[email protected] wrote:
Why should this be the case? It's perfectly possible to perform a search for some user and delete all data about the process afterwards.
Which means while the search is going on (the queries have been issued to the dependent search engines but the replies have not yet been received) the metasearch engine keeps track of where the query came from so it knows where to send the aggregated response. This is in the nature of having another computer do one's computing instead of using one's own computer to do the job.
With SASS there's no way to verify whether the server deletes data about the search, only the service owner really knows. With metasearch engines there's an additional problem if users are identifiable using search query data or a combination of query data and hosted resource logs (for those search engines organizations that also own hosting resources like Google, Microsoft, and Yahoo each do) because it would be possible to correlate search queries and hits on hosted resources.
We could increase privacy with a search service where the client could do searches against a copy of a database hosted on the user's computer. Thus the user could search that database without revealing those searches to anyone unless the user wants to reveal what they're searching for. The database files could contain a wide variety of data on indexed websites so even if someone knows what database files someone has it's not clear what a user is looking for. And the database files should be freely shared so one doesn't have to get updates from any one particular place. So long as each database file is signed with a trusted key, one should be able to get database files from anywhere.
Research is needed to determine if most users searches could work well enough with a small database to make this a reasonably private search system that is sufficiently effective for providing good results. I expect new additions to search engines would take longer to get to some users, but that could be helped with increased database file sharing.
