Rate the readability of your blog

By Grant Jacobs 20/12/2010

(or anyone else’s blog or website!)

Taking my lead from Colin Schultz’s latest article who uses Google’s new readability score to rate the readability of some science blog sites and networks (but not our’s), here’s how Google ranks Code for life:


Or an averaged readability score of 1.96 (see Schultz’s article), i.e. Readability = [ (1 x Basic score) + (2 x Intermediate) + (3 x Advanced) ] / 100. (Careful readers will note the the individual percent scores are rounded, so dividing by the total of the individual percentage scores, rather than a fixed 100 might be better and leaves me with a score of 2.00.)

Google’s readability ranking rates the text in three levels: basic, intermediate and advanced. Colin’s survey shows many popular science bloggers, like me, have a majority of articles under ‘intermediate’.

To try this on your own blog (or other sites) instructions are on the Google help page. Start with an empty google search form (a brand-new search with no search terms in the URL) and put ‘site:URL’ in the Google search textbox, or search ‘site:URL’ where ‘URL’ is the site whose readability you wish to examine;* next click ‘advanced search’ (to the right of the search box); under ‘reading level’ choose ‘annotate results with reading levels’; finally press ‘advanced search’ (bottom right of page) to get the readability displayed.

Once you have set google to display the readability of the first site, it will continue to present a readability plot for each new search result until you click the blue cross below the right end of the search textbox.

You could use the year (2010) as a search term but in my limited testing it doesn’t make a lot of difference, at least not for blogs with a reasonable volume of output.

Have fun exploring the readability of your favourite blogs. As an example of what a more technical blog looks like, here’s how Mystery Rays from Outer Space fares (readability=2.89):


By contract Bioephemera leans more on the readable side (1.76):


Instead of plotting readability levels, you can limit search results to specific readability levels.

Let me add a digression. Google’s help page on this new readability option suggests ’A scientist searching for the latest findings from the experts may want to limit results to those at advanced reading levels.’ While this might help lift more formal science blog articles rise above the noise, you’d still have to eliminate crank websites. The word-based score will likely also rank crank websites as ‘advanced’ because their efforts to sound technical, and hence creditable. (Having seen silly claims for creditability from some people that follow crank websites, I wouldn’t be entirely surprised if some of these people claim ‘advanced’ readability scores indicate creditability, silly as the idea is.) Here, for example, is what we get for the articles at–as Orac might describe it–’that warehouse of woo’, mercola.com (score=2.44):


I say ‘blog articles’ in the previous paragraph because to locate articles in the peer-reviewed literature, you’d expect you’d be better off with the more established means of searching the literature such as PubMed, Google Scholar or the like. As for science popularisation sites, we’ve already seen that many rate ‘intermediate’. This doesn’t mean that they are less creditable, but that they are targeted at a wider audience.


If you use Mac OS X, you may find it useful to use shortcuts to copy URLs. To copy the URL of a link control-click on the link: a pop-up menu will appear – choose ‘Copy link’.

* A quirk (read: bug) means that when you move to the advanced search from the basic search page, it’ll take the search terms in the web browser’s URL (location field)–not the textbox–into the advanced search. As a result you will want to either start with an empty search page with no search terms in the URL, or start by first searching the URL you want to test (thus moving your search terms into the browser’s location field). In my opinion Google’s form-processing software should be checking if the search testbox is not empty and if so, using these contents. Either way, you’ll have to work around this.

Older articles on Code for life (these from my first month of  blogging):

Scientists can’t write?

Book excerpt – Losing the faces of your wife and children

Autistic children and blood mercury levels

Three kinds of knowledge about science and journalism

Forgetting older science

Monkey business, or is my uncle also my Dad?

0 Responses to “Rate the readability of your blog”

  • Hy, Grant, what about doing a little survey of NZ science blogs for the readability (mine worked out at 2.2. Really must try for shorter, simpler posts). And perhaps comparing them with some of the well known international science blogs. Pharyngula for example rated quite low at 1.46. Mind you one problem could be that the google ranking might include comments on each post. In the case of Pharyngula this might bring more into the basic grade just because they are brief and there are so many of them.

  • The University of Auckland is 63% advanced and 33% intermediate – hope that doesn’t mean prospective students can’t understand it enough to apply!
    I could get quite carried away with this searching lark….

  • Hi Ken,

    You probably have a lot of those fancy words from philosophy skewing your scores 😉

    Thought of doing a survey as I wrote the article—about where I wrote my little aside “but not our’s”!—but my time is just too short until after Xmas.

    Saw the comments to Schultz’s post pointing out that comments are almost certainly included after I wrote my piece. I agree with your thoughts on Pharyngula. Given the sheer number of comments Pharyngula gets it’d be hardly surprising if the comments overwhelmed whatever ranking a Pharyngula blog article itself would get! Among other things he’s got that ‘permanent’ thread thing of his that he rolls over each time it runs up 1000 comments. (I think I’ve only got past 50 once, or thereabouts…)

    “Anglo-Saxon” writing might impact on that too. (Think of the sort of language debates—arguments, too—usually feature…)

    I could try measuring posts with and without comments, which might give an idea of the contribution of the comments, and a means to subtract what their (typical) effect is out. But that’d be even more work! (For sciblogs.co.nz, given that we don’t get a lot of comments, the effect might not be that great.) It might not be a lot harder to write a Perl script to calculate the readability scores in a tailored way that takes comments into account… and, erm, that sounds like more work again…

    Back to the grind…

  • Hi Anna,

    Others have expressed thoughts that scores of 100% on just one level means something isn’t right, and I agree. Statistically it’d be pretty unlikely. Perhaps google can only see your top page? (I assume you’re measuring your own site not the syndicated material at sciblogs.)

    I could get quite carried away with this searching lark….

    Same thing plotting ngrams. I wonder if there are good crime-related ones? Now you’re probably wishing I hadn’t written that… 🙂

  • Hi Anna (me, again…!)

    Usually I have the sense to check before I write… trying out your sciblogs blog you rank 100% intermediate. Maybe you are> statistically unlikely and remarkable consistent!