Shipped any bad parts lately?

The scandals and Harvard's response have focused on financial transgressions. But ethics certainly applies to product quality. A classic ethical failure is dramatized in Arthur Miller's 1947 play, "All My Sons," in which shop owner Joe Keller orders shipment of 120 aircraft-engine cylinder heads, even though visual inspection showed hairline cracks. The result: 21 pilots are killed.

The play convincingly describes the intense pressure that led to Joe's abhorrent act: "Every half hour the Major callin' for cylinder heads, they were whippin' us with the telephone." And the motives and rationalizations he offers are all too easy to sympathize with: He acted to benefit his son; and of his Army Air Force customers, he says, "I never thought they'd install [the defective heads]...by the time they could spot them I thought I'd have the process going again."

The temptation to buy time to improve "the process" is easy to understand, if not defend. However, Joe had ordered a superficial weld to cover the cracks, making it unlikely that the Army would detect them.

But would his behavior have been any less detestable had he not disguised the defects? What if, under the pressure of the major's demands, Joe's employees had unknowingly shipped 120 bad parts?

What if the company inadvertently shipped one bad part out of a batch of 120, or of 120,000? No one can answer a resounding "no" to the question posed by this column's headline—some nonzero defect rate is inevitable, and one out of 120,000 sounds reasonable.

It's difficult to know where to draw the line. An ethics class for test engineers might address questions such as these:

1. Assuming health and safety aren't at risk, is it ethical to ship 120 products that you know are bad as long as you know you can replace them with good products under warranty?
2. Is it ethical to ship 119,999 good products and one bad product (you don't know which is bad) as long as you can replace the bad one under warranty?
3. Is it ethical to ship 119,999 bad products and one good product (you don't know which part is good) as long as you can replace the bad parts under warranty?
4. If you answered "yes" to 2 and "no" to 3, then what percentage of bad parts would it be ethical to ship?

My answers? "No," "yes," "no," and "it depends." Send me your comments, and to read what I think the answer to question 4 depends on, see below.

Contact Rick Nelson at rnelson@tmworld.com.

Web Exclusive—Analysis of Four Questions

The four questions above hint at the difficulty of quantifying ethical behavior. Professional codes of ethics tend to provide qualitative guidelines. The IEEE Code of Ethics, for example, mandates "making engineering decisions consistent with the safety, health and welfare of the public [and] to avoid real or perceived conflicts of interest," among other things. And the Software Engineering Code of Ethics and Professional Practice, promulgated by the ACM and the IEEE, mandates that "Software engineers shall act in a manner that is in the best interests of their client and employer consistent with the public interest."

Texas A&M University provides a collection of engineering ethics case studies that help demonstrate the difficulty in applying such broad guidelines to specific practices.

 Here are my answers to the four very simple questions I presented above:

1. Knowingly shipping 120 bad products out of 120 total is essentially lying, and it's hard to imagine a scenario in which such behavior would be ethical. The IEEE code requires that one "be honest and realistic in stating claims or estimates based on available data," and deliberately shipping known-bad products can hardly square with a claim of honesty.
    
2. Given that perfection is impossible, some test escapes are inevitable, and the shipment of some low percentage of defective products seems ethically acceptable. That said, if safety is a consideration, other factors come into play.
     
3. Shipping mostly bad parts is ethically equivalent to shipping all bad parts. If your customer pleads for one good part, it would be up to you to determine which one is good, and ship that alone.
     
4. If you are shipping components to an OEM, the acceptable defect rate is probably a matter of contract. If you're selling to consumers, though, there will be no such contract. Retail stores never disclose information such as "one in twenty of these DVD players will prove to be defective when you get it home, and after you spend several hours trying to hook it up, you'll have to bring it back and stand in line at the customer service-counter to get a new one, which also will have a one-in-twenty chance of not working."
   
   Even if you are meeting your numerically defined contractual obligations to a customer, you may not be acting ethically. You may know more about the safety implications of your components than your customer does and thereby incur ethical obligations that exceed your contractual and legal ones. The "TV Antenna Tower Collapse" case presented on the Texas A&M site presents one such example.