AltaVista has launched a search-only site today which follows on improvements the company made to its core database of web pages about a week ago. Called Raging Search, the new site delivers fast and uncluttered listings.
"This is for the search enthusiasts, for the tech and web savvy people looking for pure web results," said Rajiv Parikh, AltaVista Search's marketing director.
The move to launch a search-centric site is significant, because it goes completely against the trend search engines have followed to date. Usually, when a search engine has become popular, it begins adding on "portal" features such as news headlines, horoscopes or free email in hopes of keeping users within the site.
Among the major search engines, the notable exceptions to this have been Google, GoTo and Northern Light. Google, in particular, has won praise from searchers I talk with that are impressed with the quality of its results and the concentration on search. GoTo has also attracted a significant audience and is able to eschew portal features because its pay-for-placement system actually means the search engine makes money when users leave its web site.
For the moment, Raging Search has no banner ads. Instead, AltaVista plans to make money with ecommerce and affiliate links that appear at the bottom of its search results page. Text ads, such as those Google uses, may also be tried in the future, AltaVista says.
Raging Search allows users to customize results in several ways, such as to see up to 50 listings at a time, to display a more compact format, to filter out adult content and to set language preferences, among other options. By default, only one page per web site is displayed in the results, but more can be seen by using the "Results from this site only" option.
Raging Search uses the same web page index as does AltaVista itself, which means there is no need to submit to it separately. However, expect that there could be variations when running the same searches in both places, especially for multiple word queries. Raging Search is designed to be a sort of test bed for improvements that may migrate over to the main AltaVista site. Consequently, a different ranking algorithm may be in use, or Raging Search may process the query in a different manner.
As for the web page index, AltaVista has expanded it to 350 million pages, which the company says are the best on the web. AltaVista has begun using a new "connectivity" graph that analyzes links from over 1 billion pages. From this, the top 600 million are crawled, then the index is reduced to 350 million pages when duplicates, dead links and spam are removed.
Having this better collection of documents should mean an overall improvement in the results users see, AltaVista says. To back its claim, the company says it was ranked the overall winner in a new relevancy survey carried out by ZD Labs. The survey itself has not yet been released, so I can't comment on it further. I'll follow up on it in a future newsletter.
AltaVista has also made small tweaks to its ranking system, primarily to refine and improve its usage of link analysis. The company has also been cracking down severely on doorway pages and says that they simply are no longer allowed in its index.
"We do consider doorway pages to be spam," said Tracy Roberts, marketing director for the AltaVista network. In the past, search engines sometimes have reversed themselves on their anti-doorway page statements, depending on how a doorway page is defined. Roberts specifically cited doorways as pages with little or no real content to them. As I reported in the last newsletter, "machine-generated" pages produced by doorway page programs such as WebPosition have also been targeted.
As always, I'd advise concentrating on content. If you want to be found for a particular topic, then create a page rich in information for that topic -- not just a page that's rich in keywords but which has no inherent value. Think of it this way -- a doorway page is not a destination for users. When users come to one, they are usually moved quickly to another page within the web site. So if you are in doubt, ask yourself whether you'd spend 30 seconds or so reading the page you've created. That's actually a pretty long time, and if you would, then your page probably has enough real information on it to escape being considered a doorway.
In another change, AltaVista has now enhanced its directory to rank sites based on popularity, using links as the means to measure this. It's similar to the system I described at Google last month -- see the article below. Additionally, the directory now includes listings from both LookSmart and the Open Directory. Any sites out of LookSmart will specifically say "From: LookSmart" after the site description.
Overall, I'm pleased to see the new standalone search service. There's definitely a demand for pure search, and even though AltaVista estimates the demand is relatively small, its nice to see them cater to this audience. It will be interesting to see if other services mimic the move.
Search Engine Sizes
This page compares reported sizes of major crawler-based search engines. It hasn't yet got the new AltaVista numbers posted, but it will shortly. It has been updated to reflect the new 500 million page index Inktomi says will go live later this month. Also see the "Numbers, Numbers -- But What Do They Mean?" article for a further explanation of how extra pages are spidered in order to increase relevancy. AltaVista is now taking the approach described for Inktomi.
Missing Pages At AltaVista
The Search Engine Update, March 3, 2000
Have you gotten the dreaded "Too Many URLs submitted" message? This article explains a bit more about what to do. AltaVista still says email contact, as described, is the best way to resolve a problem.
Google Adds Directory
The Search Engine Report, April 4, 2000
The Search Engine Update, April 4, 2000
Describes how Google is using link analysis to relevancy rank listings at the Open Directory
WebPosition, April 2000
This issue of WebPosition's newsletter has specific advice for its users relating to AltaVista and doorway pages generated by the program.