AltaVista

» see altavista.com as it lived, on the wayback machine

~ the goodbye ~

altavista launched on 15 december 1995 as a demo project at digital equipment corporation. DEC was a hardware company. they had built a powerful new server (the alpha 8400) and needed a public-facing demo to show what it could do. one of their engineers, louis monier, had built a web crawler. crawl the entire web, index it, expose a search interface. that was altavista.

this sounds normal in 2026 because google does it every day. in 1995 it was new. previous search engines (yahoo, lycos, infoseek) used directories or limited crawls. altavista crawled aggressively and indexed everything. by 1996 it had over 30 million pages indexed, which was about a third of the entire web at the time.

google launched in september 1998 with a better ranking algorithm (pagerank). within five years google had displaced altavista. compaq acquired DEC in 1998. yahoo bought altavista in 2003 for about 140 million dollars. yahoo ran altavista in slow decline for ten years before finally turning it off on 8 july 2013.

i used altavista for the first half of 1999 when i was getting online with dial-up. then i switched to google. the switch happened pretty fast. once you tried google, altavista's results felt cluttered and slow.

~ the chronicle ~

~ what it actually did ~

altavista crawled the web. it parsed pages. it built an inverted index. when you typed a query, it returned a list of pages containing your search terms, ranked by some combination of term frequency and recency. you could use boolean operators (AND, OR, NOT). you could use phrase quoting. you could use the "near" operator to find words within a few words of each other.

the boolean syntax was actually useful. you could write "linux AND tutorial NOT redhat" and get a focused result list. modern search engines have mostly removed these operators in favor of natural-language queries that depend on the algorithm doing the right thing. the altavista approach was less smooth but more controllable.

the result page showed about ten results per page. each result had a title (linked), a snippet (the page's first lines or a relevant fragment), the URL, and a small modification date. the page header had the search box and a count of total results. result counts often went into the millions, which felt impressive in 1996 even though most of those results were not useful.

~ how google won ~

the standard story is that google had a better ranking algorithm. that is true but incomplete. google had several advantages.

pagerank. google ranked pages by how many other pages linked to them. this was a much better quality signal than term frequency alone. for many queries, google's first result was the obvious right answer. altavista's first result was often a spam page with the right keywords stuffed in.

simpler interface. altavista's homepage by 1999 had become a portal: news, weather, stock quotes, a shopping section, plus the search box. google's homepage was just the search box. when you opened google you searched immediately. when you opened altavista you scanned around looking for the search box.

no ads on the search page. google's early years had ads in a clearly-marked sidebar but the main results were ad-free. altavista had inline sponsored results that were not always clearly distinguished. user trust dropped accordingly.

speed. google's pages loaded faster. the difference was visible at dial-up speeds. when search is the thing you do most often online, two seconds saved per query adds up.

~ how it died ~

yahoo bought altavista in march 2003 for 140 million dollars. they kept the brand alive but in 2004 yahoo switched the underlying search engine for altavista to its own technology, then later to bing in 2010. so altavista from 2004 onwards was just a yahoo skin on top of someone else's search.

on 28 june 2013 yahoo announced altavista would be shut down on 8 july 2013. it was part of a bigger yahoo housecleaning round under marissa mayer. several yahoo properties were retired in the same window. nobody noticed altavista going. the audience had been gone for a decade.

~ what remains ~

the case study. altavista is the original example of a search engine getting beaten by a better algorithm. every search company since has been compared to "altavista vs google". the lesson is that ranking is the product. the rest of the search engine is plumbing.

the boolean operators. modern search engines hide them. some power users still know to use site: and quotes and intext: in google. those operators are descendants of the altavista syntax. google still respects them, mostly, even though most users have never heard of them.

the era. there was a brief moment when search was new and everyone was figuring it out. altavista was the dominant search engine of that era. by 2003 the era was over and google ran search. the altavista years are the period when a search engine could be a research demo for a hardware company. that does not happen anymore.

@}-->--

              )
             (_)
            .-'-.
            |   |
            |   |
            |   |
            |   |
          __|   |__   .-.
       .-'  |   |  `-:   :
      :     `---'     :-'
       `-._       _.-'
           '""""""

« previous  ·  random  ·  next »