Archie was the first search engine, built in 1990 to index files on computers connected to the early internet

Archie came before Google, before Yahoo, before any search engine you have ever used. A computer science student named Alan Emtage built it at McGill University in Montreal to solve a real problem: the internet had thousands of computers sharing files, but no way to find them. Archie worked by automatically visiting those computers, reading their file lists, and storing the names in a searchable database. You typed in a filename, and Archie told you which computers had it.

Archie was not a search engine in the way you think of one now. It did not read the content of files or rank results by relevance. It only searched filenames — the actual names of documents and programs sitting on remote computers. But it was the first tool that let you search across multiple computers at once, and that made it the ancestor of everything that came after.

Key Takeaways

  • Archie, created in 1990, indexed filenames on computers across the early internet and let users search for files by name.
  • Gopher, launched in 1991, added a layer of organization and made the early internet more browsable, though it still required knowing what you were looking for.
  • The World Wide Web and HTML, introduced in 1989–1991, created a different kind of document that could link to other documents and display formatted text.
  • Web search engines like WebCrawler (1994) and Lycos (1994) came later and could actually read the content of web pages, not just filenames.
  • Google (1998) succeeded because it ranked results by how many other pages linked to them, rather than just matching keywords.

Why Archie existed and what problem it solved

In the late 1980s, the internet was not a single connected thing the way it is now. Universities and research labs had computers that could talk to each other, but there was no central directory. If you wanted a file — a piece of software, a dataset, a document — you had to know which computer had it. You had to ask someone, or you had to try computers one by one.

Archie automated that guessing. It ran a program that connected to computers running FTP (File Transfer Protocol) servers, read the list of files on each one, and stored those filenames in a database. When you connected to Archie and typed a filename, it searched that database and told you which computers had the file you wanted. Then you could connect directly to one of those computers and read it.

The name "Archie" was a play on the word "archive" — the tool was meant to be an archive of archives. It ran on a computer at McGill and was available to anyone on the internet who knew how to connect to it, which in 1990 meant people with technical training and institutional access.

Gopher came next and made browsing easier

A year after Archie launched, the University of Minnesota released Gopher in 1991. Gopher was not a search engine — it was a way to organize information hierarchically, like a menu system. Instead of typing a filename, you navigated through nested menus: choose a category, then a subcategory, then a document. Gopher servers held text documents, and you could move between them using straightforward commands.

Gopher was easier to use than Archie because you did not have to know the exact name of what you were looking for. You could browse. But it still required knowing roughly where to look — you had to navigate the menu structure, and different Gopher servers organized their menus differently. There was no single search across all Gopher servers, and there was no ranking of results.

Gopher became popular quickly. By the mid-1990s, thousands of organizations ran Gopher servers. But Gopher had a weakness: it was designed for text documents, and the University of Minnesota began charging licensing fees for commercial use. That opened the door for something else.

The World Wide Web introduced links and changed everything

While Archie and Gopher were solving the problem of finding files, a physicist named Tim Berners-Lee was working at CERN (the European Organization for Nuclear Research) in Switzerland. In 1989, he proposed a different system: documents that could link to other documents, displayed in formatted text with images. He called it the World Wide Web.

Berners-Lee wrote the first web browser and the first web server in 1990–1991. The web used HTML (HyperText Markup Language) to format documents and HTTP (HyperText Transfer Protocol) to move them across the internet. Unlike Gopher, the web was designed to be decentralized — anyone could run a web server, and anyone could link to anyone else's pages. Berners-Lee released the technology freely, with no licensing fees.

The web grew slowly at first. In 1991, there were only a handful of web servers in the world. But by 1993, the Mosaic browser made the web graphical and straightforward to use, and adoption accelerated. Suddenly there were thousands of web pages, and they were all linked together in ways that were hard to predict. You could not browse a menu structure anymore — you had to search.

The first web search engines arrived in 1994

WebCrawler, launched in 1994, was the first search engine built for the World Wide Web. Unlike Archie, which searched filenames, WebCrawler actually read the text content of web pages. It sent out a program called a crawler or spider that followed links from page to page, downloaded each page, and indexed the words in it. When you typed a search query, WebCrawler looked for pages that contained those words.

Lycos launched the same year and worked the same way. Both were significant improvements over Archie and Gopher because they could search the actual content of documents, not just filenames or menu structures. But they had a problem: they ranked results by how many times your search words appeared on the page. A page that mentioned your keyword 50 times ranked higher than a page that mentioned it once, even if the second page was more useful. This made it straightforward to game the system by stuffing keywords into pages.

Over the next few years, more search engines appeared: AltaVista (1995), Infoseek (1994), Excite (1995), Yahoo (1995). Yahoo started as a directory — humans organized websites into categories — but added search later. By the late 1990s, there were dozens of search engines, and choosing which one to use was a real decision.

Google changed how results were ranked

In 1998, two Stanford University students named Larry Page and Sergey Brin launched Google. Google used a different ranking system called PageRank. Instead of counting how many times your search words appeared on a page, PageRank counted how many other pages linked to it. The idea was straightforward: if many pages linked to a page, that page was probably important and relevant.

PageRank worked better than keyword counting because links are harder to fake. You can add keywords to your own page, but you cannot easily convince other people to link to you unless your page is actually useful. This made Google's results more relevant, and people noticed. Google also had a cleaner interface — just a search box, no banner ads or cluttered menus — which made it faster and easier to use.

Google was not the only search engine in 1998, but it was better. By the early 2000s, it had become the dominant search engine. Most of the others — AltaVista, Lycos, Excite — eventually shut down or were bought by larger companies. Yahoo tried to compete but eventually gave up and licensed Google's results.

How search engines work today builds on those early ideas

Modern search engines still use the basic ideas from those early days: crawlers that follow links and read pages, indexes that store the words on those pages, and ranking systems that try to figure out which pages are most relevant. Google added machine learning and artificial intelligence to the ranking process, and it now considers hundreds of factors — not just links, but user behavior, page speed, mobile compatibility, and more.

But the core idea is the same. When you type a search query into Google, Bing, or any other search engine, you are using a tool that is a direct descendant of Archie. Archie searched filenames. Google searches the content of billions of web pages. But the problem they are both solving is the same: helping you find what you are looking for in a vast collection of information.

Frequently Asked Questions

Why did Archie disappear if it was the first search engine?

Archie only searched filenames on FTP servers, and as the World Wide Web grew in the 1990s, people stopped using FTP for most things. Web pages became the main way information was shared, and web search engines like WebCrawler could search the actual content of those pages. Archie became obsolete because the internet itself changed, not because a better search engine replaced it.

Was Google really the first search engine to use links for ranking?

No. A search engine called Backrub, which was Google's original name, was the first to use link-based ranking. But Google was the first to do it at scale and with a clean interface that regular people wanted to use. Other search engines existed before Google, but Google made the idea popular and profitable.

Could you search the entire internet with Archie?

No. Archie only indexed computers that were running FTP servers and were willing to be indexed. Many computers were not, and Archie had no way to know about them. The internet in 1990 was also much smaller — thousands of computers instead of billions. Even so, Archie could not search all of them.

Why did Yahoo start as a directory instead of a search engine?

Yahoo's founders believed that humans could organize the web better than automated crawlers could. They hired people to visit websites, read them, and sort them into categories. This worked well when there were thousands of websites, but it became impossible when there were millions. Yahoo eventually added search and licensed results from Google because human organization could not keep up with the web's growth.

Do search engines still use crawlers?

Yes. Google, Bing, and other search engines still send out crawlers that follow links and read pages. The crawlers are much more sophisticated now — they can read JavaScript, understand images, and detect spam — but the basic idea is the same as it was in 1994 with WebCrawler.