Monday, November 16, 1998; Page A24

Putting more government information on public view via the Internet sounds terrific in theory. It's only when you get down to actual cases that alarmed "Yes, buts" start surfacing and missteps multiply. These come generally from the same quarters as would opposition to the release of the same information by traditional, nonelectronic means, and they represent the same mix of good reasons, bad reasons and plain inertia that keeps so much government information inaccessible.

Two recent slip-ups over information posted on government Web sites showcase the serious questions raised by the possibility of making government data genuinely, in some cases globally, transparent. The first was the Environmental Protection Agency, which attracted howls of protest last month when it announced it would place on its Web site information about potential "worst-case scenarios" at major U.S. chemical plants, including maps and data about chemical inventories, potential release of toxins and possible danger to neighboring populations.

The arguments for making such information available are obvious enough -- safety of the local residents, research into environmental threats. Nor is it surprising that chemical industry representatives would oppose such publication.

The more interesting criticism, though, comes from law enforcement authorities ranging from the FBI and CIA to the national fire chiefs' association. All argued that the Internet publication of such information would constitute a blueprint for terrorists. The EPA has said now that it will study more carefully what parts of the chemical companies' data can be disseminated in light of these concerns, which go well beyond what would govern the release of such data on paper.

The Bureau of Labor Statistics fell into a different pitfall when its Web site inadvertently released key data on quarterly job creation; the premature release rocked the stock market. The result, though, was similar: The agency closed part of the site and retreated, promising investigation of what it characterized as a simple error.

It also will look at the general problem of what kinds of "supplemental data" can be loaded onto the public site and on what basis. In both cases the best outcome will be the continued pressing outward of limits on prompt and convenient disclosure. But these periodic misadventures are reminders that the transition to such disclosure is fraught with more than just technical difficulties.