This week, all three of the world’s top general search engines touted the addition of deep links to their search results, although Google has been actively experimenting with deep links since this time last year. The basic premise is this: For Web pages that have named anchors above selected subsections — for example, <A NAME=”Details”> — the search engine is capable of generating subheadings in its search results that link users directly to those subsections, or at least to subsections whose titles imply they may have some bearing upon the query.
The fact that deep links are now official features of Google, Yahoo, and Bing search may not be nearly as relevant today as the fact that all three services made their announcements almost in unison. It’s an indication of an actual race going on in the search engine field, reminiscent of the horse-and-buggy days of the early ’90s when Lycos and AltaVista were vying against Yahoo for search supremacy. This despite the fact that Yahoo is due to be utilizing search results generated by Microsoft’s search engines Real Soon Now.
But is all this just more bluster? Now that there at least appears to be a race once again in online search functionality, are the major players merely vying for position in the consumer conscience, or are they really doing something in the laboratories? Betanews sought out to see whether we could run into any real-world examples of deep linking improving the quality of search results…knowing full well that since Web pages have evolved over time to become shorter, the use of named anchors has fallen out of favor along with Netscape Navigator.
Although we managed to uncover deep-linked pages through Google and Bing just by accident this week, it’s in hunting for them intentionally that deep links seem to scatter like startled squirrels. We started with some purposefully general, broad queries that should stir up old, existing pages.
“Abraham Lincoln”, for instance, turned up zero deeply-linked citations within the first 100 returned by Google. Yahoo, however, provided a deep-linked reference on page 1 to the Abraham Lincoln Bicentennial Commission entry on Wikipedia — one of Yahoo’s partner sites for its drill-down filtering feature. No more deeply-linked entries appeared in Yahoo’s top 100, though. Bing also pulled up zero deeply-linked references in its top 100.
What if we refined the context — would deep links crawl out from under the woodwork? For Google and Bing, no — adding “Stephen Douglas” as a criterion, for instance, did precisely nothing to draw out any deeply-linked results. It’s hard to imagine that, over the long years that pages have been published on the Web, no single page on the Lincoln/Douglas debates of 1858 ever used named anchors. It could, instead, be an indication that the anchors on those pages have yet to be indexed.
But what’s this? Yahoo turned up one deeply-linked reference to the debates as item #1 in its search results, and another reference to Stephen Douglas as #5. Once again, though, both are from Wikipedia; and although Wikipedia’s shallowly-linked reference to the debates was #3 on Google and #2 on Bing, Yahoo’s high placement along with links to history books for sale on Amazon as #2 and #3 (including links to reviews, obviously as a favor to another Yahoo partner, Amazon) indicates that Yahoo’s success on this test may not be by virtue of a highly reformed index, but instead a special arrangement with partners.
If Yahoo were as active in indexing its deep links as the Wikipedia results imply, then it would have attached deep links to other entries that it did turn up, from wiki-based systems such as this item on Stephen Douglas from Conservapedia.com — #12 on Yahoo’s list.
Which brings up an important point: As many recently published Web sites use open source Wiki software, which do generate named anchors for their subsections, there’s really no excuse for search engines that profess to provide deep links to have avoided indexing them. Since we’ve seen occasional deeply-linked examples pop up from out of the blue on Google since September 2008, you would think at least Google would have had time to index such pages before declaring this part of its service officially live.
Next: Can we find any unique deep links on Google or Bing?
We found our first intentionally-sought-after, deeply-linked reference on Google for the query calculus. After Wikipedia, the #2 entry returned by Google was to this page on Calculus.org, a collection of links to multiple other pages and Java applets for teaching the fundamentals of calculus.
Yahoo linked to Calculus.org as #2, and Bing as #1. But neither one provided deep links, which indicates that neither of these sites has gotten around to indexing much else than Wikipedia. Other very general queries we tried — “hedge funds”, “Jerry Goldsmith”, NASCAR — failed to turn up deeply-linked references to any other site but Wikipedia.
The arguments in favor of using named anchors to enable deeply-linked Web pages come from software engineers and SEO experts who believe anything that can help elevate a site’s keywords will increase its page ranking. Indeed, there’s some evidence of this as Calculus.org did place highly in Yahoo and Bing, even though deep links did not appear in their search results.
Even here, though, the advocates of deep linking suggest that the naming of anchors and sub-pages be conservative, and concentrate on a few keywords. One such advocate is Web developer Ravi Shanker, who has built Deeplink.us, a search engine for selected deeply-linked sites only, so far with a smattering of participants.
Opponents to deep linking include bigger Web publishers whose revenue streams rely on readers being able to find their articles via the routes those publishers create for them. They don’t want readers skipping over the top of the front page, which is prime real estate for advertisers — or at least, it used to be.
But neither the supporters or opponents of deep linking may have anything to gain or fear from this week’s developments among the leading search engines. Let’s face it: Folks who really want to see articles of questionable validity written by any number of individuals with unknown reputation, perhaps plucked at random from a street corner, will go to Wikipedia directly — they don’t need Yahoo or Bing or Google to direct them there. So the idea of promoting a new research feature that, in essence, serves as a platform for elevating someone else’s site in particular, may not be all that useful to everyday users after all.