Is consistency truly the hobgoblin of little minds, as Ralph Waldo Emerson claimed in his essay, “On Self-Reliance?” The savvy searcher is more likely to expect inconsistency from search engines. Expecting inconsistency in the interface, functionality, features, and even results keeps searchers sane. Remember, too, that the exact quote is about a “foolish consistency.”
The rapid pace of change at Google and other search engines must mean that they find their own consistency foolish enough that they must constantly change it. Indeed, several search engines have recently made changes to their search interface and results presentation. The changes range from minor to a near-complete overhaul. Users are left to figure out the changes and find new ways to make search work. Apparently, search engines enjoy keeping their users guessing, as evidenced by the hundreds of tests run by Google and Bing and the major changes appearing at other search engines.
In the early days of web search engines, searchers were amazed and somewhat frustrated to discover that AltaVista, for example, would give different results for the exact same search (and this was at a time before personalization and localization took precedence). A group in a hands-on class would all run the same search on the same search engine, and students would get different results.
Big Data Encourages Experimentation
Fast-forward to 2013, and the differences in results are far more noticeable. Such a large number of searches occur on both Google and Bing that it becomes very easy for them to run experiments on all sorts of aspects of the search interface, search results, ranking, ads, and design. With millions of searches per day, a search engine can quickly gather Big Data about differences in searcher behavior. The companies can run an A/B test to analyze searcher responses to a change and then use that data in the decision process.
In the past year, both search companies have mentioned how frequently they run such experiments. In October 2012, Google’s chief economist Hal Varian mentioned that Google searchers may see a dozen or more experiments on any search (searchengineland.com/just-testing-google-searchers-may-see-up-to-a-dozen-experiments-141570; Varian’s quote occurs 26 minutes into a video recording of the talk). Many of these experiments are so minor that most searchers will not notice any difference. It could be a few pixels difference in spacing within a result, the color of a link, or the arrangement of ads. Or it could be a more significant change to the relevance ranking or placement of results on the page.
The Bing Search Blog describes a very similar process in a June 2013 posting (bing.com/blogs/site_blogs/b/search/archive/2013/06/20/boards.aspx). “On any given day, Bing runs hundreds of experiments that help us improve the search experience. Most are nearly invisible, ranging from small tweaks to the search ?algorithm, to changes in our ad selection, to optimizations in the color, size, and placement of features.”
While many of the tests may seem inconsequential, some end up making bigger waves. The change in location of the cache link during the past few years is one example. Those changes were based on experiments. One result was movement at both Google and Bing of a visible cache link on the search results page first to a more hidden location in the right-hand page previews, and then to the current location (at least at the time of this writing) of a small green triangle after the URL that must be clicked to display the cache link. Changes to the advanced search pages, Google’s move of the search tools from the left to the top, and the expansion of subsite links are some of the more significant changes that derive from analysis of data from the tests.
In this constant experimentation and analysis of search interactions Big Data, Google performs more than 5,000 experiments per year that can result in more than 500 updates per year for both the organic search results and another 500 for the ads displayed. The search engines’ goals are laudable—to improve the search experience. However, for the frequent searcher, it can be disconcerting to see such variations and inconsistency. The loss of search features (from the +, to the tilde, to disappearing search tools) can make it harder for the knowledgeable searcher who was planning on using that advanced search feature. You have to first notice the difference and then find an alternative way to get the same type of results, or at least some that end up having the desired information.
More Experiments, Lasting Longer
At both Google and Bing, I have noticed more experiments in the past year—and they seem to last longer. It can be difficult to tell if it is an interface change or just an experiment, although experiments are rarely announced, while major feature changes usually rate a post in the search engines’ blogs. Before Google removed the site previews, one of my browsers was consistently showing that experiment while other browsers (and other computers when I was not logged in) did not. A bit later, in Chrome, I was seeing another experiment that removed the black (or gray) bar at the top and replaced it with an Apps button, somewhat more similar to the Google App approach. That experiment did not (yet) turn into an adopted change.
With 5,000 experiments and only 500 updates, it may be a good estimate to say that only 1 in 10 experiments will lead to an actual visible change. Meanwhile, at Bing, search history links for logged-in users moved from the lower right to just below the search box. On search results pages, suggested searches are now sometimes displayed below the search box.
How much of this type of new design will still be in effect by next year is anyone’s guess. It certainly creates a challenge for teachers who plan on showing features (now defunct) such as Google’s tilde for related terms (Google originally called it synonym searching) or search tools for translated pages, similar pages, and pages with images. Since there is rarely any advance notice, you can plan a lesson on such features only to see Google remove those options right before or after the training session. But it is also a teachable opportunity to remind others that features and services may or may not continue to be available. The recent demise of popular services such as Google Reader, iGoogle, and the + symbol for exact searching are good examples of this.
The Big blekko Change
One advantage of being a smaller search engine is that it is easier to make a major change without as much of an impact. Launched in late 2010, blekko has become known as an alternative search engine with a smaller, more focused database that aims to include only quality websites. Initially, blekko had extensive link and crawl data that later moved behind a paywall as SEO data. During the past few years, blekko has made several incremental changes to its search interface and results display. However, at the end of May 2013, blekko announced a major site redesign.
Its blog post says that the redesign includes three main features:
While none of these sound that all that major, the main screen and the search results pages are all vastly changed, missing many of the features previously available, and emphasizing the multiple categories.
The new look and interface is completely different from the old one and is based in part on its mobile app, izik search. Instead of trying to get users to add slashtags to search terms, the new search results page automatically suggests top categories (based on blekko slashtags) on the left, each with a different color. The categories, starting with “Top Results” each have a few results displayed with the option to display more by clicking the down arrow, which also minimizes the other categories. To see even more results, searchers can scroll sideways using the arrows. Choose another category to open that one and minimize the others. Images and related searches may show up in the right column, and for some searches, a definition or quick answer from Wikipedia may show up above the other results.
The interface is worlds apart from the old blekko and most other search engines. It is encouraging to see this type of experimentation and reconceptualization of how search results can be presented. However, other search features have disappeared. The web search bill of rights that blekko posted in June 2011 includes several principles that no longer appear to be reflected in the new design:
“Ranking data shall not be kept secret”
“Web data shall be readily available”
“Advanced search shall be accessible”
Neither ranking data nor web data is linked from the search results. The advanced search capabilities of using slashtags can still be used, but they are much less accessible now that searchers need to know what slashtags are available for use. Some opened categories display a pencil icon in the top-left corner. Clicking the pencil will bring up information about the category (slashtag) including the listing and number of sites, a ranking number, and an SEO link that just runs another search.
The SEO data that was available for free, but changed so that it was available only to those who paid, is not linked from any of the standard results. SEO data does appear to be available, for now, at seo.blekko.com, but no new accounts are being accepted. The ability to see a cached copy of a page or links to a site is no longer available. Grep the Web (blekko.com/webgrep) was a novel search approach of requesting a specific advanced search, which is then voted on and scheduled to run. But it has posted no new results since December 2012.
While the new design is more graphical and categorical and features a sideways scroll, many of the strengths of the old blekko are now gone. It may help to attract new searchers, but it also runs the risk of losing those who liked the old features.
The Russian Islands
Yandex has also recently announced a significant change. Yandex runs search engines in Russia, Turkey, Ukraine, Belarus, and Kazakhstan as well as an English-language search engine. The different country site variations have seen a variety of search interface changes over time.
In June, Yandex announced a more unusual addition with a new interactive approach to searching going by the name of Yandex Islands. Other search engines, particularly Google and Bing, are using rich snippet content in search results, and together with Yandex, support schema.org markup. Such results can better structure in-page content such as authorship, medical data, or popular media. Yet Islands aims to go further and help with interactive and transactional searches. It is what Yandex calls a “goal-oriented search platform.”
When fully implemented, Yandex searchers will be able to book appointments, buy tickets, pay bills, and make reservations from the search results page. Yandex (company.yandex.com/press_center/blog/entry.xml?pid=22) describes it as a “search results page, where each search result is a standalone block of interactive information—an Island. These blocks are the first step to the user’s search goal and can be anything from factual information to purchase buttons or order forms.”
While some of the examples look similar to quick answers at Google and Bing, Yandex is taking a different approach in two ways: allowing for greater interactivity and allowing website owners the ability to control more of the appearance. As the blog post puts it, Yandex is offering “an opportunity to choose what information from their site should be featured on Yandex’s search results page. Through our Yandex Webmaster service … owners can markup their pages and see what their Island will look like on Yandex’ search results page.”
While the interactive snippets are based on schema.org markup, Yandex has expanded its own system and made it available to webmasters (github.com/bobuk/islands/blob/master/interactive-answers-eng.md) with a form description language and guidelines for API real-time interaction with Yandex. However, to see the Islands in action, reading Turkish helps, since Yandex’s first rollout of the new interface is limited to Yandex Turkey. The company plans to expand to Russia, Ukraine, Belarus, and Kazakhstan later in 2013, but it did not state whether or when the feature would be included on the English-language version. Maybe libraries could use interactive snippets to create the ability to check out ebooks, pay overdue fines, or request documents, all from within the search results.
In June, Yahoo! launched a “new, modern design” on its main U.S. search engine. Yahoo! still uses Bing’s database of webpages, but it provides its own interface and design for results. The main changes add a common top bar, move the search results up higher and to the left by getting rid of the tabs, and deliver results faster. Related searches moved from the left side to be displayed under the search box. These changes were all based on weeks of testing, but most users may not notice a big difference.
As we all get more used to changing interfaces, new web designs, and updated sites, it becomes easier to tune out the inconsistencies and to accept the new interfaces and pay less attention to minor changes. In these days of rapid prototyping, constant redesign, and perpetual change, it may be unfortunately foolish to hope for much consistency anywhere on the web.