Is Yahoo! (re)teaching us the art of search?

Yahoo! recently revamped the search mechanism on Yahoo.com, and it now includes Search-Assist. The feature is similar to Google Suggest, which has yet to make it out of Google Labs (beta). Although not a defining accomplishment, the functionality demonstrates that Yahoo! may still have something to teach us about the art of search.

There are two primary differences between Yahoo! Search-Assist and Google Suggest, and these variations illustrate that Yahoo! is paying close attention to human behavior in search habits. Whereas Google has gone to great strides to analyze information relationships through complicated algorithms and PageRank, Yahoo! is helping users discover hidden results.

These two screenshots illustrate what I mean. The most visited article on d’bug to-date is a rather short and utilitarian piece that I wrote over a year ago highlighting an unusual browser bug. More than a thousand search keyword combinations are used to find this post, so I decided to try my hand at a few popular variations with both Google Suggest and Yahoo! Search-Assist.

Unlike Google Suggest, Yahoo! is formulating results based upon keyword criteria, as well as concepts related to my search. It just so happens that IE6 was the root cause of my frustration for the aforementioned bug. Seeing the keyword ie 6 within the concepts pane, I begin to realize that I might be on the right track. I also recognize immediately that users have attempted similar search patterns. Clicking on ie 6 brings me to the results page where d’bug is the fourth entry.

Google Suggest will only bring me back results that match identical search terms. This approach negates both human error, and it promotes the misconception that even though I do not know exactly what I want, a search engine can tell me precisely what I need. Yahoo! takes one step back, and delivers an intelligent alternative — a search assistant that makes assumptions, which ironically tend to be more accurate than assertions.

Getting things done

From the Yahoo! Search Blog:

One thing we’ve learned since launching our own algorithmic search engine back in 2004 is that at the end of the day, people really don’t want to search; they want to get things done. Today, we’re launching an all new Yahoo! Search experience that gets users the answers they’re looking for quickly and easily, and often in one search.

This approach and attitude could slowly give Yahoo! a competitive advantage. While Google remains an unconquerable search giant, there is no better adversary than the one already waiting in the wings. Many naysayers continue to write Yahoo! off as old-hat, but a determined focus on serving basic human needs through innovations like Search-Assist could breath new life into the company.

Services such as Yahoo! Answers and Yahoo! News are also given high praise for simplicity and accuracy. Even though Google continues to roll out a barrage of similar offerings, the behemoth is often perceived as an absentee landlord. Until the cable is out, the air conditioner is busted, and the cockroaches are multiplying, there is no solution in sight. Also once hailed as an advertising powerhouse, serious publishers shy away from the strict rules and small profits surrounding AdSense. This is the perfect opportunity for an “underdog” like Yahoo! to steal the show.

Search is still short-sited

To be realistic, Yahoo! Search-Assist is nothing revolutionary, albeit useful and effective. Ultimately, search results are still tracked by users the same now as they were a decade ago. The “Golden Triangle”, or “F” pattern, is well documented in several studies. Both theories suggest that the most prevalence is given to those Web site links and summaries located at the center of the triangle. For everyone but Chinese visitors (who read right to left), this would be the top-left of the screen.

For the search engine elites, playing too much with this formula can be risky. Toying with the method with which we search is acceptable, so long as the way in which we view the results remains pure. However, given the amount of liquid cash at hand, Yahoo! should invest some research dollars into alternative methods of displaying results. Google has been investigating unconventional interface configurations for search results via Experimental Search since May of 2007. No word yet on the success of this endeavor.

Of course, the final frontier for search is personalization through artificial intelligence. This is not the sort of portal personalization hailed as the next best thing before the Internet bubble burst. It is the sort of personal search that is anything but short-sited or generically suited for all users. At the TechCrunch40 2007 conference, one contender rose up to meet this challenge — Powerset.

Whether or not Powerset can upset the balance in the search universe will be unknown for quite some time. Yahoo! could still manage to compliment Search-Assist with a comprehensive suite of search tools. This would help to close the gap on Google, or it could serve to pull the wool over the lamb’s eyes just long enough for another search engine to steal the show.

Leave a Comment

Comments are moderated. No profanity. Only <a>...</a>, <blockquote>...</blockquote>, and <code>...</code> are allowed.

Seperate paragraphs by pressing the "Enter" key twice, or press it once to break to a new line.

5 Comments

#01, Oct 09 2007

Mark Johnson

I’m pretty excited about Search Assist because it offers a real step-up over Google Suggest (which, in itself was an amazing innovation). There are definitely ways that Powerset can compliment Yahoo’s technology — in providing meta-data about answers, instead of just what’s in the query and by offering more information from a deeper analysis of the index. Stay tuned for more =)

#02, Oct 10 2007

Brian

Hi Mark,

Thanks for stopping by :)

I think there is a lot of excitement around Powerset, and people are ready for a renaissance in search. It is an area that has seen little movement in the last decade. An intelligent shift in how search results are returned, as well as a change in the mechanisms that guide those results will surely be welcomed.

Please keep me informed of any happenings on your end — I’m excited to see how Powerset evolves.

#03, Oct 10 2007

Mihai Campean

Indeed Powerset got themselves a real challenge out of building a natural language search engine. I hope that they will really manage to make this idea work because this indeed will represent a huge leap in search engine evolution.

#04, Oct 12 2007

fred

Great article, I might try yahoo again. However, I will like to point out this little error.

“Chinese visitors (who read right to left)”

The Chinese language abandoned right-to-left style long time ago. The only modern language that writes RTL are Arabic, Hebrew, Farsi, and Urdu.

#05, Oct 12 2007

Brian

Hi fred,

Thank you for that correction. I think the right-to-left reading style was something I came across while researching the original “Golden Triangle” tests. For the life of me I can’t find it now. So it might be that because right-to-left reading is still culturally prevalent it had an impact on the results. That would only be my own speculation though.