Searchmarking with JRB

of :

Search Journal



Time's Up (sort of) 

My Google Fast officially ended yesterday, and truth be told, I have not missed Google one little bit. For a straight-out, old-fashioned, general search I have to say I've been every bit as satisfied by the results from engines like AltaVista, Excite, Alltheweb, and similar kin. In fact, one engine in particular, Teoma, has gain particular favour and could easy replace Google pound for pound, day by day.

There was a time when Google was head and shoulders above the rest, but nowadays the benefit is harder to distinguish. At this point Google real benefits aren't found in the general search, but in services like Google Scholar and Google Local. But technically speaking these services are more akin to dedicated search engines like the ACM-DL, and one might argue outside of the realm of comparison for this experiment.

So, I've decided to continue my fast for a while longer. Who knows how long.

Clustering 

Now here's an idea with some legs to it. Consider a search for the term "apocalypse". For a search like this context is everything. Is the intent to discover speculation on the end of the world, information on the Apocalypse Wrestling Federation, fan sites for a defunct heavy metal band, or a guide to the White Wolf Werewolf game?

Clustering offers a practical and highly usable solution to this kind of problem. Cluster engines provide methods to define a context for your search by offering up topic lists that help users further define the intent of their keyword query. By excluding resources that fail to match the defined context the user is given more meaningful, relevant results. Furthermore, it allows less popular results to bubble up to the forefront, surfacing on that ever-important first page of results.

The technology is a bit immature right now, but the potential is obvious. And the potential is not limited to merely producing more focused, relevant results. Clustering engines can serve as an educational tool by introducing users to the scope, depth, and breadth of a topic, and providing insight into the very nature of each facet.

Some popular cluster engines (warning: some metasearch using google):

Two years back, Applied Semantics had a pretty good jump start on the cluster market with their Oingo engine. However, they got bought out by Google, and seem to have been consumed in whole to fuel the rise of AdSense and Google Sets. Too bad.

Metasearching 

I remember being quite enamored by the first metasearch engine I encountered, Metasearch.com, back in 1995. Back then the results you'd get from one engine would differ so strongly from the results of another that the idea of a unified interface that allowed you to simultaneously search several engines at once seemed powerful, efficient, and wise. Fast forward to 2005 however and it would appear that a lot of the usefulness of metasearching has waned.

So many engines currently produce such similar results that metasearching now seems a bit wasteful. I can only imagine the resources involved in running the same query against multiple indexes, comparing, contrasting, and analyzing the similarities and patterns. To net out with results that are more-or-less identical to the results produced by any of the individual engines involved makes me wonder if this technology might be a dead-end of sorts, cursed to stand as a novelty at best.

Theoretically there is at least one advantage: it does help deprive any particular engine of overt influence. The authority of a single source is diluted by its blending with the competition. I'd like to believe this offers protection against googlification, but alas, the practical effect does lend much credence to this idea.

Some popular metasearch engines (warning: some use google):

Links to more metasearch engines can be found via searchenginewatch.com

a9.com disqualified 

I'm forced to disqualify a9.com from the list of google alternatives as it's apparently "enhanced by Google"". Too bad, the ability to customize the interface with additional results columns for web, image, and book (etc.) searches was a really nice feature.

The Googlification of Society 

Society has a dependence on Google that I find, well, a little frightening sometimes. It's like we all defer to some divine sage, an ever-cognizant fountain of all that is known and wise and good. We put stock in the idea that Google's reach allows it to accurately reflect the knowledge topography of the internet. We also hope that somehow its sensibilities allow it to evaluate and rank information in a way that is reflective of us. And in the process we submit to gentle manipulation, a subtle brainwashing, and usher in the "Googlification" of society.

This phenomenon seemed very overt to me today. On my way to work the cab driver mentioned having "Googled" for something before getting into his taxi that morning. 10 minutes in the door and one of my co-workers mentioned having to "Google" for something when he got back to his machine. "Ask Google" I was told several minutes later in response to a question over ICQ. By 11am I figure I must have heard the word Google 15 or 20 times. Nail in the coffin: shortly thereafter I was on a phone conference with a client walking them through lessons-learned in setting up a Google Appliance.

Imagine if all these people had been saying "Master", "His Divine Shadow", "The Government", or even "CNN". Can't speak for anyone else but that could conceivably get me running for the hills with a rifle on my back. Yet somehow sweet, innocent, trustworthy little Google slips in under the radar.

So, I'm going on a Google Fast. For the next 14 days my new best friends are:

Atom Site Feed

This page is powered by Blogger. Isn't yours?