Following a phone call during his trip out West, Allen Stern at CenterNetworks had this to say about Surf Canyon:

“Compared to some of the other search engine addons I’ve reviewed in the past, Surf Canyon offers an interesting and innovative model. Surf Canyon is trying to solve the problem of too many search results for your search queries.”

Also, Kathy Martyn had a rant this morning that went like this:

“Surf Canyon purports to add other engines to the adds functionality and extends search results when you use Google Search… The best part is it also shows results from deeper within the various engines as well, so you don’t have to scroll through page after page of results. Surf Canyon puts them right there (where they should be) on page one! Awesome. We’ll see how useful that is, This is a great time saver and one of those, “why didn’t they think of this sooner” type things since usually when the first 20 Google finds are garbage or adverts.”

Finally, here’s a little feedback from Seguílaflecha.com, translated for you with the help of Google:

“Some results can be so effective that you can save a lot of time in terms of seeking to find and bring together the most relevant results with respect to what is being sought.”

Late last night we released the latest version of our application.

The first important change is that we altered the frequency with which we deliver “recommendations.” With previous versions we would always deliver at least one recommended search result in response to a user action. With v1.1.3, however, all recommended search results must satisfy a minimum relevancy threshold. If none do, then none will be displayed. The exception is that clicking the bull’s eye will always generate at least one recommended search result, as before.

For Firefox, we also changed the navigation to subsequent pages of results at the bottom of the screen. The pages are still rolled out downward, however, the “More Results” link has been removed in favor of the initial navigation provided by the search engine. Here are, respectively, Google, Yahoo! and Microsoft Live Search:

Lastly, this version of the application is compatible with recent changes made to the Google search results page. Earlier version are, therefore, no longer supported. If you don’t have the latest version, please upgrade!

Ed Rasimus, who spent 23 years flying fast jets for the USAF, 10 years doing computer software reviews for ZDNet and who wrote two books on his combat experiences, had this to say today on his blog:

“Searches with Google or Yahoo are enhanced with a nifty extension called Surf Canyon. Get a Google hit list and choose a site as you normally would. When you return to the list, Surf Canyon will have added a sub-folder of additional sites supplemental to the one you just looked at. The drill-down research potential is incredible.”

We are big fans of Ed.

Greg Hollingsworth at How To Split An Atom also had some great things to say in his Top 10 list of Firefox plugins for Networking Addicts:

 “This is a really nice add-on that significantly enhances my search engine results… It makes it a lot easier to dig into your search results, thus shortening the window between searching for and finding what you’re really looking for.”

John Brandon at Computerworld added Surf Canyon to his very short list of “back to school” 2.0 websites. He recommends Surf Canyon for “students who need a powerful search tool for research.”

Brian Deagon at Investor’s Business Daily wrote a piece on search startups entitled, “Many Would-Be Google Killers Boast Big Weapon — VC Bucks.” Surf Canyon was one of the companies profiled and Mark Cramer, CEO, had the following to offer:

But no one has broken into the big leagues since 10-year-old Google. “You have to have something that’s not just a little better, it has to be hugely better,” said Mark Cramer, chief executive of SurfCanyon.com. “It’s extremely difficult getting people to keep coming back.”

So difficult that Cramer, after striving to develop a search engine, changed course and delivered a search application instead. The application, which is downloaded, sits on top of search engines, observes what a person is searching for, filters through it and aims to deliver more relevant results.

“We could have launched a search engine, but the challenge is enormous,” he said. “You can’t do it on the cheap.”

Update ( 10/10/2008) – Front and center, smack dab in the middle, here’s the front page of the August 7th edition of Investor’s Business Daily.

Cuil just launched their search engine which boasts the largest index on the internet with 120 billion pages. (121,617,892,992 to be precise, as posted on their home page.) While the exact numbers are not always made available, Google, Yahoo! and Microsoft also all have 10s of billions of pages in their indexes. Having as comprehensive an index as possible is a fabulous thing, and a very important prerequisite for search since you can’t find anything if it’s not in there, but it does not solve the problem of putting 10 relevant results on page one.

In their paper entitled “Beyond the Commons: Investigating the Value of Personalizing Web Search,” Teevan et al. make the observation that:

“Web queries are very short, and it is unlikely that a two- or three-word query can unambiguously describe a user’s informational goal.”

Ambiguous intent combined with an exploding quantity of content on the internet makes it increasingly difficult to put all of the relevant results on page one while simultaneously eliminating those that are not pertinent.

Very few people venture past the first page of search results to find what they want, so returning hundreds of thousands or even millions of results is of little value to the user. (You cannot look past the first 1000 even if you wanted to.) Even if the user is particularly motivated, the process of digging through page after page of results is nothing short of tedious, which is the reason users will either quickly turn to reformulating their query or abandoning the search.

The problem is too many results!

The solution to the conundrum is to have a greater understanding of the user’s intent in order to more precisely focus the results. One way to achieve this is to get the user to explicitly specify intent by entering more keywords, although getting people to change behavior is not easy. Another way to achieve this is to implicitly infer intent through the type of long-term personalization offered by Google, although this too has a number shortcomings.

The most effective way to resolve this issue is to implicitly infer intent from real-time behavior signals and then immediately re-rank the results, through the use of instantaneous relevancy calculations, so that the most pertinent results are moved to the top while the less relevant are suppressed. Surf Canyon‘s Discovery for Search™ is such a solution. Disambiguating intent “on the fly” not only enables users to continue searching with their current behavior, but no search histories or profiles are required. Furthermore, the signals are strong so that the results can be reordered dramatically and the user can actually “see” the process working, creating a more encouraging and perhaps entertaining search experience.

Today Surf Canyon was awarded “Cool Site of the Day” by… well… www.CoolSiteoftheDay.com. They’re quite cool, too.

Today the latest release of Surf Canyon was unleashed upon the world. The most important development is the enhanced compatibility with the Firefox version of Google Preview, which is available on addons.Mozilla.org. Google Preview “inserts preview images (thumbnails) of web sites into the Google and Yahoo search results pages” and these previews are now available with Surf Canyon’s recommend results as well.

We’re unaware to what extent there is precedent for two Firefox add-ons collaborating to produce an enhanced experience that is greater than the sum of its parts, but Edward Ackroyd at Google Preview provided us with code and instructions that enabled our application to first identify whether or not Google Preview is installed and, if so, insert the appropriate previews in the correct places. We’re pleased to be working with him.

Additionally, the UI for the new version has been translated into Russian, which looks very cool.

David Shaw at Bloggle announced the new release and had this to say:

“[Surf Canyon] really does improve on Google’s search results in real time… Best of all, you don’t move from Google – no new web page… It works extremely well, quickly and smoothly. The results are there if you want them, but if you choose not to use Surf Canyon’s recommendations, your page hardly looks different.”

In other news, Surf Canyon was selected as the Tip of the Day for July 17th by BootLog.co.uk, which has this to say:

“Obviously it can’t work miracles but after trying it for a couple of weeks I have to say it’s quite impressive and more often than not its suggested results get you to where you want to go a lot quicker then trudging through pages of hits.”

One of the techniques that Surf Canyon uses to determine if a search result is relevant to your query is to examine synonyms. Princeton University provides Wordnet, a structured lexicon that acts as a dictionary and thesaurus. There are a few open source Java libraries that provide a Java API to Wordnet. We tried using both JAWS and JWNL, but neither of them provided the response time that the Surf Canyon algorithm requires. JWNL supposedly provides an in-memory map version of the lexicon, but it seems that no one, including us, has been able to get it to work.

These open source libraries implement much more functionality than what we required. All we want to do is pass in a word and get back a Set of Sets of synonyms for that word. For example, if you pass in the word “fair,” the returned Sets should include the Set of synonyms that mean “evenhanded,” another Set of synonyms that mean “carnival,” and another set of synonyms that mean “attractive.”

Following is a 100-line Java class that implements this functionality very quickly. First, make sure your CLASSPATH includes the directory that contains the Wordnet database files. The class reads in 4 of the Wordnet database files. (The file names it uses are the file names from the UNIX distribution. The Windows distribution uses different file names. For example, the UNIX data.verb file is called verb.dat on Windows.)

[Read more →]

v1.1.1 of the extension was released today. Here are the major improvements:

• Now works if your browser does not accept cookies
• User interface translated to Deutsch and Português
• Craigslist enhancements
• Various patches and performance enhancements