Comments on the role of testing

ecp@acm.org (Erik Petersen) writes:

"Will the product accurately track financial information? If it does not, "stop ship" authority is required and may be needed. As an example, handheld computers used in their hundreds to track collection of tens of millions of dollars a day for a multinational's deliveries, where I had "stop ship" and anticipated using it but luckily the code quality improved remarkably at the close of the project."

Brian Marick replies: "I agree that certain situations require the testing team to be a barrier between development and the customer, one that can stop shipment. In those situations, the cost of a bad decision to ship is so high that the inefficiencies of this division of labor are tolerable."

Cem Kaner writes:

"I disagree. I think that the test group needs the power of free speech. If the company is about to make a terrible decision, the test group needs the power to speak to any decision-maker in the company, to get that person to influence or change the decision. In my experience, this usually works.

"Directly stopping shipment works sometimes too, but it creates an opportunity for the responsible decision-makers to be irresponsible. If a don't-ship decision will be seriously unpopular, the project manager (or VP or whatever) will be tempted to let Mikey (the test group) do it. Over the long term, this dynamic can be seriously damaging."

Got something to add?