The methods for certifying software quality continue to multiply. Popular, process-based approaches-- such as ISO 9000 and SEI-CMM--make software publishers take oaths concerning which development standards and processes they will use. These approaches often require auditors to spot-check a publisher's documentation and oaths. Yet even if a certification auditor can verify the publisher's veracity, that verification alone does not guarantee high-quality software.

Given these problems, the author proposes a certification methodology that does not employ publisher oaths and auditors. This certification method employs automated processes to greatly reduce liability while also eliminating the need to dispatch human auditors. The process harnesses the testing resources of end users, drawing upon proven tech-niques such as the methods that made Linux the most popular and reliable of all Unix flavors. Totally product-based, this proposed process assesses how well behaved the software is, not the maturity of the processes used to develop the code.