Monday, June 29, 2009

If We Fixed Cars In the U. S. the Way We Fix People

You would typically get auto insurance (including insurance for major repairs) as an employment benefit, and you'd never see the money. If you lost your job, you'd either have to buy very expensive insurance on the open market, or just hope your car didn't break down before you got another job.

If you had insurance, all you'd have to pay at the repair shop would be a small copayment of twenty or thirty dollars. The insurance would cover the rest.

Many people with a car problem would first have to take their car to their primary care mechanic (PCM), no matter what was wrong. If it was the brakes, your PCM would have to write you a referral to a brake specialist before you could get your brakes fixed. Of course, you could just pay for the repair yourself, but that would cost many thousands of dollars.

You could do minor repairs yourself, but for something major like a new starter, you couldn't just walk into the parts store and buy a new one. You'd first have to get a handwritten note from your PCM. Then you'd take the note to a special parts store that has college-trained salespeople licensed to sell the higher grades of auto parts, and buy the part from one of them.

Every auto repair shop would have huge filing cabinets with multicolored file tabs sticking out of them, one file for each car. All the important records on your car would be written by hand.

Minor repairs would be made in small shops, but major repairs would take place in a few giant facilities in each metropolitan area, where hundreds of cars would be collected together for repairs. Getting an estimate for repairs at one of these mega-shops would be next to impossible. Instead, you just take your car in, wait till they fix it, and hope your insurance will pay for it.

Reading a repair bill would be an exercise in mystification. Even professional accountants couldn't explain why the repair shop would charge $674.92 for replacing a U-joint, the auto insurance would pay $407.17, and you would end up owing $103.37.

Mechanics would check for the most unlikely problems even when the issue was a simple one. You might take your car in because the wiper blades needed replacing, and the mechanic might do a stress test on the windshield and charge the insurance company $400.00 for it. But you could rest assured that your windshield wasn't about to shatter spontaneously, which happens maybe once in every 30 million passenger miles.

Auto mechanics would have to go to school for six to eight years after their BS degree and pay back hundreds of thousands of dollars in school loans before getting their license to fix cars. But they would earn almost as much as lawyers do, if they could keep ahead of their own soaring insurance premiums. If you didn't have a government-sanctioned auto repair license and you tried to fix anybody else's car, you could go to jail.

Gasoline additives would be a multi-billion-dollar business, funded with a combination of government support and private money. Universities would have special auto-repair branches and auto-repair research departments to develop new additives and repair techniques.

Mechanics could be hauled into court and made to pay millions for faulty repairs. Mechanical malpractice law would be a lucrative branch of the legal business.

Eventually, things would get so screwed up that the U. S. government would wade into the mess and take over large portions of it, promising to manage things better than the former private owners did. Of course, that's exactly what has happened with the U. S. auto industry already—so maybe it's time to wind up this little fantasy. . . .

Health care in the U. S. is a complex system that can be viewed as a technology. And it is fraught inside and out with ethical implications, so I think the current debate over what should be done about health care is fair game for an engineering ethics blog, broadly defined.

Of course, people aren't cars, and the analogy between car repair and health care breaks down if you examine it seriously. But sometimes casting a familiar situation in a new light will reveal problems in a way that more people can understand. For example, it is a fact that most people don't know they pay an average of $400 a month for health care—either as a deduction from their paycheck, or a contribution from their employer, or typically both. If you had to write a health-insurance check for $400 from your own pocket every month (as some self-employed people do), that single change would bring the reality of health-care costs home in a way that no number of TV ads will do. Most people do pay for their car repairs out of their own pocket. If we had the direct price information that would let us pay attention to the quality and price of health care in the same way we evaluate an auto mechanic's services, it would do a lot to reduce needless health-care expenses.

Eliminating the employer tax deduction for health-care costs, if done the right way, will be a step in this direction. But not if it is done just as a way to raise revenue and make a government-funded public insurance option more attractive.

As the debate develops, it seems to be coming down to a single question: Who can exert more effective discipline on the health-services sector: a set of government bureaucracies that will ration and regulate the system, or an enlightened public which is allowed to see the costs of health care directly and make their own judgments as to which insurance plan and caregiver is best for them? The latter option still has a role for government, possibly as the provider of last resort for insurance or services for those who can't afford them, and as a check on rampantly malicious behavior on the part of powerful institutions.

Either there will be major changes in the way health care is delivered in this country by later this fall, or there won't be. Sometimes no change is better than a bad change. After all, we could be worse off. Surveys indicate that most people are reasonably contented with their present state of health care. The trick will be to help the groups who are under-served without worsening the status of those who like what they have already—and not do something else awful like run the government into bankruptcy in the meantime.

Monday, June 22, 2009

Should You Worry About Oil Sands?

Gasoline prices have been going up in recent weeks, and one reason they haven't gone up more is oil sands in Canada. As Michael Levi, writing in the current issue of Slate, points out, not too long ago Canada took over from Saudi Arabia as the country supplying the most imported oil to the U. S. And a big factor in Canada's oil exports is their production of oil sands.

It turns out that under a large chunk of northeastern Alberta, Canada, the Athabasca oil sands deposit harbors the oil-sands equivalent of the rest of the world's conventional oil reserves combined. What are oil sands? Sometimes referred to as "bitumen," the hydrocarbons in oil sands are tar-like substances that are too thick to flow out of the formations they are found in, as conventional petroleum does. Instead, you have to dig the stuff out with conventional open-pit mining techniques and then process it to make what is called synthetic crude oil. From that point on, the stuff can be treated more or less like regular petroleum.

Because of the added complexity of extraction, oil sands have only recently been exploited commercially on a large scale, and they still form only a fraction of Canada's total oil output. (Venezuela also has a large oil-sands deposit, but the government of Venezuela is not as favorably disposed toward the U. S. as the Canadian government is, to say the least.) Oil sands have received some black environmental marks, both for the relatively large amount of greenhouse gases that result from producing a barrel of oil-sands petroleum compared to a barrel of conventional oil, and for the problems that used water and mine tailings cause.

So what is the right thing to do about oil sands? Do the environmental issues dominate and make us swear off them altogether? Or should we just arrange some very long-term contracts with Canadian produces and quit worrying about the shrinking oil reserves in the rest of the world?

As Levi points out, neither of these extreme alternatives is wise. While the production end of oil-sands operations does make more carbon dioxide, you still end up burning oil at the consumption end, so economies or conservation elsewhere in the system can make up for the additional burden at the production end. Pollution of water and destruction of land due to poor open-pit mining practices are concerns, but if I know our Canadian friends, they are on top of those issues and have found ways of dealing with them.

All the same, it would be the height of complacency to say along with the rich fool in the New Testament parable, "Soul, you have oil sands saved up for many years. Drive, drink, and be merry." I have said before that for reasons of national security, the U. S. ought to devise a long-term plan to move gradually and even profitably toward increased energy independence. With the global oil market the way it is, namely much like a pickup basketball game with no referee, we are taking chances with our economy by relying too much on unstable parts of the world for our energy supplies. Although some weird things go on in Canada, as a whole its government is a lot more reliable than, say, Iran's, so on balance it's a good thing that we are getting more oil from Canada than we get from Saudi Arabia or other Middle Eastern countries. However, the best outcome would be to move, deliberately and without serious damage to the economy, toward a situation where fossil fuels are gradually phased out in preference to an electricity-based energy economy, perhaps powered largely by nuclear plants. But that's just my opinion.

Nuclear has the dual advantage of not emitting any greenhouse gases and of using fuel that is not primarily found in politically unstable parts of the world. Nuclear plants use uranium, and some types can even produce more fuel than they consume. And it turns out that more than half of all the uranium produced in the world is mined in two countries: Canada and Australia, both of which we get along with pretty well.

The difficulty, as always, is how to implement such a plan in a democracy where short-term considerations tend to dominate political discussions. Somehow we can always agree to keep troops here or there for another few years, but we can't agree on a plan to reduce our dependence on oil so much that we wouldn't need to play global policeman so much. The new administration has made efforts in this direction, but with so many other irons in the fire, energy independence is going to take a back seat (to mix a few metaphors).

In the meantime, we can be grateful to our Canadian friends for developing oil sands in an environmentally responsible way. For the next few years we will need to buy oil from somewhere outside the U. S., and Canada looks like a nice place to get it from.

Sources: Michael Levi's article "Living on Canada's Oil" can be found in Slate at I also used material from the Wikipedia articles on oil sands and uranium.

Monday, June 15, 2009

Health Care as Systems Engineering

This summer promises a great debate over health care in the U. S. It is pretty sure to be great in the sense of historic or significant; what is not so clear is whether the outcome will be great either in the sense of good and positive, or in the ironic sense ("Great! That's just what I needed!"). One perspective that may help us judge the quality of the debate and the outcome is to view health care through the lens of systems engineering. But since health care deals with the most intimate aspects of human life, ethical considerations also show up at all levels, from individual decision-making to nationwide policy.

It is well known that the U. S. pays more per capita for health care than most other industrialized nations, but by many measures we are not that much more healthy than the other countries are. In other words, we're not getting what we're paying for, if we think that spending more health dollars per person will make everybody that much healthier. Viewed as a system with inputs (health care money and resources) and outputs (people treated by the system), our system does not work as efficiently as it could.

Over the last several weeks I have read of certain parts of the country where the health outcomes per Medicare dollar (which is an easy statistic to obtain) are much better than the national average (e. g. Green Bay, Wisconsin, where President Obama recently spoke on the issue), and other areas where they are much worse (e. g. McAllen, Texas, which a recent New Yorker article identified as in the second most expensive county in the country, measured in Medicare-dollar-per-patient terms). One reason for these identifications appears to be the idea that if we can just figure out what the good places are doing right, we can replicate these successes and do away with whatever bad or evil mischief is going on in the expensive places. How likely is that to succeed?

Atul Gawande, the author of the New Yorker piece, thinks the root problem in McAllen lies in the disconnected, revenue-driven nature of the medical culture there. He says many doctors view their practice as a way of making money, and if you want to practice medicine that way there are few barriers to stop you. By contrast, he cites places like the Mayo Clinic, where doctors on salary receive no incentives for ordering extra tests, and participate in meetings designed to improve patient care systematically by coordinating it to eliminate needless and duplicative tests, among other things. He admits, though, that discouraging the former behavior and encouraging the latter will be a long, tricky process.

I think a key element in the solution, if one can be found, lies in a careful study of incentives and disincentives. Although people can't always be relied upon to do the rational thing, most people will make choices they perceive to be in their own best interest. Of course, perception can be distorted through propaganda and so on, but especially where pocketbook matters are concerned, most people make fairly optimal decisions if given the opportunity to do so. One trouble with health care as it exists today is the same problem I have noticed with the college-textbook market: the people who pay for the goods or services (students or patients) are not the people making the decisions (professors or doctors). Although one doctor quoted by Gawande says allowing patients more of an economic stake in medical decision-making is like relying on "the sheep to negotiate with the wolves," it doesn't have to be that one-sided. If people were in economic control of their own health-care expenditures rather than having to rely on their employer (if they have a job in the first place), I think some way could be developed so that the Mayo-Clinic-type coordinated operations and their lower cost per patient could be packaged to be more appealing in an open market, compared to the multiple-stop-shopping of places like McAllen. The comparison is a little unfair, but think of shopping at Wal-Mart versus going to a third-world village market with its street of shoemakers and street of roasted-goat vendors. The streets full of private vendors are colorful and make for great vacation photos, but if all you want is a pair of shoes you'll go to Wal-Mart.

Of course, efficiency can be carried too far, and if we let even the Mayo Clinics coalesce into one giant monopolistic medical provider, the outcome is likely to be bad. But an appropriate level of market openness in which consumers could see economically efficient, good care for what it is, and choose it, would avoid the coercion and potential for debilitating bureaucracy that so many proposals involve.

Healing is a deeply ethical activity. The oldest known professional code of ethics—the Hippocratic Oath—deals with the ethics of medicine, and many religious leaders such as Jesus made healing an important part of their ministry. The problems Gawande and others have identified when healers start to put money over patient care merely demonstrate that the system, whatever form it takes, must have professionals in it whose philosophy or faith makes healing an end in itself, not primarily a means to wealth. Whatever happens to U. S. medical care after this summer's debate, I hope the designers do not lose sight of the fact that, like doctors themselves, they cannot fix the problem the way you would fix a balky lawnmower engine. All they can do is try to create an environment for people of good will to do even better than they are doing now.

Sources: Atul Gawande's article "The Cost Conundrum" appears in the June 1, 2009 issue of The New Yorker.

Monday, June 08, 2009

The Air France Crash: More Questions Than Answers

The crash of Air France flight 447 from Rio de Janeiro to Paris on the last day of May is bad news for a number of reasons. The deaths of all 228 people on board make it the worst air disaster since 2001. And while advancing technology has enabled investigators to recover a limited amount of flight data through a remote data link that was operating at the time of the crash, the deep waters where the plane went down may prevent the recovery of the "black box" containing voice and detailed data recordings.

What do we know today, eight days after the crash? There were thunderstorms in the area that night, and early speculation centered on the possibility of a lightning strike to the plane. Although lightning hits planes hundreds of times a year, relatively little damage usually occurs and most modern aircraft can be considered essentially (though not totally) lightning-proof. Evidently, there was a satellite or other radio-mediated data link which was continually feeding certain types of flight data to the ground. Examination of this data shows that the flight speeds during the last minute or so of the flight became "incoherent," followed by a loss of cabin pressure and failure of electrical systems. While this information will be helpful in deciphering what went wrong, it apparently lacks the detail that flight data recorders can preserve. One can imagine a day when such radio links will take the place of, or at least duplicate, the capabilities of flight data recorders so that mechanical recovery of the black box will no longer be so urgent. The black box is designed to emit a sonar signal for 30 days after the crash, so the underwater recovery crews gearing up to find it are operating under tremendous time pressure, not to mention the water pressure at depths exceeding 20,000 feet. The box may never be found.

Recent news reports have focused on the fact that the plane's Pitot tube, the device that measures airspeed, had not yet been replaced with a newer model as the plane's manufacturer Airbus recommended. A Pitot tube is a small tube that faces directly into the airstream. The difference in pressure between the air inside the tube (which is blocked off and registers what is called "stagnation pressure") and the ambient or "static" air pressure, is an indication of airspeed, which is the most important kind of speed to know about when you are trying to fly a plane. These days, when most parts of a flight except for landing and takeoff are under automatic control, the airspeed data from the Pitot tube forms part of an elaborate computer-controlled feedback loop that maintains constant speed, altitude, and other flight characteristics.

The old saw about computers regarding "garbage in, garbage out" goes double when a feedback loop is involved. If enough ice forms on a Pitot tube to plug up the entrance, the indicated airspeed goes way below what is actually the case, and either the automatic pilot guns the engines inappropriately or the real pilots may take incorrect action based on faulty airspeed data. This is exactly what happened to an Argentine DC-9 flight in 1999, which resulted in a spectacular crash, killing all aboard.

Although I have no independent information on this, I hope that modern aircraft such as the Air France A330 that crashed have more than one means of measuring speed: either a second Pitot tube (which of course would be just as likely to ice up as the first one), or other means such as radar altimeter and speed measurements or GPS-based airspeed indicators. But whether the autopilot takes all these other inputs into consideration, and whether the real pilots do too, I don't know. Clearly, if the Pitot tube was involved in this crash, the right thing to do in the circumstances wasn't done.

All flight-critical Pitot tubes have heaters to prevent icing, but evidently the one on Flight 447 was deficient or non-optimal in some way, or else Airbus wouldn't have recommended replacing it. Of course replacements can be recommended for all kinds of reasons, some of which have nothing to do with safety. All that will come out in the investigation report, which will take months or more to complete.

Despite this crash, the general trend in air safety has been a positive one. More people fly every year, and so the safety record per passenger mile is even better than the raw statistics on crashes would indicate. But this record can be maintained only through the painstaking work of investigators, engineers, regulators, inspectors, and the pilots and crews who actually do the work. Most of the time the system works well, and the silver lining in every accident is the fact that it carries with it potential answers to problems that need to be addressed to improve safety even further. I just hope they are able to recover the flight data recorders in order to develop a complete picture of what went wrong, and to teach us how similar situations can be avoided in the future.

We will revisit this accident when more information is available, and in the meantime, our sympathy is with the relatives and friends of those who lost loved ones in this tragedy.

Sources: I drew on reports from Fox News at,2933,525117,00.html and an Associated Press report obtained from Yahoo News at, as well as the Wikipedia article "Pitot tube."