1 2 3 4 5 6

sites with buyers.[12] Burst!'s approach is a bit different from DoubleClick's, in that it specifically has chosen to work with smaller companies and sites. Burst! matches advertisers to its network of more than 500 sites by looking for network members whose demographics fit the advertiser's desired profile. The company characterizes its sites as independent content producers--homegrown sites covering topics like genealogy, fashion, fishing, or gourmet cooking.

Jarvis Coffin, Burst!'s president and a former director of advertising for the Los Angeles Times, says, "We are entirely betting on the fact that the Internet is going to be about a tremendous diversity of content" (telephone interview, 1996). Burst! works on commission; it stores the ads--small banner graphics--on a server at BBN Planet, one of the original contractors for the precursor to the Internet. When you load a page from one of Burst!'s sites, a few lines of code in the Web page call up the ad from Burst!'s server, load it, and count the hit for auditing purposes.

All those advertising services will require another category of middlemen: third-party auditors to verify the number of times a particular company's banners have been seen on a given site and on which pages. And another: organizations who are concerned about the future of truth-in-advertising. Britain's Advertising Standards Authority, which responds to public complaints about ads and false claims, regards Web sites run by British companies as falling within its territory.[13] Similarly, medical authorities, including the Food and Drug Administration (FDA), are concerned about the potential for selling unapproved drugs and quack health cures in an unregulated environment.[14]

One non-regulatory solution to this might be some kind of branding by trusted organizations. In an international medium, you can only control sites run by companies from your own country. Under a branding scheme, sites containing reputable information could be endorsed by third parties with known reputations, such as the FDA and its counterparts elsewhere. This is similar to the ratings and classification systems already being discussed for Web sites, where you might choose a ratings system based on the known agenda of a particular church, government body, or cause organization. Surveying all those Web sites is a time-consuming job, but here again, a standardized interface so that any organization could custom-build a database of sites and plug it in would go a long way toward ensuring maximum user choice. Today's specialized software packages have already been criticized for blocking material that doesn't obviously fit the categories they say they're blocking, such as sex and bomb-making instructions.

An alternative to blocking software is a different class of middlemen that already exists on the commercial online systems: moderators and sysops, who get paid to monitor specific areas to make sure they conform to what's expected of them. Moderating is a hard enough job (trust me, I know) that few people want to do it for any length of time without pay. The business model required to support paying these people, which awards them royalties according to the amount of time people spend in their areas, is what keeps CompuServe and kept America Online so comparatively expensive. AOL's conversion to a flat-rate service at the end of 1996 requires the service to find new ways to pay its content providers; the most likely will be through selling advertising space. What effect sponsorship will have on online debate remains to be seen.

Some traditional middlemen are likely to survive, even though early predictions were that they would no longer be necessary. These are primarily information specialists, such as librarians, researchers, and journalists. The thinking was that no one would need these filters because the information would all be right there on the Net, and good computerized searching would make it easy to sort through. But anyone who's ever used a computer search program like Altavista knows that the results it produces, often as many as 10,000 or 20,000 hits, are only the beginning. Sifting through those to find the good matches is difficult; the many mistaken conclusions drawn from such searches by those seeking to sensationalize pornography online are a testament to that. One of the more interesting experiments in early 1997 is HumanSearch, a search service set up by a Rhode Island college student and his father with about forty volunteers. You tell them what you want to know; they search for you using a variety of techniques.[15]