Google's Founding and PageRank: How Links Became Search's Currency
In 1998 two Stanford students built a search engine that ranked pages by their links. Read how Google and PageRank began and what it means for your website.
Before Google, finding something on the web was a frustrating task. Directories and early search engines such as AltaVista, Lycos and Excite could index a lot of pages, but the results often seemed random or were easy to manipulate. In 1998 two graduate students at Stanford University suggested a different way to measure what mattered on the web, and the search business was never the same.
Before Google
The earliest tools for finding pages were Archie, which indexed file names, and Gopher-based search tools. As the web grew, Yahoo started as a hand-edited directory of links in 1994. Search engines like WebCrawler (1994) and AltaVista (1995) could search the text of pages automatically. Ranking by text alone had a clear weakness: a page could repeat a keyword hundreds of times and appear to be relevant. Site owners learned to do exactly that.
BackRub: the research project
Larry Page and Sergey Brin met at Stanford in 1995. Page was interested in the link structure of the web. Each link from one page to another could be seen as a kind of citation, similar to how academic papers cite earlier work. Page began a project to collect the links, which he and Brin called BackRub, because it analysed the "back links" pointing to a page. Brin joined him to work on the mathematics and on how to store and process all the data.
What PageRank does
The algorithm, named PageRank, partly for Larry Page, gives each page a score based on the pages that link to it. The central points are:
- A link from one page to another counts as a vote for the second page.
- Votes from pages that are themselves important count for more.
- The score is calculated across the whole web, through repeated calculations until the values settle.
The two published their approach in 1998 in a paper titled "The Anatomy of a Large-Scale Hypertextual Web Search Engine", and a related paper on PageRank was written with colleagues at Stanford. The patent was assigned to Stanford University.
Google the company
The name Google is a play on "googol", the number one followed by a hundred zeros, hinting at the scale of information it was meant to organise. Google Inc. was incorporated on 4 September 1998. Early work took place in a rented garage in Menlo Park, California, after the founders received early investment, including a cheque from Andy Bechtolsheim, one of the founders of Sun Microsystems. The first public version of the search engine was already used at Stanford before that.
Several features made Google stand out:
- Better results through link analysis.
- A very plain page that loaded fast on slow connections.
- Servers built from cheap commodity hardware, designed for failures to be expected and handled.
Google started selling text advertising linked to search queries in 2000 with AdWords, which grew into its main business. The company went public in 2004.
The side effects
Once links counted, people started to create them artificially. Link farms, paid links and comment spam appeared. Google responded with a long series of updates and policies, and introduced the nofollow attribute in 2005 so that site owners could mark links they did not vouch for. Today, PageRank is only one of a very large number of signals used in ranking, and Google stopped showing the public toolbar score years ago.
What this means for your site
| Principle from the PageRank era | Practical advice today |
|---|---|
| Links are endorsements | Earn links by publishing useful content, not by buying them |
| Speed mattered from the start | Keep pages light and fast |
| Quality beats repetition | Write for people, not for keyword counts |
How crawling and indexing work
PageRank is only one part of a larger process. A search engine first discovers pages by following links, which is called crawling. It then analyses and stores what it finds, called indexing, and finally ranks pages for each query. For a site owner, the first two steps are the ones you can directly help. Make sure your pages are reachable through ordinary links, that your server responds quickly and reliably, that your site map is accurate, and that you do not block crawlers by accident with a misconfigured robots file. A crawler that repeatedly receives errors or timeouts will visit less often, and pages it cannot reach cannot be ranked at all.
That is why uptime is not just a convenience for visitors. A site that is frequently down for hours is one that the crawler learns to treat cautiously.
Search has changed, the basics have not
Modern search results use language understanding, location, device type and many other signals, and the pages shown for a query are far more varied than the ten blue links of 1998. Even so, the founding principle remains: search engines try to reward pages that other people find useful. A site that answers real questions clearly, loads quickly and is easy to navigate on a phone is in a good position, whichever signals are used.
The lesson for site owners
Google's founding idea is still at the base of search: pages that other people choose to link to are more likely to be useful. A site that is online and quick whenever a crawler visits also does better than one that is slow or unreachable. Fast, reliable hosting is part of that, and a CDN can improve loading times for visitors far from your server. Our WordPress hosting is a good fit if your content lives in a WordPress blog.
