Recent research published in the September 2006 edition of Information Processing & Management entitled A study of results overlap and uniqueness among major Web search engines showed that 84.9% of first page search results were unique to a particular search engine. These results are broadly in line with those released by jux2 - a public meta-search engine..
The researchers were working in collaboration with Dogpile.com, another meta-search engine which provided them with some of the query terms used in the work.
While Sci7 agrees that with results of this research in so far as meta-search engines are nesscessary, there is a problem with the existing public services, in that in order to obtain feeds from the major search engines they have to licence the data at significant cost. To support this, many run adverts, often, as in the case of Dogpile, they add the advertising links directly into the search results.
One way around this problem is to run the search aggregation software on your own computer - Apple users will be familiar with “Sherlock”, an application which searches a wide variety of popular and specialised search engines. In a business setting - a web-based meta search engine on an intranet site can provide the ideal “meta-search”, such a search facility can also be customised to reflect a particular business’ requirements. Examples of customisations include searching internal company information, as well as adjusting the weightings applied to particular specialised searches.
Opensearch - from a9.com is emerging as one standard for integrating search results from various sources, and it has a head start on Google’s GData which could well emerge as the winner, even though it has broader applications and is not intended primarily for this application.
There are two open source projects which may provide a useful base for implementing a personal or business meta-search application.