Secure software development

In connection with this month's NoticeBored awareness materials on the security aspects of software development, I've been listening to a podcast by Ralph Hood and Kim Howell (two Microsofties) about how both privacy and wider information security issues are integrated into Microsoft's development practices.

From a non-US perspective, the very idea that privacy and security are "opposite sides of the same coin" seems a little weird. For most of the rest of the world, privacy has long been acknowledged as a subset of information security, being essentially the confidentiality of information about specific individuals. But, as host Julia Allen mentions in the podcast, the US is still shifting from the idea that it's perfectly OK to collect all sorts of personal information from people and use it as you wish.

One of the interesting approaches discussed in the podcast is that personal information needed purely for aggregation or statistical purposes should be collected and held only temporarily, then deleted as soon as possible. Personal information in server logs, for example, may be parsed out, analyzed and deleted by a regular process. While this would make post-hoc validation of the data difficult, this slight drawback is outweighed by the privacy advantages for those who supply their information.

The podcast is one of the excellent Security for business leaders series by CERT at Carnegie Mellon University. An impressive range of podcasts is available to download.