In 1994, an error in the floating point arithmetic logic on Intel’s Pentium chip was discovered. The small bug could result in arithmetic error at the 4th decimal place. Pentium chips were sold starting in 1993, and this bug was discovered by Intel testers in June of 1994. Intel kept quiet about the issue until it became public in November 1994. Intel did not develop a plan to replace faulty chips until mid-December. Intel held off on replacing customer’s Pentium chips because it was going to lose money - in fact, they lost about $475 million after finally deciding they would replace faulty chips at no cost.
The issue presented by this case is whether and how to inform customers about product defects. Pentium chips were rolled out in March 1993, and the mistake was discovered the following year. Although Intel claimed the bug would only occur with a probability of 1 in 27 million, it didn’t take long for the error to start affecting its customers.
As a large corporation, Intel’s actions affected many people around the world. Stakeholders in the decision by Intel to withhold action once the bug was discovered include all those who purchased computers that use the Pentium chip, those who depended or used the computation outcome of the Pentium chip, Intel’s employees and shareholders, and computer companies who use Intel processors in their products.
Intel is a publicly-traded corporation with clear economic incentives for making the decisions it did. Having to issue a recall on all defective chips would certainly have a large negative financial impact on the business including possible loss of business to competitors. Also, as a publicly-traded corporation, Intel must be concerned with keeping shareholders happy. Producing a defective chip could negatively impact Intel’s stock price. Keeping stock price high and pleasing shareholders is certainly an incentive for hiding the defect from the public.
Furthermore, Intel had to consider the social impacts on their decision to reveal the defect. As a major chip corporation, Intel had to uphold their reputation. By acknowledging that their engineers failed in correctly designing the chips, they would also be acknowledging the incompetence of the company to deal with this problem and their disregard for consumer satisfaction.
From a utilitarian point of view the best outcome is what determines what is right or wrong. Microprocessors in general have many flaws in its system even though it doesn’t directly interfere with general usage, and they are generally not detectable by the user. If this were the case, then a utilitarian would support Intel’s decision in not reporting the issue. Reporting the problem would not change much of the operation of the computers and would require Intel to use up resources to fix the problem. Yet, the Pentium case is another matter. This problem did in fact affect the user operation, but Intel decided not to report it. By not reporting it, the flaw continued to negatively impact users, making Intel’s decision to keep quiet the wrong choice.
Another matter to consider is the justice test. Is there an existence of a fair distribution of benefits and burdens? For most consumers, this faulty chip did not affect their daily computer usage. In fact, many would not have discovered it. If Intel was not required to replace these chips, the majority of the consumers would not suffer from this defect. However from Intel’s perspective, there would be major consequences if the defects of these chips were announced. Not only would it cost the company monetary costs, it would also harm their reputation. The burden that Intel would carry outweighed the one of consumers.
Lastly, one must also consider the virtue test in this matter. Do Intel’s actions represent the type of company it wants to be? According to Intel’s stated values, they emphasize customer orientation, results training, risk taking, great place to work, quality, and discipline. The two matters that pertain to the current situation are customer orientation and quality. Intel failed to inform its customers of the flawed chip. The company was not customer orientated and in fact tried to hide the truth at first. Furthermore, the quality of the chip seemed questionable. From a customer’s point of view, he or she may expect the chip to have no issues at all. However from an engineer’s perspective, a perfect chip does not exist. There is no way Intel can possibly inform customers of all the bugs in a chip, especially if the issue appears to be trivial. The quality of the chip may have met the standards of the engineers, but it did not meet the expectations of consumers.
A possible solution that might benefit most stakeholders is to insert a description of the error into the instruction manual. Just like developmental microprocessors, Pentium’s 1000-or-so page instruction manual could contain a description of the error. The updated manual could be mailed out to all its Pentium customers. If we apply the justice test to this scenario, this choice would inform customers who find the error that Intel is aware of it and the customer could contact the company for a replacement if necessary. Customers who do not catch the error most likely will not need it, although this is a form of lying (violates the virtue test, since it could be regarded as information withholding). Intel would still have to pay for replacement and the sending out of updated manuals, but it would be more feasible than recalling all chips and replacing them.
Another possible solution is to implement a software solution. As described in one of our articles, MathWorks was able to change its software to accommodate for the processor error. Intel could reach out to commonly used calculation platforms and suggest that their programs implement a similar software correction.