In the latest Webmaster video, Matt Cutts answers a question that could have come from me (but actually came from someone named Leah in New York):
Google changed the search engine market in the '90s by evaluating a website's backlinks instead of just the content ... Updates like Panda and Penguin show a shift in importance towards content. Will backlinks lose their importance?
There have been some arguments – notably from Russ Jones in the comments on this post (extrapolated on by Rand Fishkin in a recent Whiteboard Friday) – that links will actually increase in value over time.
In the below video, Matt Cutts says the opposite is true. It’s not around the corner, but Google will eventually put less emphasis on links – not, however, purely because of link spam, but because most links to a site aren’t the best measure of the value of a single page for a given search.
Let’s walk through what he actually says in response to her question.
Cutts: "Inevitably what we're trying to do [as a search engine] is figure out how an expert user would say, this particular page matched their information needs."
Translation: Google wants to provide the best “answer” to the user’s “question” (whether or not the search query is technically a question). In a perfect world, the algorithm would be able to deliver that answer as well as an expert (or all-knowing god) in that field.
Cutts: "Sometimes backlinks matter for that; it's helpful to find out what the reputation of a site or a page is."
Translation: Historically, Google has depended heavily on reputation and authority when it comes to ranking pages – i.e., the New York Times can rank for whatever it wants. Sometimes backlinks matter in determining reputation, he says. They’re “helpful” – not the be-all, end-all.
Cutts: "But for the most part, people care about the quality of the content on that particular page, the one that they landed on."
Translation: To be a great search engine, Google needs to know what people really want when they want it. There’s a difference between authority and popularity, and for some searches, popularity matters more. And links aren’t necessarily the best way to measure popularity. (The NYT isn’t always going to have the best story.)
Cutts: "So I think over time, backlinks will become a little less important."
Translation: Cutts is admitting that links aren’t necessarily the best way to find the best “answer” for every search question. Or, even if links are the best measure we have now, they’re not the best possible measure. If Google wants to stay #1, they’ll have to change and get better.
Remember when we talked about porn? People don’t link to porn sites; they don’t have “authority” in the Googley sense. So how do they rank porn? And what other verticals might be porn-like, as it were? In other words, in what areas of search would people care less about reputation and authority? Cutts seems to be admitting that reputation isn’t that important to the average user, but the relevance of the content is.
Then Cutts says that Author Rank is helpful for identifying experts in a field, but …
Cutts: "Even if we don't know who actually wrote something, Google is getting better and better at understanding actual language. So one of the big areas we're investing in is trying to figure out how to do [something] more like a Star Trek computer, so conversational search: the sort of search where you can talk to a machine and it will understand you, so you're not just using keywords."
Translation: Google wants to get better at understanding what users want so it can be better at delivering that. His example is that if someone searches “how tall is Justin Bieber,” then immediately afterwards, “when was he born,” Google should know that “he” refers to Justin Bieber.
Sounds great (if not exactly Star Trek level), but unfortunately it seems like this is getting worse before it gets better. Google has gotten really bad at answering long-tail search queries. For example, the other day I googled “ten books that might be poetry,” looking for a particular article by that name. I didn’t use quotes because I wasn’t sure if I had the title exactly right, and I didn’t know if the number was written out or not.
As it turns out, I had the title right, but Google ignored half the words in my query and just returned a bunch of random articles that included the words “ten,” “books,” and “poetry” – the article I was looking for was many pages deep into the index. I had to refine the query to include the author’s name to get it on the first page. (I only remembered the author’s name because it was me.) If we still need links, maybe it’s because Google is still so bad at conversational search.
Cutts: "As we get better at understanding who wrote something and what the real meaning of that content is, inevitably over time, there will be a little less emphasis on links."
Translation: Again, the idea here is that the best way to find the right answer for a given question is to fully understand both the question and the answer. Reputation will always matter, but not as much as the content of the page and whether or not it’s the best answer.
Cutts: "For the next few years, we will continue to use links in order to assess the basic reputation of pages and of sites."
Translation: Google still needs links because they haven’t figured all this stuff out yet. But note that he said “for the next few years” – not the next decade. That means linkless search (or much-less-link-dependent search) isn’t too far off.
Note, also, that there’s an opportunity here for the “next Google.” A smaller, more agile company, one that isn’t so wedded to links, could come along and figure out conversational search before Google, and that would change everything.