Posted January 16, 2006 10:50 am by with 7 comments

Tweet about this on TwitterShare on LinkedInShare on Google+Share on FacebookBuffer this page

One of the most successful concepts of 2005 was most certainly “tagging�?. Blog posts, search results, news items; if it’s on the web, you can bet that someone has tagged it as something they’d like to go back to or share., Digg and Furl are just some of the many tools available to web users, looking to keep track of their favorite content. But with so many different tagging alternatives, and so many people tagging, how do you keep track of the most relevant content and where can you go to find a consolidated database of the best tagged “stuff�?? That’s where Wink comes in to save the day!

Wink is a search engine that lets you find the results that others think are most relevant. It combines results from all the most popular tagging sites and then adds Google to fill any gaps. The result? Wink uses its TagRank(tm) algorithms to determine which pages are the most relevant while allowing uses to add their own rating.

We caught-up with Wink founder and CEO, Michael Tanne, and asked him to explain the concept behind Wink and his vision of how “tagging�? will impact search results to come.

[Andy Beal] What are Wink’s goals for 2006?

[Michael Tanne] First, we’re working on getting the word out about Wink – it’s a different kind of search engine and it’s not exactly mainstream yet, but for early adopters and social media enthusiasts it has been delivering good results. Second, we will continue to increase the number of sources that we have to include more social bookmarking sites, and other relevant sources that incorporate tags. We’re also going to be working with other groups to advance some of the standards we’re working on for exchanging tags, such as and Finally, we’re going to continue to work on improving the quality of our search results by improving our algorithms and index.

[AB] At what point will Wink be able to solely use its tag search results?

[MT] Wink doesn’t intend to ever rely on Wink tag results alone. Our defined purpose is to provide users with a search across as much of the user intent on the Web as possible. That means indexing bookmarks and tags, ratings and reviews, and other user-generated content from across a large number of sources. Some amount of user intent will be expressed on Wink itself, because it’s logical to do that in the context of search, but people also do these things at other places where the context is right for that, such as on a shopping site, using a social bookmarks tool, on a social news site, etc. And we think that is just as important and valid as tagging they do at

[AB] What do you foresee as the “tipping point” for Wink?

[MT] As far as a tipping point, that probably happens when there are enough users doing enough tagging around the Web and on Wink, that users who do a search on Wink see a measurably better search result – it’s a critical mass of tags, so to speak. At that point there is more of an incentive to use Wink, and that use leads to more user input, which leads to a better experience, and so on.

[AB] What are your thoughts on the future of vertical search? What role will vertical search engines (like Wink) play in the overall search engine landscape in 2010?

[MT] At this point, and for the next several years, vertical search engines will do better within the context of their particular domain than the large search engines. The large search engines have done a great job in large areas such as local and news, and at including quick answers in context – flight numbers, area codes, etc. But when it comes to searching for travel or jobs or events, those are specialized areas where the user is searching for certain fields (or facets) such as location, company, airline, special meals, window or aisle, degree required, etc. Big search engines have trained people to enter a few words into a search box and get 10 results back. They may introduce vertical search but it will be an area where they don’t innovate as fast as those focused on a terrific experience in that specialty. Perhaps some of those will get acquired. Over time, though, I suspect the major search engines will expand and gradually encompass vertical search.

Wink – and companies like Technorati and Feedster for blog search – aren’t so much vertical search, as they are searching a type of information rather than a particular vertical subject area.
Technorati searches blogs – about any subject. Wink searches tags – about any subject. There is room for these approaches for the foreseeable future.

[AB] What measures does Wink have in place to protect its search results against spam/malicious tagging?

[MT] In the world of anti-spam for email, there’s a company Cloudmark [disclosure: Tanne is an investor] which operates a network in which users block spam when they see it and then models take over from there, blocking the spam for millions of users. There’s great power in having an army of users who know spam when they see it. They can be more effective at blocking spam than algorithms alone, as long as is it very easy for them to block things they consider spam, and a model to apply that blocking across the service.

[Update: Michael expands on Wink’s answer to spam in the “Comments” section of this post]

[AB] How does Wink determine relevancy? Can you elaborate on the relationship between the standard search algorithms you are using and your TagRank(tm) algorithms?

[MT] Ahhh, you’re asking for the recipe to the secret sauce? I can just say a little about this. We think that user intent can be inferred from user input such as tagging (but that’s just one kind of input) and the amount a Web page is tagged suggests its relevance. That’s helpful but not enough by itself. You have to take into account who tagged the page and in what context. Also relevance tends to change over time. We have thought through and implemented (and patented) a robust set of algorithms around this, and are tuning and refining every day.

[AB] Which of the big search engines do you see align most with your goals for Wink? Which do you think would find the most value in Wink’s offering?

[MT] So far, Google has been the most aligned and we have a good partnership wherein they provide core Web search for us and we serve ads using AdSense (which of course they share in the revenue). That isn’t to say Yahoo or MSN wouldn’t be good partners but we have found Google to be very helpful so far. We see that Yahoo has taken quite an interest in social bookmarking (by acquiring and social search (with MyWeb and Yahoo360). It’s evident they get this area and are invested in it, but I don’t know if that means they’d be a better partner or more of a competitor. These things have a way of evolving over time

[AB] Thanks for taking the time to tell us more about Wink!

  • Thanks for the insight on Wink Andy. I do have concerns about the “spam/malicious tagging” question. I don’t feel it was answered. How would you stop a company with black hat intentions from creating multiple valid accounts from within, Digg and Furl?? The same sort of thing Google has come up against in it’s Pagerack alg. This seems even harder to attack. I myself can think of two or three ways to create multiple accounts from multiple IP addresses, automated of course, to manipluate the system. In my opinion, I hope Michael can gain traction fast before the black hats start to degrade the service.

  • Good question, I’ll see if he has a more in-depth answer.

  • Wink is going to have a tough time with blackhats when they are pulling in from multiple sites as said before. Sometimes as been proven with Google algorithms are just not enough. Thats why when I came up with Seekum I built it with one thing in mind. The people are always right. It is the only algorithm with only 1 variable.

  • I asked Mike to comment on Mike’s question above. Here is his response…

    Great question. You’ve raised the question of spammers who attempt to promote sites by manipulating other services. Where there’s an audience, they are bound to come. Wink does need to prevent bringing in spam from other sources, as do the search engines when they decide whether to index a site.
    The model we are building is designed to classify spam even if it was tagged by several accounts, and filter it from our input. It’s typically become an arms race between clever tricks to manipulate results and algorithms to block the spam. With user input we add users’ voices in addition to the algorithms.
    That said, we have seen that some sources are more susceptible to spam than others and if a source became overwhelmed with spam we may consider excluding that source. That’s the approach – sorry if I can’t go into too much detail about our specific algorithms. As you said, we too hope Wink gains traction before the black hats try to degrade the service.

  • Thanks for the follow-up Andy and Michael. I assumed that Michael does have a plan for black hats and that he certainly cannot give out the details. The arms race he refers to is similar to the problem the US military is facing in Iraq and faced in Vietnam. Without clear identification of the “enemy” makes fighting the war even harder. That said, I wish Michael all the success and hope that this social approach to indexing becomes as valuable as it’s potential.

  • Matt

    “Also relevance tends to change over time. We have thought through and implemented (and patented) a robust set of algorithms around this, and are tuning and refining every day.”

    Any hints of finding that patent? I ran some searches and didn’t come up with much.

  • Pingback: IrishWonder’s SEO Consulting Blog » Blog Archive » Social Bookmarking and Tagging for SEO()