". Yet groups of autonomous specialists often produce "only a cacophony of incompatible decisions and overlooked errors". What we really need are first-rate teams.
With his remarkable range of talents and the looks of a medical soap star, Gawande is almost too good to be true. As a surgeon, he takes on about 250 cases a year. He is an associate professor at Harvard Medical School, training both medical students and surgical residents. He also runs a research centre at the Harvard School of Public Health.
And, just in case that doesn't sound like enough for one person, he is also a staff writer on The New Yorker magazine - a position many professional journalists would kill for - with a rare gift for conveying the pleasures and pains, the practical and moral dilemmas, of life in the operating theatre.
"As I entered medicine," Gawande observes, "I felt as if I was running into more and more questions that didn't have ready answers and, if I was going to be good at my job, I needed to find a way to work them out. Writing helped me grapple with the inevitable messiness of trying to bring science to the complexity of individual lives."
He points to a number of crucial concerns that are seldom discussed in teaching and textbooks: "How you handle a mistake, how you understand fallibility and the limits of your own ability, how we cope with the rising levels of information we're supposed to digest. The biggest areas of taboo are often those where we have the most discomfort." They also represent some of the central challenges for medicine and medical education.
Gawande's second book, Better: A Surgeon's Notes on Performance (2007), includes an account of a two-month working trip he made in 2003 to India, his family's ancestral home, just before he "officially began life as a general and endocrine surgeon".
"What is your preferred technique for removing bladder stones?" he is asked at one point.
"My technique is to call a urologist," he replies.
Such American levels of specialisation and "superspecialisation" are, of course, an unimaginable luxury in much of the world, where doctors constantly have to improvise new techniques with wholly inadequate tools and resources. Yet everywhere, notes Gawande, and perhaps especially in the West, medicine remains "troubled by human failings" such as "avarice, arrogance, insecurity, misunderstanding". Even as healthcare advances and ignorance retreats, mistakes resulting from ineptitude remain tragically common. Many of them arise from slips as basic as doctors failing to wash their hands.
The figures are simple but terrifying. Gawande's new book cites research indicating that "the average patient required 178 individual actions per day, ranging from administering a drug to suctioning the lungs, and every one of them posed risks. Remarkably, the nurses and doctors were observed to make an error in just 1 per cent of these actions - but that still amounted to an average of two errors a day with every patient." Bringing that figure down could save thousands of lives.
In 2006, Gawande began to devote serious attention to this goal, when the World Health Organization asked him to help develop a global programme to reduce avoidable deaths and harm from surgery.
Part of the answer, he now argues, comes from checklists. This hardly sounds like a gripping topic for a book, but The Checklist Manifesto makes the case for a surprisingly radical revision of how we see doctors, how they see themselves and how they are trained.
Something of the rationale is captured in a great story about a rock star who demanded a bowl of M&Ms backstage at every concert, but with all the brown ones removed. This may sound like the most tiresome sort of narcissism, but Gawande suggests it was much more reasonable than that.
Setting up in a new stadium required 18-wheel trucks and vast girders that could easily cause lethal accidents. The contract contained hundreds of safety clauses, so anyone who missed the one about "no brown M&Ms" couldn't be trusted to get everything else right. There is method in the madness even of "power-mad celebrities" - the method of the checklist.
"Pilots in training learn that they are fallible," says Gawande, "and that they can counteract it only with the discipline of procedure, the discipline of checklists, to make themselves more effective.
"In medical schools, we don't teach that. It's all about technology and knowledge, not about how you cope with inevitable failures.
"We don't train people for working in teams. It is only at rare places such as the University of Nevada, Reno that they combine the medical and nursing schools."
One of today's key tasks, therefore, is "replacing the cowboy" - in the basically positive, American sense - "with the pit crew".
The kinds of checklist Gawande envisages are designed to prevent all-too-common elementary errors (opening up the wrong side of a patient, getting drug dosages wrong, failing to have the right equipment to hand). But they are also based, he says, on "a deeper set of values".
"The checklist is a cultural-change device. In order to be effective, it brings about a change in status for everybody in the room. If people introduce themselves before the operation, it makes them much more likely to speak up later if they are puzzled about something. People are allowed to voice their doubts, which is an important part of having a successful team."
Yet anything that flattens the hierarchy in the operating theatre tends to attract fierce opposition. Many surgeons strongly resist the idea of introducing themselves to unfamiliar members of the medical team.
Can we put this down simply to the arrogance of surgeons who, as the saying goes, are "sometimes wrong, never in doubt"?
"Of course we have arrogance!" Gawande replies. "You have to have self-confidence to go into a room and cut somebody open and tell them they will be better off - especially since you know things will sometimes go wrong.
"If you didn't have a degree of arrogance, you'd be a poor surgeon. But if you didn't have a degree of fear, you'd also be a poor surgeon. Surgeons who have no fear have an overweening confidence - and that can be dangerous."
Yet alongside the arrogance of some surgeons, Gawande points to a wider public need for lone, self-sufficient heroes just at a time when individual autonomy has become an often dangerous ideal.
Take the case of "the miracle on the Hudson" in January 2009, when Captain Chesley Sullenberger managed to crash-land his plane in the river with no loss of life.
"We wanted to see Sullenberger as the pilot who improvised everything about saving that flight and therefore single-handedly rescued those people," says Gawande. "But he said it was about adherence to protocol and teamwork.
"That wasn't just modesty. He and his co-pilot went through a series of checks even before they left the ground and walked through what would happen if the engines failed. Because he could trust others to play their roles, he was able to focus on the things for which he did need autonomy, notably where he was going to land the plane."
So both the public at large and doctors themselves need to grow up and stop seeing doctors as superhuman. But there are also crucial lessons for medical education and research.
"We have pushed towards operations that require incredible levels of skill as we minimise incisions, using laparoscopic and then robotic devices," observes Gawande. But while medical schools love turning out "star" surgeons who can perform such demanding procedures, this misses the crucial point, he says, that "for an operation to be done well across the entire country, it needs to be something that can be readily replicated.
"We've focused on the results in the best of hands rather than asking about the actual results in the broader world. We don't measure how many deaths have occurred because of complications in surgery across the whole nation. It's not just about how we teach, but what we teach, and how we design medicine to actually work."