aka: Google Learns to Read Like a Ten-Year-Old
Note: I do not have concrete evidence nor have I tested the following.
This is simply a theory I formulated by vaguely (and quite likely inaccurately) recalling my studies in linguistics as a graduate student, my recent reading of some of Google Director of Engineering, Futurist and linguist Ray Kurzweil’s work, a recent comment made by Matt Cutts during his keynote address at PubCon in Las Vegas, my career in search marketing and a weird dream I had where Larry and Surgey gave me all of Google’s secrets carved into clay tablets, asked me to quickly memorize them then bury half in the Sahara Desert and send the other half to outer space in a time capsule.
What is Google Hummingbird?
Google Hummingbird is Google’s newest algorithm. It differs from other recent algorithm updates such as the infamous “Panda” and “Penguin” updates in that it is a completely new algorithm whereas the previously mentioned were changes made to the existing algorithm.
As I understand it Hummingbird is meant to allow Google to more accurately decipher a searcher’s intent for the purpose of returning more relevant results. It does this by using an algorithm equipped with natural language processing to better understand and respond to search queries that are more conversational in nature and more representative of natural language use.
According to Google representative Amit Singhal:
“With more complex queries, the algorithm can better understand concepts vs. words as well as relationships between concepts.”
The “complex queries” referenced above are a reference to the spoken queries that one would use when using Google’s voice search. The idea being that the type of queries one will likely use when conducting voice search differs fundamentally from those that one may type on a keyboard. The go-to example to illustrate this seems to be the pizza restaurant scenario. This scenario suggests that an individual looking for a place to get pizza may ask “where is the best place to get pizza in Chicago,” for example, whereas if they were to type this query into a search engine it would likely be different- Pizza Restaurant Chicago or something similar.
Matt Cutts on Google’s Reading Level
In Matt Cutts’ recent keynote address at Pubcon in Las Vegas, he stated that “Google can now read at the level of a ten year old.” To me this is amongst the most important concept to note when considering any changes hummingbird may levy on search and search marketing.
So how does a ten year old read?
Context clues, right?
I remember trying to read books way beyond my reading level at ten years of age but as long as I was able to isolate and decode the meaning of at least a few of the main words I could usually figure the rest out. Heck- come to think of it, I still read like this half the time!
So when a question such as ”where is the best place to get pizza in Chicago” is asked via Google’s voice search meaning is derived by using context clues. Words such as “pizza” and “Chicago”, our main keywords, are also the words that carry the most meaning. Certain question words such as who, what, where, why etc. further qualify searcher intent by adding an additional layer of meaning to the query, for example:
So by learning to distinguish between filler words, meaning making words and question words, a search algorithm can derive meaning from and endless number of combinations. So in short, Google Hummingbird is what it looks like when a machine learns to read, and apparently this is similar to how a ten-year-old learns to read.
Actionable Points for SEOs
When you think about it in the above context it becomes quite simple and allows us to formulate some actionable points. Most importantly? Don’t forget that Google is at the reading level of a ten-year-old. I think we would be under-minding Google’s drive, however, if we said: “okay, that means we have to write all of our content at a 10 year old reading level now, and we will definitely rank number one.” I would say this is not the case, far from it, and while Google may only “read” at the level of second grader, this is still only one part of the equation, and as the ten-year-old me mentioned above did everything he could to pull meaning out of those big, dusty novels by using context clues, so does Google. Of course we are talking about the usual suspects:
This is more important than ever. With a speech based algorithm reading at a second grade level we need to give Google all the context clues we can. As always, best practices come into play here. Make sure your title tags are consistent and accurate in their description of the page.
I have always known linking out to authority, relevant sites is a must, but I have never thought much about the anchor text of these outbound links. Lately though I have began seeing pages rank for anchor text terms of outbound links to authority sites even if the ranking domain, let alone URL, and zero inbound links with this same anchor text.
My personal favorite. Back links are not going anywhere and remain and will always remain one of the greatest indicators of the quality of a site or page as well as what the page is about.
I think the idea of a link as “a vote” is ready to get tossed, however, and in its place we should put the idea of a link as a pathway to allow the flow of information. I don’t know if that is the best way to describe it but I think it is a start.
For example, in academia there are generally main works that serve as the origins of each discipline. Each subsequent generation of scholars derives from this main work and in turn references it, the next generation does the same but also references the newer works, and on and on. For example, Jung referenced, added to, agreed with and disputed the works of Freud and eventually many more scholars did the same. The end product is a body of knowledge in the form of written works, all relevant to one another, and all “linked” to one another.
I think this is how we have to think of linking. We have to break down the “walls” of our individual websites and join the conversation. In this context linking is easy. Forum comments make sense if you are commenting on relevant sites and adding value to the conversation. Guest posting becomes valuable if you are doing it in a way that adds value and is relevant to the hosting site. If you happen to be writing about something and can reference the information on your own site why wouldn’t I?