The Betterment of the Human Condition

The Betterment of the Human Condition? (Part II)

We have been considering the ways in which engineers, contrary to the layman's definition with which we started, do harm to the rest of humanity. The second category I want to consider is `expected but unintentional harm'. At first it isn't clear how there can be such a category: if you expect an action to be harmful, yet you do it anyway, how can we say the harm is unintentional? Let me give an example which will show what I mean:

Dow Chemical develops a medicine for cancer. In the majority of cases, the medicine will cure the cancer. In all cases, the medicine has serious side-effects: it causes hair loss and intense irritation of the gastro-intestinal tract, often leading to ulcers. All these effects are known in advance, and it is also known that in about 10% of the cases, the patient will suffer the side effects without their cancer being cured. So certainly those 10% are suffering harm, and it was expected in advance that they would suffer harm, yet no-one intended that they should suffer.

Engineers frequently find themselves in the situation of the chemist who develops such a drug; there are very few new technologies that are entirely without side-effects. For example, the civil engineer who builds a dam is providing a renewable power source that may replace fossil-fuelled or nuclear alternatives, yet the dam may also destroy salmon habitat, and rotting vegetation in the newly formed lake will release methane, exacerbating the greenhouse effect.

As another example, Jacquard's development of the automatic power loom made high-quality cloth available for many who couldn't previously have afforded it. Yet it also created such miserable conditions for the mill workers that they would risk hanging to smash the looms. Should Jacquard be held responsible for their misery?

A third example: the electronic or computer engineer who designs a terminal is exposing the user of the terminal to low-level electromagnetic radiation. Is this doing the user harm? Some members of the public believe that it is. Few health studies have confirmed their belief; but then, much of the data used in such studies comes from those with a vested interest in finding no harm.

There is a standard way of handling situations in which a given course of action has known costs and benefits. We calculate a single number, the ratio of benefit to cost, and proceed if this number is greater than unity. Unfortunately, there are often practical difficulties in applying this simple measure.

One difficulty is that to calculate the cost/benefit ratio, both costs and benefits must be expressed in the same terms -- usually, though not necessarily, as cash figures. How do we do this in cases where the costs or benefits include human lives? Well, what is the cash value of a human life?

One option is to say, ``You can't play the numbers game with human lives. Each life is unique and of inestimable value. Attempting to treat lives on the same basis as dollars is both cold-blooded and ridiculous.''

But this really won't do. For example, medical administrators have a responsibility to save lives, and they have limited resources to meet this responsibility. If they are to apportion their resources rationally, they must be prepared to compare the results of different strategies.

And in any case, we can demonstrate that we all agree that there is an upper bound on the value of a human life. For example, suppose research showed that, by making structural alterations to a car, we could reduce the likelihood of a fatal accident by 1% -- that is, given 100 cases in which accidents had led to fatalities, the proposed alterations would reduce this to 99 fatalities. The alterations cost $1,000 per car. Would you buy the altered car yourself? If you were the government, would you require all car manufacturers to make these alterations?

I suggest that you wouldn't pay $1,000 for such a slight improvement. After all, the chance of being in an accident is pretty low anyway, otherwise nobody would drive. Are you going to pay $1,000 to reduce it by 1%?

There are about 6 million cars sold per year across North America, so the total cost of the changes would be 6 billion dollars. About 60,000 North Americans die in motor accidents per year, so the changes would save 600 lives. So we agree that it's not worth spending $10 million to save a life.

Note that, in this hypothetical example, you're estimating what it's worth to save your own life. Presumably you would not pay more than this to save someone else's life, and you wouldn't expect anyone else to pay more than this to save your life. So it seems safe to take $10,000,000 as an upper bound on the value of a human life.

Another question we have to resolve before we can apply cost/benefit analysis is, what is the value of a future benefit? Suppose, for example, I want to borrow $1,000 from you, to be paid back in one year. How much should I pay you back?

There are a large number of things you will want to consider: how much do I want the money? What else could you do with the money if you didn't lend it to me? How reliable do you think I am? Am I offering you any collateral on the loan? By considering these factors, we may be able to come to an agreement. That agreement will almost certainly say that, if you're going to lend me the money, I'll have to pay you back more than I borrowed. We could think of the extra amount I pay back as a rental that you're charging for the use of your money. Another way of thinking of it would be to say that $100 a year from now is worth less than $100 right now -- if it weren't, you should be satisfied to lend me the money for no interest. (Note that this has nothing to do with inflation.)

This simple observation is known in engineering economics as `the time value of money', and a lot of elementary economics is concerned with deriving formulae to compare the value of future cash flows with their present values. In general, we say that getting a sum of money M in a year's time is equivalent to getting a sum M/(1+i) right now. Applying this argument repeatedly over n years, we conclude that getting $M in n years time is equivalent to getting M/(1+i)ⁿ right now.

Now we've established this principle -- which is really used as the basis for most economic calculations -- let's go back to the cost of human life. We've established that a human life is worth less than $10,000,000. What's the life of a person born in the year 3000 worth now? Well, in the year 3000 it's worth $10,000,000. What amount of money in the year 2000 is equivalent to $10,000,000 in the year 3000? Applying the formula we've just derived, it's

$10,000,000/(1.05)¹⁰⁰⁰ = $6 * 10^-13

So to save the entire global population of 10 billion in the year 3000, it's worth spending at most 6 * 10^-13 * 10¹⁰ = 0.6 cents now. For example, if you were an astronomer and detected a comet that would impact the earth in 1000 years, and if we had the technology to stop the comet at zero cost, it wouldn't even be worth making a phone call to recommend using that technology.

This conclusion is surprising, but it isn't a trick; this is the conclusion that standard economic analysis delivers. In general, standard economic analysis sets a very low value on benefits that lie a long way in the future. This can have practical consequences; for example, one of the engineering decisions we are currently facing is the storage of high-level radioactive waste for tens of thousands of years. In planning storage facilities like Yucca Mountain, the cost of improved storage has to be weighed against the cost of the damage done by leaking waste centuries or millenia in the future.

Human life is not the only thing that's difficult to figure into cost/benefit calculations. For example, how do we assign a cash value to a particular level of atmospheric pollution? Further, how do we deal with the common case where the people who pay the costs are not those who reap the benefits? This becomes particularly difficult to decide when the beneficiaries are the ones who pay our salary.

A further question is, how broadly are we to interpret the notion of `harm'? Certainly, if we can show that a particular technology produces cancer in a significant number of people, that is something that must be taken seriously. But what of less tangible forms of harm? Suppose it can be shown that the construction of a new highway system will put local merchants out of business, and thus destroy the sense of community in a small town. Should we attempt to measure and reduce this harm, or should we regard it as inevitable, and hence unimportant?

Consider the following four test cases:

After several successive years of drought, the GVRD decides to construct a new reservoir to serve the Vancouver area. A preliminary study shows that there are only two feasible sites. Site A involves flooding a small valley. The valley is uninhabited, but a local First Nations band objects to its being flooded, on the grounds that it is a burial ground and sacred site. Although the valley has not been used for burials within the past four hundred years, the band asserts that it was so used in the more distant past. On the other hand, anthropologists from a local university say that the band only came into the area a few centuries ago, and has never used the valley for burials.

There is no significant public opposition to the use of the alternative site, Site B.

If Site A is used, the cost of the reservoir is estimated to be $120 million; the cost of building in Site B is estimated at $200 million.

What procedure should be followed to reach a decision in this case?
You are working for a pulp and paper company in northern Canada. The company is a wholly-owned subsidiary of a multinational conglomerate. You report to the plant manager, Smith, who reports to the conglomerate's head office, located outside Canada. The company employs 150 people, which is most of the adult population of the local community.
In the course of your work, you have become aware, because of your knowledge of the processes used in the plant, that the effluent from the plant contains a very high concentration of a mercury compound that could be dangerous. In fact, since the plant has been discharging this material for 25 years the river is thoroughly unfit for drinking or swimming downstream from the plant, and you suspect that the curious illness reported in a community downstream is actually Minamata disease, which is caused by mercury contamination. The classic symptoms are loss of coordination, spastic muscle movements, and eventually death. You suspect that the fish in the river are contaminated with the mercury, and have spread the contamination to all the downstream lakes. To remedy the problem would involve drastic changes to the plant, costing at least $5 million. At present no-one knows of the suspected problem except you and Smith. You have discussed the problem at length with Smith, and he has confided that the head office considers the plant to be only marginally profitable, so an expenditure of this magnitude is simply not possible. The head office, he says, would simply close the plant, causing massive unemployment, and would probably cause the workers from the community to abandon their homes and seek work elsewhere, since the alternative would be severe poverty. What should you do? [This problem appears on p. 207 of `Canadian Professional Engineering Practice and Ethics' by Gordon Andrews and John Kemper, ISBN 0-03-922875-4, 1992.]
You are employed by NASA. One night you are awakened by a phone call at 4 am. The space shuttle Discovery, carrying a crew of seven, is experiencing problems. A number of tiles have been lost from the underside of the shuttle, making it almost certain that the shuttle will burn up on re-entry. After hurriedly examining all possibilities, you and other NASA engineers can identify only one possible solution, which is to lighten the shuttle by jettisoning the lead-shielded radio-isotope auxiliary power generator. If this is jettisoned, the shuttle has better than a 50:50 chance of landing safely. However, the jettisoned generator will vaporise on re-entry, spreading a fine dust of radioactive cesium in the upper atmosphere. It is estimated that the resulting radiation will lead to between 10 and 50 cancer deaths worldwide over the next five years.

The cash value of the space shuttle is about two billion dollars.

There is no time to investigate other possibilities in more detail; you have to decide whether or not to jettison the generator within the next half-hour.
You work for a government agency that develops genetically modified foods. Over the past five years, the company has created a new variety of rice that will grow on marginal soils, yet requires very little fertiliser. Exhaustive safety testing has been performed, and no ill effects on the environment or on human health have been seen. It is estimated that introducing this rice in southern Asia will save the lives of about a quarter of a million people per year.

During an informal conversation, the genetic engineer who developed the rice confides to you that, to achieve her results, she introduced short segments of animal DNA into the rice genome, including genetic material from cows and pigs. While these segments will not have adverse health effects, their presence is likely to make the rice unacceptable to many of its intended beneficiaries. Therefore, she says, she plans to alter her records so that the source of the added genetic material will not become public knowledge.

What, if anything, should you do?

Next: Betterment III

Previous: Betterment I