the private sector, it is the committee's opinion that the result will be a patchwork of databases of uneven or unknown quality. Larger, better-financed private-sector organizations or industry sectors may develop occupational information systems to meet their needs, but these systems would, naturally enough, cover only those occupations and descriptive variables that would be of most current interest to them. They may, again understandably, maintain proprietary control over the information to retain a competitive edge. Smaller organizations and sectors may have no systems at all. Trying to put together a national occupational database by building on such a collection of independently developed databases is doomed to failure. Developing and maintaining a common, national occupational database allows private organizations and individuals to apply their resources to augment, enhance, and build on the national information to compete in a global marketplace. Competition is not eliminated, it is likely to be raised to another level—a good thing for the entire country.
One missing element in the present vision of an occupational information system is a closer tie-in of the day-to-day labor transactions and the occupational information system. If daily recruiting, hiring, and firing activity could be linked with occupational categories and, in turn, with the associated skills, abilities, and other attributes of the categories, then trends in desired or required occupations and occupational attributes could be more dynamically monitored. Furthermore, historical data could be accumulated that would be extremely useful for disentangling relatively minor or momentary trends from longer-term shifts in the world of work, something that we have shown to be a difficult undertaking. It seems to us that many of the pieces for such a linkage are already or nearly in place.
To achieve a dynamic, accurate occupational information system, some compromises will no doubt need to be made between breadth of coverage (numbers of occupational categories), depth of coverage (number and types of attributes of occupations), and currency of information (frequency of updating). One possible compromise that strikes us as attractive is to forgo the routine, random sampling of small-frequency occupations or occupations that are extremely difficult to access. The cost per bit of informa-