← all stories

Before the Web · Era 2 · Into the Internet

1989–1995sold to AOL for ~$15M · 1995

WAIS

Wide Area Information Servers

A system for full-text search across documents on the internet before the web appeared: it was built in 1991 by Brewster Kahle and Harry Morris at Thinking Machines together with Apple, Dow Jones, and KPMG. In 1995 AOL bought the company, WAIS Inc., for ~$15 million, and the founder later put the proceeds into the Internet Archive and Alexa Internet.

Founders Brewster Kahle · Harry Morris · Bruce Gilliat
Domains —
searchopen-standardacquihireprotocol

In brief

WAIS was a system that, even before the web existed, searched not by file names but inside the text of documents on servers all over the internet: it was released in 1991 by Brewster Kahle, a supercomputer engineer at Thinking Machines, together with Harry Morris, and Apple, Dow Jones, and KPMG Peat Marwick served as partners in the pilot. Within a year the system grew to 10,000+ users in 24 countries, and in 1995 AOL bought WAIS Inc. for $15 million in stock. For Kahle the sale became starting capital: with that money he and Bruce Gilliat built the Internet Archive and Alexa Internet, and for years the commercial company fed data to the nonprofit one.

How it started (the founders)

Brewster Kahle joined Thinking Machines Corporation in 1983 and until 1989 was a lead engineer on the Connection Machine, a massively parallel supercomputer that Kahle himself, years later, described as the first search engine in the world. That, rather than an abstract idea of search in general, is where WAIS came from: there had to be a way to find information on a machine designed for parallel processing of huge volumes of data. Development began in October 1989, the date given in RFC 1625, which the WAIS developers wrote in 1994. Harry Morris became the co-author.

Rather than a solo effort, WAIS was put together from the start as a partnership of four organizations, each contributing its own type of resource instead of money: Dow Jones supplied premium business content, Thinking Machines the full-text search engine, Apple expertise in interfaces, and KPMG Peat Marwick its own staff as a test audience; fifteen Peat Marwick employees tested the prototype for more than three months, building it into their daily work. The first public release came out in April 1991. The protocol was built on top of Z39.50-1988, a standard for library card catalogs, but instead of the full OSI stack the team assembled its own protocol, inspired by Z39.50, on top of ordinary TCP/IP. By the end of 1991 WAIS had more than 10,000 users in 24 countries, and in six months the number of databases went from a handful to 160+ in eight countries. In 1992 Kahle and Morris founded WAIS Inc. estimate; in RFC 1625 of June 1994 both are already listed as employees of WAIS, Inc.

Year-by-year timeline

Lesser-known but significant facts

  1. WAIS Inc. made Ross Perot's presidential campaign its first truly profitable client. Kahle met Perot Systems head Mort Meyerson, built a system in a few days to coordinate data from 50 states, and by November 1992 WAIS was already running inside Perot Systems itself; it was this pilot, not the open protocol on its own, that made the company profitable.
  2. The four partners of 1991 paid not with money but with different types of resources. Dow Jones provided content, Apple interface expertise, KPMG Peat Marwick employees as a test audience, and Thinking Machines the computing engine; the first WAIS round was a barter of competencies, not an investment.
  3. WAIS officially cataloged itself. A special server, "Directory of Servers," held information about all the other WAIS servers on the internet, an idea that later inspired directory protocols such as GILS and WHOIS++.
  4. After the sale to AOL the WAIS team did not dissolve but spread out into several different companies, some of which are still alive today. Developer Georgios Papadopoulos founded Atypon in 1996, still an active infrastructure company for scholarly publishers, and François Schiettecatte's version of WAIS grew into ScienceServer and was bought by Elsevier.
  5. Alexa Internet was not only a startup Kahle co-founded but also a commercial data donor to his own nonprofit Internet Archive. Starting in 1998 Alexa passed its web crawls (beginning with a copy of the archive sent to the Library of Congress) to the Internet Archive, which means the same founder was simultaneously building a commercial product that collected traffic data through a toolbar and a free public archive that received that data. One column considers the image of an archive built out of nothing a simplification rumor.

Legend vs. the record

The first growth lever

WAIS had two levers, one after the other. The first was openness: in 1991 WAIS came out as a free Unix release, and without any involvement from Kahle himself, people around the world began setting up their own WAIS servers. In eight months the number of databases grew from a handful to 160+ in eight countries, and users to 10,000+ in 24 countries. This gave the project distribution that a small company could not have paid for itself; later CNIDR's freeWAIS, funded by NSF, continued down the same path.

But an open protocol did not bring in money by itself. The second and far more concentrated lever was a single client: in 1992, thanks to Kahle's acquaintance with Perot Systems head Mort Meyerson, WAIS Inc. built in a few days a system for coordinating data for Ross Perot's presidential campaign across 50 states. The pilot was so successful that by November the system was installed inside Perot Systems itself, and it was this project, not the open servers around the world, that made the company profitable. Openness brought reach; one high-profile client brought revenue.

The paired story

WAIS and Archie were two answers to the same question of 1990–1991: how to search the internet before the web. Archie (McGill University, 1990) indexed only the names of files on anonymous FTP servers (see the Archie dossier). WAIS (Thinking Machines, 1991) searched inside the documents themselves through a protocol based on Z39.50. Both systems gave way to the web in the mid-1990s, and both had a commercial life, but very different ones: Archie's creators founded Bunyip Information Systems in 1992, sold licenses to providers, and dissolved the company in 2003 (see the Archie dossier), while WAIS was built from 1991 as a partnership of four companies, and in 1995 AOL bought WAIS Inc. Kahle put the money from that sale into the Internet Archive and Alexa.

Parallels today (projects from the catalog)

What a builder can take from this in 2026

  1. Give the protocol away for free and sell a service on top of it. The free 1991 Unix release gave WAIS 160+ databases in 8 countries within eight months; openness was not charity but the cheapest distribution channel.
  2. Gather partners by type of resource, not by check size. Content, engine, interface, test audience: WAIS got four different things from four companies, none of which was an investment in itself.
  3. One visible, unconventional client can pay for a company faster than thousands of small ones. A presidential campaign made WAIS Inc. profitable within a few months; the open protocol provided reach but not revenue.
  4. Tying yourself to an existing authoritative standard is borrowed trust, not an obligation to follow it to the letter. WAIS pointed to the library standard Z39.50 but in practice built a simplified protocol on top of TCP/IP rather than the full OSI stack.
  5. A winner's technical superiority often comes down to one specific capability rather than overall quality. Contemporaries saw the main difference between the web and WAIS in one thing, hypertext links between documents; competitors should look for exactly these narrow but decisive gaps instead of trying to be better on every parameter.

Discrepancies and what we could not verify

Sources (primary first)

Secondary, known secondhand: historyofinformation.com; money.cnn.com ("Amazon acquires three Internet firms," April 26, 1999); loc.gov (LC Information Bulletin, November 1998, the Alexa donation); Medium (Brendan Halpin, criticism of the Alexa/IA link, unavailable: 403).

Internet History Keeper

Every historical dossier here is free and open, and it will stay that way. If you want the series to keep going (new dossiers, checks against primary sources, the English edition), support it with a subscription for $9.90 a month. It is support, not access, and you can cancel at any time.

Become a keeper — $9.90 a month →

The same thing, about today

The same breakdowns, but of projects launching right now: what the product is, where the first users came from, how they charge. The card is free, the full dossier is $5 (the dossier itself is written in Russian).

Browse the dossier catalog →