The big problem is that while we crunch the same numbers in the same ways from job to job, each project usually requires us to do something a little differently -- to collect different kinds of data, or to devise new comparisons of facts we have uncovered.