Metasearch Standouts and Flops

By JEgererJEgerer (1261516658|%a, %b %e at %I:%M%p)

Metasearch is an interesting topic because it is an area that offers such rewards in theory — but we haven't really realized the superior performance from these tools we might expect. Below are some reasons to use and not use metasearch, and a couple standout tools that actually deliver something you won't find on Google or Yahoo! alone.

Metasearch

For the purposes of this analysis I used 'unfiled' = unified, to describe an engine that simply provides an interface for quickly utilizing multiple engines.

'Filed' = integrated, then describes those sites that file that index the engine together with a backend aggregating and sorting mechanism.


Unfiled Search

In the world of unfiled metasearch the results are subdivided by their origins, there is no back end aggregating and sorting mechanism that tries to cherry pick the best results from each engine.

So then we have to ask ourself if relevance isn't being increased, what benefits would we seek from an unfiled meta-engine to make it a value added tool?

Well, it has to make the job of searching whatever engines it searches easier than doing on your own. So if it can deliver on speed, organization of results, or ease of query.

Let's analyze SearchIO, one of the better unfiled engines on this basis:

flickr:4150794935

Speed

You'll have to test it out or take my word for it, but queries definitely seem to take longer on the meta engine than on the home engines. We can't ignore that the meta site is doing multiple searches at once, but lets be real.. if I were to duplicate what the unfiled meta does with tabs I would be opening up a new tab and searching my next engine while the engine loads results (if I even have that long). All in all I find duplicating the results using tabs of a site like DogPile which searches the 4 main sites takes me all of 5-10 seconds longer.

This is advantage meta, but its so immaterial compared the shortcomings to be discussed.

Organization

Impressions of the interface:

Well… it's pretty cluttered. The internal framing necessary to subdivide the search engine tabes (Google, yahoo, etc.) takes up a lot of the viewing pane. I think I'd rather just have nice clean tabs for each site so I could see more of the results.

Second, notice all the sponsored ads and advertising — you won't find this on one of the main search engine pages (ie Google.com). This is a problem I found was common to meta engine and I think it just reflects their traffic level. If they saw the traffic Yahoo! and Google did they would have other revenue streams, and ample opportunity to spread their advertising a bit to reduce clutter.

Advantage tabs, its sooo much easier to browse the results and frankly… less annoying and these search engines are more aesthetically pleasing even in their basic web design than their meta counterparts.

Ease of Query

This is where meta really drops the ball in my opinion, and it goes largely unnoticed. Take a minute to explore the query below on SearchIO.

allinurl:ford motor company

Perfectly acceptable syntax for google, but Yahoo! syntax would require use to design this query as 'inurl:ford and inurl:motor and inurl:company'. The meta engines do not reconcile these differences, if you'll click the Yahoo! tab you'll notice the query is completely ineffective.

This will become a significant problem as your average web crawler becomes more and more or an expert.

Its easy to see past organizational or speed problems, but when the functionality isn't there, its tab browsing all the way for me.


To me the type of unfiled engine that is useful is one which guides your search to a wider array of sources than you would accomplish by tab browsing. Meta engines that search the big 3 or 4 search engines are useless in this respect, but an engine like Joongel does bring something to the table.

flickr:4150793495

Here there is such an array of sources that I would never think to look at if I was browsing meta-style with tabs.

Filed Search

Filed meta-searches by their definition offer something of value in the way they aggregate and sort results. The idea being that recall and precision will be higher than on the individual sites by virtue of a greater document set and superior results ranking.

flickr:4150796305

Speed

Same as above, not really worth spending much time on because the difference is so immaterial.

Organization

The same problem with over advertising exists with filed engines, even the best ones like Info.com. The problem with internal windows is obviously not applicable, so they are a step up from unfiled engines on organization, but still a step down from Google or Yahoo! on aesthetics.

Ease of Query

This is a hard dimension to measure because I don't have access to their backend systems. But it appears that filed engines suffer from the same non-universal syntax problems that unfiled engines do.

On these dimensions it appears filed engines are just as weak, but that's not the case, some filed meta engines do in fact produce materially more relevant results. Our classroom analysis showed that, Info.com, picture above was one such engine.


To me the relevance difference was not enough for me to switch over to a meta engine in general, and I even prefer to do tabbed browsing when searching multiple engines yet. The theory behind these tools is sound and in the future I expect greater implementation to increase traffic to these sites. In the meantime the real value to meta engines is when they sort an obnoxiously large amount of sources for you, like with Joongel.

Unless otherwise stated, the content of this page is licensed under Creative Commons Attribution-NonCommercial-ShareAlike 3.0 License