"One of our senior partners, Mr. Jones, is about to retire," an engineer from a large construction company confided to me at a press dinner, "and it s my job to be sure that his domain expertise stays with the firm after he's gone." Although this engineer didn t say so outright, he seemed skeptical of ever accomplishing his mandate. "After 25 years of reviewing design drawings, Mr. Jones understands how all the elements of a plan fit together. There's got to be a way to get his special knowledge into our mapping system, but . . . ," his voice trailed off, leaving the sentence unfinished. After a covert glance around the table, he whispered that his firm was even willing to switch GIS software packages if they could find a product that allowed their spatial data to encapsulate their own domain rules. That conversation typifies both the emerging need for and the lack of awareness about geospatial data modeling.

The need for data modeling has always existed and, in fact, is satisfied by all file-based GISs now in popular use. In file-based systems, however, the data are separate from the domain expertise. Domain experts are people so passionate about their profession that they know every detail, past and present, about their organizations daily operations.

This system of operational details is also known as business rules. Municipalities in particular are often staffed by domain experts who store their cities' layouts as intricate, private mental maps. For instance, when asked about what appeared to be an error in the addressing system of her city's street centerlines, a permitting expert in Oakland, California, explained exactly how the city's street network had begun and how the branching construction from a backbone street resulted in increasingly large addresses in all directions away from the backbone. She knew, from memory, the date and material of construction of every street in the city (give or take a year) and saved my data-conversion team weeks of research by identifying (again from memory) every unregistered road in a dataset of 14,000 street segments. "Ah yes, that s the 2000 block of Miller Street," she said, barely concentrating, "and on a grade that steep, they had to build it with concrete." When this expert retires, her city will lose an amazing resource.

Living legends with a zest for their discipline bring their areas of expertise alive for everyone around them. Unfortunately, it's not possible to compress domain experts, along with the data and systems they administer, into a digital file so that their knowledge can be unpacked for use even when they call in sick. Or is it?