Where is Search broken?
Every Wednesday on AltSearchEngines we pay a virtual visit to the offices of an alternative search engine. Today we stopped by Bessed, and CEO Adam Jusko.
Late last year, in announcing plan for his wiki-inspired search engine, Wikipedia co-founder Jimmy Wales called search “broken.” But is it? Is the current gold standard, Google, really in such need of improvement as to justify Wales’ project and the proliferation of new search startups over the past year? Or are these contenders / pretenders fooling themselves? (Or just hoping to eat enough Google scraps to reach the magic 1%?)
To justify its existence, each search startup has a different spin on what will make search better. Human-powered search engines like; Bessed (my site), Mahalo, ChaCha and the planned Jimmy Wales effort believe that human intervention is the answer, while engines like Hakia and the much-publicized, never-quite-released Powerset are betting on semantic search (itself an attempt to answer queries in the way a human would). And of course many of these engines are trying to harness the potential power of social networking to create better results and gain a wider audience.
In short, all of these engines are trying to give you search results that would match the best suggestions of an expert on whatever topic you’re researching. That’s great. But exactly what existing problem are we trying to solve? Where is Google not doing the job that you as a searcher expect it to? Where do you think search is broken?
Last month I blogged about Mahalo, its similarities to Bessed, and my jealousy of its venture capital backing. Aside from my sour grapes, the most important point I brought up, and the one that elicited a response from Mahalo’s Jason Calacanis, was the fact that we disagree on where searchers are feeling the pain.
From Bessed’s perspective, search is “broken” in the long tail, where a lack of trusted sites and trusted pages leaves Google open to being gamed by site owners who cross-link aggressively and where Web sites with good content get buried due to a lack of SEO savvy. It’s a lot harder to be the top result for “Hawaii vacations” if your site is devoid of value than it is to be ranked first for “Mickey Mouse pillow case.” As such, Bessed is spending more of our resources on improving results for the “long tail.”
Calacanis disagreed, saying that Mahalo’s research showed that searchers are more frustrated with the most common searches, where aggressive marketers are successfully getting spam ranked above the best sites. This is presumably why Mahalo’s stated push is toward providing better results for the more common searches.
My feeling is that the majority of search startups are targeting the long tail. No one needs ChaCha’s human guides to find a flight to Honolulu, but they might use them to find out the distance from New York to Honolulu. The point of semantic search is to understand that you want information on a TV show when you type in “this week’s Lost” versus giving you random pages with “Lost” and “weeks” in them. But semantic search doesn’t offer much extra benefit when you’re searching for info on Lindsay Lohan’s latest stint in rehab. From where I’m sitting, the “short tail” seems adequately covered.
However, while those of us looking to be a Google alternative (or adjunct, or search of last resort) have hunches about where search is “broken”, and some of us can point to research that backs us up, none of us really knows. So here’s a chance for you to tell us (and Google, too) where search is still hurting.
What do you want that today’s search engines do not deliver?
August 22nd, 2007 at 3:47 pm
Do two searches on Google:
1. Paris Hotels.
2. four seasons hotel paris
On the first search they miss all the great sites in their top 10-20 searches. You can compare their results to Mahalo for Paris Hotels and see how much we improve their results by taking out spam and pushing up amazing sites and information.
http://www.mahalo.com/paris_hotels
Now, look at search two on Google and you can see when you do a long-tail search for something very specific Google nails it most times.
I know you think the long-tail is where the help is needed–and help is needed everywhere I agree–however the 75 hours of users testing we’ve done say the exact opposite. The top searches are EXACTLY where there is too much information and too many bad sites pushing down the amazing sites.
Can you show me examples of what you’re talking about, because I’m really shocked they we are getting the exact opposite information from the market. Perhaps we’re doing something wrong on our side!
all the best,
Jason
August 22nd, 2007 at 7:47 pm
[...] Efron Where is Search broken? » This Summary is from an article posted at Alt Search Engines on Wednesday, August 22, 2007 [...]
August 23rd, 2007 at 10:31 am
Jason,
I’ll give you an example of what I’m talking about, although it’s not an exact example. One of the things we see often is that when people are searching for specific products, the first search engine results for those products are often comparison shopping sites like Nextag, Shopzilla, Shopping.com, etc. Once you click through to those shopping sites, you often find that they either (1) only show the product from a single merchant partner, so there’s no basis for price comparison so no value is added (and I should add that the one partner they feature often shows up in the regular search engine listings, so for example Amazon in effect shows up twice in the first two or three search results) or (2) the shopping comparison site doesn’t even have that product at all, but has created a ghost page with no content but the right title tag and is just coasting off of its larger SEO reputation, i.e., being a “trusted site” in Google’s eyes.
I agree with your example for “Paris hotels”—Mahalo’s results are better. But the question in my mind is whether people are frustrated by what they find on Google or Yahoo for that query. If a searcher is happy to go off to ParisHotels.com and use that directory to drill down to a hotel, then the question becomes whether Mahalo doing it better is enough or whether Mahalo (or Bessed or any other engine) needs to focus on the searches where there is real frustration.
As I acknowledge in my post, your research says the searcher frustration is at the top; I wonder if that frustration is bad enough to look for an alternative. My belief is that neither you nor I know for sure—which is the whole point of my asking for feedback on where search is broken. Unfortunately so far it’s just a conversation between you and me again.
(By the way, I think it’s interesting you use the “Paris hotels” example, because I believe that’s the exact phrase Jimmy Wales used as an example when he originally called search “broken.” You two must be on the same wavelength, despite his calling Mahalo uninteresting.)
August 26th, 2007 at 11:08 am
Search can be improved at the top because PageRank can always be improved through personalisation, collaborative filtering, spam removal and other means.
But it’s most broken at the bottom, where Google doesn’t understand the semantics of web pages - what’s a 5* hotel, and where is paris. Moreover its the structure of web pages that poses the biggest problem: a page with links to hotels is NOT the same thing as a hotel listing on a single page. That’s why vertical search engines are useful: they understand structure and semantics, alas only for certain categories and not all of world’s knowledge.
August 29th, 2007 at 7:29 pm
[...] Bessed [...]
September 5th, 2007 at 11:35 am
Adam: Can you give us three product examples of where this happens?