Google’s New ‘Reader Level’
by Thomas Doane | Content-Manager at SmallBox Web Design
As of last week, Google’s Algorithm has developed a modicum of literary taste. Their new ‘Reading Levels’ feature grades the text of websites to filter search, according to whether the prose contained in the search results is ‘basic,’ ‘intermediate,’ or ‘advanced’.
In order to use the reading filter, click in to advanced search on Google’s Home Page.
After you’re in Advanced Search, open up the drop down about halfway down the page that’s labeled ‘Reading Level,’ which is the first option in the ‘Need more tools?’ column.
You can choose not to display your reading level, you can choose to annotate the results of your search with reading levels, or you can choose to filter your results so that only pages graded ‘advanced’(for example) will appear in your search results.
Google’s project manager, Nundu, explains the development as follows:
“The feature is based primarily on statistical models we built with the help of teachers. We paid teachers to classify pages for different reading levels, and then took their classifications to build a statistical model. With this model, we can compare the words on any webpage with the words in the model to classify reading levels. We also use data from Google Scholar, since most of the articles in Scholar are advanced.”
Google’s grading process is pretty steep, by the way. Only 5% of articles from the New Yorker–the chosen publication of the literati on this side of the Atlantic–were scored as ‘advanced.’ 57% of their articles scored ‘intermediate,’ and 36% scored ‘beginner.’ My own favorite publication (which I thought would score higher than the New Yorker) scored even worse.
What demand is this new development responding to? Google explains the utility of this tool in their official blog. “This…new advanced search feature…categorizes results by reading level. For example, if you’re writing a college paper on [herbivores] you can refine search to see only advanced material, or if you’re a grade school teacher preparing for a class on [herbivores] you can refine to see only basic material.”
For years people have complained that the internet has been lowering the quality of public discourse. There has been some legitimate concern that since the internet drives the media, and the internet has not previously been able to refine search according to the quality or intelligence of discourse, the quality of news coverage has suffered. (This is only one aspect of that whole line of thought. A more important factor is how difficult it is to monetize news consumption online.)
So: up until now, Search Engines have been very literal-minded creatures. They have–if you will–been ‘philistines.’ This means that metrics for attracting audiences, up until this point, have not been able to measure how intelligent the copy on a particular site is. With the introduction of the reading-level tool, the quality of prose on certain types of sites may see a come-back as one of the factors determining readership. For example we see that the quality of prose on CNN’s news site outclasses Fox by a roughly 63% margin.
Fox News, in fact, is only faring slightly better than Disney’s site.
I would imagine that this tool will become more nuanced as time goes by.
Is this likely effect business websites? Well, of course, it depends on the website. If you’re a locksmith or a pizza parlor, this is not going to effect the volume of your seach traffic. If you’re a chemical engineering lab that has a blog about new developments in chemistry, Google Reading Level’s opinion of your prose may have some effect on your traffic.