There's an interesting innovation in the search field that is an effort to leverage social participation for search result relevance. URL.com is a search engine with a difference. The goal of the service is to harness the social power of the search community in order to increase the relevance and usefulness of results.
To see it in action, let's try my default comedy test search for chickens. You'll notice that URL.com returns a consolidated set of results derived from the top 10 results on Google, Yahoo and MSN. Interesting so far, esp. w/ the ranking for each engine displayed there on the left, and the small icons at the top of the ranking columns that let you jump direct to those results, but nothing that blows your socks off....
So what's new? Well, let's sign up and see. If you sign in before you search, then click a result, the result loads in a frame, which gives you the chance to rank your result.
The frame makes sure that you visit the page that you're evaluating, rather than expressing an opinion on the list of results page. In a sense, this relies on the same sort of social contract as a wiki. "First do no harm, respect the common good" etc. Of course, search engines aren't always the premier forum for this kind of ethical web-citizenship, & I'd be interested to see what sort of tools are employed to prevent an American-Idol style "Chickens Rock" campaign from taking over the universe. (Bwahahahahahah!) I'm growing my Bo Bice hair as we speak....
Anyhow... (sanity returns...) there's a Techie About page at URL.com that explains the weights and rankings that are assigned to results. There's also an FAQ, containing the nugget that votes are superceded... If there are multiple votes from a single account, then only the most recent will count. Thus, the universe would have to be conquered by committee!
Privacy is also addressed at the FAQ. URL.com state that they only publicise the links that you vote or comment on... so I guess your personal search history is no more or less problematic than with any other search engine. The long & short of it, I think, is that searches shouldn't be regarded as private, whichever engine you use.
Here's my thought / wish for the future. The evaluation of the page doesn't seem to get at accuracy or "aboutness" as clearly as you might wish for. Your vote for "good" & "not so good" results affects the way that results from that engine are interfiled with results from the other two. It does not unequivocally state "This page was / was not about Chickens..."
Now I realise that we'd be talking 8,000,000 servers to do it this way, and that other engines approximate an "aboutness" voting system by boosting the juice of pages that are frequently clicked results for a given search, (the implication being that regular traffic = aboutness) but if you're asking for my vote, I'd like to be able to say "This page wasn't about Chickens" and have it affect the page in relation to the search, rather than the bulk of engine results in relation to the other engines.
I'd eventually like the judgement to be about the page and not about the engine.
Some challenging new issues, then, but an interesting combination of two different models / technologies, and one worth keeping an eye on.
Filed in: webtech search url.com
To see it in action, let's try my default comedy test search for chickens. You'll notice that URL.com returns a consolidated set of results derived from the top 10 results on Google, Yahoo and MSN. Interesting so far, esp. w/ the ranking for each engine displayed there on the left, and the small icons at the top of the ranking columns that let you jump direct to those results, but nothing that blows your socks off....
So what's new? Well, let's sign up and see. If you sign in before you search, then click a result, the result loads in a frame, which gives you the chance to rank your result.
The frame makes sure that you visit the page that you're evaluating, rather than expressing an opinion on the list of results page. In a sense, this relies on the same sort of social contract as a wiki. "First do no harm, respect the common good" etc. Of course, search engines aren't always the premier forum for this kind of ethical web-citizenship, & I'd be interested to see what sort of tools are employed to prevent an American-Idol style "Chickens Rock" campaign from taking over the universe. (Bwahahahahahah!) I'm growing my Bo Bice hair as we speak....
Anyhow... (sanity returns...) there's a Techie About page at URL.com that explains the weights and rankings that are assigned to results. There's also an FAQ, containing the nugget that votes are superceded... If there are multiple votes from a single account, then only the most recent will count. Thus, the universe would have to be conquered by committee!
Privacy is also addressed at the FAQ. URL.com state that they only publicise the links that you vote or comment on... so I guess your personal search history is no more or less problematic than with any other search engine. The long & short of it, I think, is that searches shouldn't be regarded as private, whichever engine you use.
Here's my thought / wish for the future. The evaluation of the page doesn't seem to get at accuracy or "aboutness" as clearly as you might wish for. Your vote for "good" & "not so good" results affects the way that results from that engine are interfiled with results from the other two. It does not unequivocally state "This page was / was not about Chickens..."
Now I realise that we'd be talking 8,000,000 servers to do it this way, and that other engines approximate an "aboutness" voting system by boosting the juice of pages that are frequently clicked results for a given search, (the implication being that regular traffic = aboutness) but if you're asking for my vote, I'd like to be able to say "This page wasn't about Chickens" and have it affect the page in relation to the search, rather than the bulk of engine results in relation to the other engines.
I'd eventually like the judgement to be about the page and not about the engine.
Some challenging new issues, then, but an interesting combination of two different models / technologies, and one worth keeping an eye on.
Filed in: webtech search url.com
Ramani said... 
TroubledYouth said... 
