Rachel Cunliffe asks when we’ll see predictive blog stats packages. “All the statistics packages I’ve seen are purely descriptive and not predictive.” So true. And, having been checking out Inrix and their predictive (automobile) traffic data, I’m wondering whether most data businesses of the future won’t have a predictive component. After all, only some uses of stats packages are audits–a lot of stats are gathered for business intelligence reasons, to gain insight into the effectiveness of actions the company takes. Every package that gathers data should have a machine learning component to help predict it too. Quicken should tell me how much money I’ll have, du should tell me when it looks like I’ll need to buy a new drive, and iChat should tell me when I’m overdue to say “hi” to my Mum.