Intrinsitivity

How Search Engines are Finding Themselves 3/30/07

Google is fast but not smart. It’s clever, but not brilliant. It’s more R2-D2 than HAL.

Scientists at Google Labs play with the future, just not in public. There is a method to their madness, though, because of course the market isn’t ready.

However, some search engine companies are already waving their wands, as it were, by introducing what appear to be magic formulas. I want to thank this article at Read/WriteWeb (RW) for bringing these different services to my attention.

One example of a new way to search information online is CogHog, a service that claims in this demo to be "The Next Evolution in Search." It searches the Internet by parsing natural language for relevant context, with a technology they brand as CognitionSearch.

I would love to chat with a search engine. In a sense, you already can do this with chatterbots. Windows Live Messenger, for example, has an "interactive agent" called Encarta® Instant Answers. Let’s call her Enca for short.

Enca can tell you anything you want to know, as long as she exists in MSN Encarta®, which it describes as–I know because I asked her–"an online Encyclopedia, Dictionary, Atlas, and Homework helper… your gateway for interactive learning on the Web." She might be a gateway, but she really leaves you stranded at the gate. You still have to read pages, in some cases, of information to find what you want.

So I asked Enca, "What time should I wake up tomorrow?" She replied, "Why don’t you ask me something that’s actually in Encarta?" Women. It was an appropriately sarcastic response, though, I must admit.

RW also brought to my attention a new meta search engine called Agent 55. Agent 55 is still the same ol’ search, but, unlike oldies but goodies like Dogpile and Mamma, this one claims to let you search "a total of 365 search-engines in 44 categories, making it the worlds most extensive meta-search service."

No doubt it is. I can’t help but immediately wonder how much over overlap might exist between these 365 search engines. There is probably a reason only three, Yahoo! Search, Google, and, to a lesser extent, Live Search, are widely known. And I doubt it has to do with a short-term memory problem.

What I like about Agent 55 is its speed and convenience. It uses AJAX to keep you posted on its search progress and also suggests a national search. For me, it suggested us.agent55.com. The only problem with this tighter search is that it is, in my opinion, too tight, searching only three search engines, Google, Yahoo!, and another new one for me, Exalead.

Exalead is worth checking out, since it seems to return results that somewhat differ from what you might expect to find at the other two. Also worth noting is that with Exalead, you can filter your search to results to those containing only audio, video, or even RSS feeds.

One website that looks more like a real live boy–or girl in this case–is AbbyMe. Like the visual meta-search engine, KartOO, AbbyMe employs Flash for its interface. Yet, unlike KartOO, AbbyMe is not a search engine. It is "a new way to create and send phone messages using the Internet." You write to someone and Abby calls that person with your message. Nifty.

This is a sign of things to come, as I have come to see them, so expect more companies to follow in AbbyMe’s footsteps. In fact, as often happens, it is probably a safe bet to assume that similar services are out there or in the works (hopefully without Abby’s longwinded preliminary advertisements). Maybe Google even has something like this up its sleeve.

One final next-generation search engine to notice is Hakia. You can read more about it in this RW article. Hakia claims to understand what your searches mean. If search engines dreamt, this is what they would aspire to be, one that not only knew how to find relevant answers with, as CogHog points out, pattern recognition, but with what we humans would perceive to be intelligent understanding.

In other words, search engines aren’t just trying to pass the Turing Test. They are trying to disprove John Searle’s Chinese Room argument.