Google Has its Own 'Live' Search Lab

It was Google that first made developers begin thinking of asynchronous JavaScript in the context of search technology, with the development and rapid expansion of its Google Earth, Maps and Gadgets features. But with Microsoft’s new, all-out effort with Windows Live Search to finally claim a serious share of searches, suddenly the question has become more pointed:

Should AJAX be employed for everyday searches, or should the search page be left uncomplicated?

Recently, Google has been floating a little trial project to come up with the answer. Entitled SearchMash, the system is essentially the AJAX version of Google Search, with some subtle adjustments that the amateur Web user might notice, and whose existence that may slip right past some veterans.

For instance, the front page gives a simple line of instructions, “Type one or more search keywords to get started.” Once you’ve typed something, this line changes to read, “Hit enter to get results.” It’s an extremely simple example of AJAX at work: responding to user input by altering the model of the page, even slightly. Sometimes, if you wait just a moment, this line will give you pointers or offer tips.

A more engaging example concerns the search results themselves. Google has always displayed the first ten searches, before showing you a “G-o-o-o-ogle” tab that offers to take you to the next ten. This task typically requires a page reload. With AJAX, the existing page can allow itself to be changed internally without being reloaded – only the changes are requested from the server.

This mode of operation is being demonstrated by Microsoft now, with Windows Live Search’s image search feature. There, you simply scroll the internal vertical scroll bar down to see more image retrieval results, each of which is loaded into memory as necessary, without forcing a page refresh and without the existing results having to be cached locally.

SearchMash tries applying this principle to textual searches. At the bottom of each batch of ten is a link that reads, “more web pages.” When you click that, SearchMash renders the next ten searches below the existing ones, so the topmost results don’t get dropped into the browser cache. There’s a tradeoff: SearchMash doesn’t tell you how far you could go before you reach the end of the list. Here, Google could be trying to find out how useful it really is for users to know whether there are thirteen results or thirteen hundred – perhaps users only make use of the first three pages’ worth.

It’s apparent that Google is tracking how such subtle adjustments to the search experience are used. From the Features page, which shows you the handful of concepts that SearchMash is testing, you can give each feature a relative thumbs-up/thumbs-down by responding to the question, “Is this useful?”

One feature whose usefulness may have yet to present itself is the drag-and-drop reorganization of search results. If you don’t like the fact that an obviously matching result came in at #3 (SearchMash’s results are enumerated), you can drag it up the list with your mouse and drop it down at #1.

While this gives you the tools you need to move BetaNews search results higher up on your browser, this doesn’t actually change how SearchMash (or Google) would enumerate the result if you tried it again. “This is just for fun right now,” reads the Features page, “but we have some ideas for how to use this.”

Google isn’t exactly being secretive about this project – it does take credit for SearchMash deep in the site’s credits, and apparently did make investors aware of its development some weeks earlier. But why isn’t this a feature of Google Labs, where the trademark could be made more prominent?

As a representative of the company told SearchEngineWatch, one of the factors the company is studying is the effect of its own trademark on its user. “One of the important factors we wanted to address was the influence that may come from Google branding,” the representative wrote. “Creating a separate site will help us gather more objective data about user response to new interfaces.”

One reason Google may be interested in the level of its own influence may have to do with how it could color user’s perceptions of the viability of new features, one direction or the other. As popular blogger David Naylor demonstrated in a recent review, the drag-and-drop feature seemed to be made less relevant by the fact that it was Google that came up with it. “Thought you’d best bite the bullet and admit your relevancy sucks, yeah?” Naylor wrote. “I can’t wait to start abusing this when they use it to help with rankings.”

As veteran technical book publisher Joe Wikert noted on his blog earlier this month, Google’s veil of faux secrecy could end up sending red flags. Citing SearchMash’s statement in the press that the Google trademark was omitted to enable a more objective measure of usage, Wikert wrote, “The existing Google search interface is so simple and clean it’s hard to think there are any elements on it that would impede their research on SearchMash. Plus, the only way people are discovering SearchMash is through blogs and other links which refer to it as a Google property, so anyone using SearchMash is coming at it with Google in mind.”

It was Werner Heisenberg who messed up the realm of classical physics permanently by proving you can’t observe a physical system without changing it. As Google proceeds to observe the behavior of its users –- whether or not it wishes to claim them -– its impact as a sociological force remains unmistakable, and perhaps inescapable.