What is the true nature of Web search? What are we really doing when we use a search engine? Are we looking for something we already know about, or are we embarking on a journey of discovery? Have you ever done a search that took you in a completely unexpected, yet satisying direction? Perhaps you have also searched, and found yourself lost on a virtual dead-end street. Have you ever wondered about the difference between these and other search experiences?

     If you think about it in depth, you will find that what seemed simple and straightforward to you, becomes quite complicated when you try to break it down into discrete components. That is the challenge faced by all search providers: making the complex seem simple. One-size-fits-all. The Google home page is the archetype of the simplistic approach; white page with query box. Behind the scenes, thousands of computers work at a furious pace, trying to decode the meaning of your specific query. Millions of calculations are made, and a list of 10, or maybe 30 results (out of umpty-million) is the output. The user has no idea how they arrived; they just exist. Take it or leave it.

     But what if you could take a deeper look into the structure of the process? What if you could view the semantic relationships between your keywords and other relevant words? What if you could explore different patterns and combinations of meanings, in a fluid, dynamic real-time interface? What if you could change the emphasis of various keywords? Would you perhaps emerge with a deeper understanding of the information you are seeking? Would you perhaps discover new concepts as you trace the threads in this inter-related web of meaning?

     I'm happy to say that these are no longer just idle questions. A new breed of search engine is on the rise, and Quintura is one of its finest examples. Quintura turns search into an interactive journey of discovery. Along the way, you will find new ways of thinking, new paths to explore, and new destinations. I like Quintura so much, I'll go out on a limb and say this is the Next Big Thing. Or rather, Quintura is an example of how the Next Big Thing will operate.

     If you are familiar with the concept of the tag cloud, then Quintura will immediately make sense to you. One of the shortcomings of the traditional tag cloud is that, as more descriptive tags or keywords are added to the cloud, it becomes cluttered and unusable. Users of the delicious social bookmarking service will probably know what I'm talking about. The convenience and free-association of tagging becomes its eventual downfall. But Quintura has a unique and very savvy way of dealing with cloud clutter. Quintura's contextual interface shows just a few of the keywords which are most relevant to your query. Mousing over a keyword will cause more related keywords to surface - and with them, a different results set. Quintura calls this navigating the layers of the cloud. I call it a fantastic way to explore semantic relationships on the Web. Using Quintura is a much more fluid, intuitive experience than using the mighty Google.

     Quintura accomplishes this using proprietary neural network algorithms. Somehow they are judging semantic relevancy based on the context of the keywords:

www.thegeniusfiles.com, on Flickr">

That is a distinctly different approach than Google's Page Rank, which scores search result Web pages based on how many other Web pages link to them. Google's approach really is more an indicator of traffic, not necessarily relevancy. In other words, you see a lot of cars driving on I94 to get from Minneapolis to Madison, so you take I94 as well. But if you really meant to go to Madison, Minnesota, then you'd be going off on a wild goose chase because those other cars on I94 are actually headed to Madison, Wisconsin. Page Rank will tend to advise you to take I94 - whereas with Quintura, you would be more likely to discover the way to Madison, Minnesota. Well, at least it's a more entertaining experience! (Note: there are some recent indications that Google is beginning to place more emphasis on semantic relationships. The display of that information is however still static). The Google Page Rank Model

Things I Like about Quintura

     Sometimes it's the simple things which make for a good user experience. I like the fact that Quintura works so well without requiring an add-on. I particularly like the fact that Quintura is easy to install into the Firefox Search Bar. Once you do that, you won't have to type in the Quintura URL any more. Just hit CNTR+K to focus on the Search Bar, then F4 to show the list of your installed search engines. Arrow down to Quintura, hit ENTER. Type your query, hit ENTER, and your results page will open directly. (To open in a new tab, just hit ALT+ENTER instead. To set new tab as default, type about:config into the address bar, and then put the following into the filter box: browser.search.openintab then double-click the value to change it to true.) An advantage to this approach is that your specific query results page can be easily bookmarked for later reference.

     I also like the tabbed interface which allows you to switch over to image or video search without changing your query.

     Another thing I like about Quintura is that you can easily convert a query results page into embeddable code for display on another Web page. They could make it even easier if they supported one-click embedding into Blogger, Wordpress, Facebook, MySpace, etc the way some other widget publishers do. I think it would be a smart use of development effort: every time you make it easier to use your service, you are also lowering the barrier to adoption of that service.

     I also like that Quintura makes it easy to share the query results page link by email. However, an address book with auto-complete would make this feature easier to use. Google's Contacts API, or some similar 3rd-party contacts API might be a good fit here. It makes sharing easier.

     I like the fact that Quintura has a query history navigation feature, and a Home button. I like the way a caption pops up when you mouse over these buttons; it's a nice touch. I'd like to see this functionality expanded, perhaps with a way to save the query history for later reference.

     Perhaps Quintura could best address the problems of persistent history, sharing, contact management, etc by becoming more social. Yes, that dreaded, over-hyped word 'social.' But by providing unique user accounts, Quintura would be laying the foundation for a whole new layer of semantic data. Surely that could only help advance the development of the search service. What if Quintura allowed you to save and tag an individual URL from a query result set? It could add a human element of semantic analysis. Perhaps such a hybrid service could take the stagnating delicious paradigm to the next level.

Things I Don't Like about Quintura

     Sometimes, the 'cloud layers' of a query don't yield much more than the initial query results. This could be due to a number of different factors, some of which may be outside of Quintura's control. Or it may be due to limitations in the Quintura search algorithms. I have no way of judging it; it's just an observation.

     In my testing, sometimes the video thumbnails failed to display properly. I'm not sure why. The text descriptions and links were there, so it was still usable.

     Quintura can be slower to use than the generic Google or Yahoo search. It's JavaScript based, and the client machine has to process the visual effects of cloud navigation. My netbook is kind of sluggish with it. A higher-spec'd computer gives a more fluid experience.

     Quintura Site Search works imperfectly with Blogger. That's too bad, because I think a lot of users of the Blogger platform would really appreciate the Quintura interface. Too bad the 'more results' page doesn't work like it should. Hopefully, this issue will be resolved soon. My requests to Quintura for comment have so far been answered with the generic "we'll consider it." If you would like to have Quintura Site Search better support the Blogger platform, let them know.

     Web search is a utility which most of us take for granted. Much of the time, we don't really think about the search engine itself. In fact, Google has dominated this space to such a degree that it has become a verb: "Google it." The presumptive access to information is definitely a benefit to society, and yet it also has its limitations. Google's simplified, keyword-based approach tends to work against its very mission to be all things to all people. To organize all the world's information, yet access it from within the confines of that little white box: this is a paradoxical mission. Making it work requires trade-offs and compromises. It requires the user to think a certain way. It implicitly works against ways of thinking which are 'outside the box.' In a functional way, it almost positions Google as the new Thought Police, all without the user ever really being aware of it.

     So if Google (and its direct competitors Live Search and Yahoo) limits the ways you can even think about finding the information you want, what can be done to overcome those limitations? One popular approach is to use Google APIs as a back-end, and put a new and different front-end on it. What does this mean, and why would you want to do this?

     There are several reasons. Google has a massive infrastructure of high-powered servers, fast network connections, and a vast index of crawled Web pages. This represents an enormous capital investment, and a very significant barrier to entry. Why duplicate that investment when Google is making it available as part of their Open Web strategy? Why not leverage it as a launching point for the next evolution in search?

     You may be thinking, but didn't I just say that Google is the Thought Police, blah blah Google bad. Not exactly. You see, on some level, it's all just zeros and ones - machine code. The interface (in Google's case, the Google home page with its familiar search box) is what facilitates and also limits your interaction with that machine code. User interface, or UI, determines what you can do, how easily you can do it, and what you can see and hear. UI can take many forms, styles and functions. Google has chosen a simplified UI in order to appeal to the widest possible audience. But by using Google APIs, you can put a different UI (front-end) on top of the infrastructure (back-end).

     This is a very rational and effective way to maximize scarce resources in order to develop a new service (lower barriers to entry). Many newcomers are using this approach to develop the next step in search evolution. ManagedQ is one such upstart which puts a visual, semantically-interactive face on Google. While that may not sound like much to get excited about, remember that it's the experience of using something which matters more than the words used to describe that experience.

     ManagedQ does several things which the generic Google Search can not do. The most immediately noticeable of these is a visual preview of search results. These are laid out in a grid of thumbnails, with a link atop each and a text snippet at the bottom of each. In of itself this isn't terribly remarkable; a lot of services do variations of previews. This leads into the second major feature of ManagedQ: something they call the "Executive Summary." The Executive Summary is a sidebar list of semantically-related keywords which ManagedQ considers to be significantly related to your original query. They are grouped by Person, Place, and Thing. This is really what makes ManagedQ interesting: when you mouse over the Executive Summary keywords, ManagedQ immediately highlights them as full and partial occurrences in previewed query results text. At the same time, a secondary query box pops up which allows you to input additional text to find in the results. You can use regular expressions and wildcards too. The effect is to allow you to quickly drill down into multiple levels and dimensions of information without leaving the original query.

     That is useful, but what if you want to zero in on one particular result page? ManagedQ allows you to do so, without leaving your search result page. They call this a "Managed Result." Then you can back out to look at more results, or further refine your query with additional keywords.

     Does this all sound a bit complicated? Consider this: try describing in words the process of entering successive queries into Google, sifting through the results, trying different combinations of keywords, repeating the search, etc until you find the information you are looking for. Does that seem complicated? But you are used to it, so it seems simple. The ManagedQ approach is very simple to use, but describing it seems complicated. It is something you have to try in order to understand it.

     Fortunately, they have made a very nice, brief tutorial so you can see exactly how it works. They have also made it very easy to add ManagedQ to your search bar; something I highly recommend doing. If you combine that with a handy add-on for Firefox called Context Search, you will be able to highlight any text on a Web page and send it directly into a ManagedQ query result page.

     If I could suggest additional features, I'd like to have the ability to save a search results page, save an individual result, annotate and tag saved results, and embed a search in another page. Some form of RSS syndication would also be nice for queries which update frequently, like news articles. It would also be nice to be able to exclude individual result previews or keywords. Perhaps some of these features will make their way into future iterations of ManagedQ. As it stands, ManagedQ is a worthy alternative to generic search engines; especially when you need to go into deeper levels of meaning.

ManagedQ: The Search Application