Summary

  • FirstGov joined a donated search engine to a government-wide public portal, but the search index was not the same thing as the agencies’ source pages or an archival record.
  • Contemporary sources described an initially separate Federal Search Foundation role in maintaining the index; later government accounts name different search providers, so the evidence supports a changing service lineage—not a single technology or custody chain.

A deadline produced a door, not a new archive

The White House announcement of 22 September 2000 framed FirstGov as a “single” entrance to federal information. The date mattered: President Bill Clinton had challenged government and industry in June to complete the portal within 90 days, and the site launched on that timetable. Its premise was practical. A citizen looking for a loan, a retirement benefit or a campsite should not have to guess which agency owned the relevant page before searching.

The launch fact sheet said FirstGov could search 27 million federal-agency web pages at once. It also said the system could search half a billion documents in less than a quarter-second and handle millions of searches a day. Those figures describe the government’s contemporary account of scope and capacity; they are not an independent benchmark or a measured daily load. A Berkeley report published three weeks later used other figures: more than 100 million government pages in the wider universe, about 30 million expected to be requested initially, and a system designed with room to grow.

The different denominators should not be collapsed into one proven count of pages actually indexed at launch.

That qualification does not diminish the design problem Brewer was addressing. By 1995, he and Berkeley graduate student Paul Gauthier had developed Inktomi from work on networked workstations. A Berkeley account described a crawler that periodically discovered new addresses, an index of more than 1.3 million documents and parallel searches across a cluster of inexpensive machines. It even reported that a four-workstation arrangement could retain access to the index if three machines failed, at reduced speed.

This was a contemporary description of a particular system, not a guarantee that every later Inktomi deployment had the same failure behaviour.

The architectural through-line is more modest and more interesting than a story of a lone inventor. Search at the web’s scale was not just a ranking formula. It required a changing index, machines that could divide work, and a way to locate pages whose owners and structures were distributed. FirstGov applied that search capability to a civic interface: visitors could search across agency boundaries rather than navigate the federal org chart first.

“Donated” did not mean every layer changed hands

The most revealing detail sits beside the public launch narrative. The White House described a search engine developed and donated by Brewer. A contemporaneous Berkeley report said the Federal Search Foundation, a nonprofit Brewer founded in 2000, would initially own and maintain the database. It added that the government would have a right of first refusal to purchase it after two years. The same report said Inktomi technology would maintain updated copies of government web pages on an expandable server network.

That account separates at least four layers: agency pages, the copies or index used to find them, the search engine, and the public portal. Calling the engine a gift does not establish that the foundation’s index immediately became government property, that the index was an official record, or that every source page moved into Brewer’s custody. Nor does the Berkeley article tell us how the first two years ended. The available sources do not establish the final disposition of that arrangement.

This boundary matters because search is a powerful convenience but a limited form of custody. An index can make scattered pages discoverable and can refresh copies; it does not, by itself, preserve the authoritative version, explain whether a missing result was never published or fell out of the index, or make one agency answerable for another’s content. The inference is about system boundaries, not a claim that FirstGov failed. The contemporary record documents an access service, not a complete government archive.

The name lasted longer than the original arrangement

FirstGov’s later history also cautions against treating a portal name as an implementation diagram. A 2006 General Services Administration announcement described FirstGov Search as a partnership with Microsoft and Vivisimo, using the MSN Search index with Vivisimo metasearch technology. It cited more than 50 million federal, state, local, tribal and territorial documents and said indexes were continually updated. This is evidence that the named service had moved to a different provider/index arrangement. It does not reveal the legal fate of the initial Federal Search Foundation database or prove what happened to every piece of Inktomi code.

The present-day USAGov account says SearchGov shares its history with USA.gov, traces that story to Brewer’s donation and says the service expanded to government agencies in 2010. It reports more than 200 million searches a year across thousands of federal websites. That current number describes today’s service as USAGov reports it; it should not be compared directly with the 2000 launch claims. A public name and mission can persist while suppliers, indexes, funding and operating responsibilities change.

Berkeley’s faculty profile records Brewer as a computer scientist whose work spans scalable servers, search, network infrastructure and government. That breadth makes FirstGov a useful career episode: a system first developed to find web pages at scale was offered as a way to make public information easier to reach. The project was also collective. Inktomi’s early development involved Paul Gauthier and a Berkeley research environment; the portal required a presidential deadline, agency participation and continuing public administration.

Access is not authority

The lasting test for a government search service is therefore not only how many queries it answers. It is whether an agency’s source page remains authoritative, how stale or broken results are detected, who maintains the index, and what happens when a supplier or funding model changes. Search makes the state easier to approach, but it cannot substitute for the state’s responsibility to publish, preserve and correct its own information.

That is the distinction behind FirstGov’s “one door.” A common entrance can reduce the cost of finding a service without centralizing the underlying records or transferring their custody to the search operator. It can be a useful public interface without becoming a legislature, a registry or an archive. Brewer’s donation helped make the doorway real; the later record shows why the doorway’s owner, the index’s custodian and the source agencies should never be assumed to be one and the same.

Sources