The Hunt for Search Engine Innovation, Part 1

Originally published in MediaPost

HOW MANY search engines do we really need?

According to the metasearch engine GoshMe, there are more than
500,000 search engines. That’s more than one for every resident of Albuquerque,
New Mexico. I dare you to search them all. If anyone will accomplish the task,
it’s Charles Knight, a search engine optimizer who has made a name for himself
publishing monthly lists of the Top 100 Alternative Search Engines.

I’ve attempted a number of grueling feats in my day. In college, I
won a challenge to see who could eat the most Deadly Chocolate Sins, a rich,
fudgy, warm brownie served at Applebee’s, and I subsequently learned that along
with a sugar high, there’s also such thing as a sugar hangover. I am also one of
few men who will admit to having endured watching nearly every episode of “The
Real Housewives of the O.C.” (the things men do for love). The weekend I spent
sorting through all of the Top 100 search engines wasn’t quite so demanding as
brownie-eating or “Housewives”-watching, but it was up there.

With all these search engines, and I have no doubt that the 100
Mr. Knight compiled were truly among the best, I was mining them to explore
where the real innovation lies. What aspects of all these engines will improve
the search experience for users over the years ahead? Even if none of these are
the next Google, Yahoo, or Windows Live Search, are there diamonds in the rough
that can be polished and adapted into the major engines’ algorithms and results
pages?

For the most part, the answer is no.

The engines on the Top 100 list can be segmented into a handful of
categories, and those categories can be further divided based on which ones will
have a low impact on innovation, and which ones will matter most the rest of the
decade. This week, we’ll look at the low-impact categories, and then next week
we’ll see which categories are more promising.

Low-Impact Engines

  • Clustering/graphic display: These engines organize search results in
    some sort of visual field. Quintura’s among the best of these, and it’s
    potentially useful for academics and brand managers, but I don’t get the benefit
    for general consumers. Gnod clusters results based on specific subjects such as
    authors, as in this example for a search on Philip
    Roth
    , yet Amazon’s recommendations are usually more than sufficient (as an
    aside, my regrets to the good Dr. Oliver Sacks for appearing on the literary map of
    Danielle Steele
    ).
  • Filtering based on categories/recommended keywords: This is one
    feature especially common in vertical search, but it’s also being used by other
    engines such as Factbites. If that’s the predominant
    feature, it’s not going to be incredibly useful, as it’s already being used by
    other engines, notably Ask.com and Windows Live Search.
  • Metasearch/aggregated search: These engines search multiple sites at
    once or individually. Dogpile, Mamma, and Goshme all are variations on the metasearch
    theme, while engines like FindForward allow more features for
    searching select sites one by one. Even if these engines are useful at times,
    Dogpile and its ilk are icons of the Web’s past, not its future.
  • User-ratings/voting: VMGO lets users rank search results. I’m
    skeptical of the longevity of this approach, as it’s too easily gamed and too
    biased toward early adopters. If an algorithm’s that good for natural rankings,
    voting won’t matter, though the whole idea of a Digg-based search engine might
    gain some fleeting buzz.
  • Q&A: These engines, like Lexxe, aim to give you direct answers to your
    questions. For the post part, the innovation here has already happened, as Yahoo
    Answers
    emerged as one of the company’s biggest success stories in recent
    years while Google Answers folded. One of my favorite entrants in the Top 100,
    Ask Vox, falls
    into the Q&A category. Built on the Yahoo Answers API, Vox is a talking
    avatar who answers your questions, and you can add in your own answers when she
    falls short (see this vanity
    search
    as an example). For added fun, Vox says on her MySpace
    page
    that she’s going out with the retired Ask.com butler Jeeves. If you ask
    her directly if she’s in a relationship, she’ll confirm the tryst, though the
    two-timer also says she’s single if you press her.

Even though these categories are low-impact, some of these engines
are innovative in their own way. Quintura keeps evolving and grows more useful
with each iteration, Goshme is awe-inspiring with its breadth, and Vox was so
much fun, I shared her with every visitor to my office last week.

But enough playing around. Next week, we’ll look to the engines
and categories that will fuel the future of search innovation.

2 thoughts on “The Hunt for Search Engine Innovation, Part 1

  1. nice work.
    reality will settle. information overload.
    InLocator.com is next evolution of pc use.
    the Internet dies taken over by the InLocator Technology. An 800 channel vertical locator cluster content distribution platform.
    MyLocator.com is the Anchor site.

  2. David,
    I’d invite you to take a look at our human-powered search engine Bessed at http://www.bessed.com.
    We’re very new, so just scratching the surface of where we want to go, but the idea is simple. Humans can create better search results than robots can. Human editors can 86 the spam, can provide a greater variety of search results and can provide meaningful descriptions that help you judge whether it’s worthwhile to clickthrough to a listed site.
    In addition, Bessed has been built on WordPress, allowing visitors to comment on search results, including requests to have their sites added or to dis other sites they think aren’t good.
    Our challenge will be scalability, but if over time we can cover 80-90% of searches, Bessed will be a valuable alternative to the robot-based search engine approach.

Comments are closed.