Reducing Your Book Buying To Statistically Ridiculous Triviality

from the how-many-words-per-dollar-is-that? dept

Earlier this year, Amazon.com got some press for revealing their "text stats" with "statistically improbably phrases," listing out phrases that tend to only appear in that particular book. There were other stats as well -- and all were about equally as useless. It appears that the Washington Post has just discovered these silly stats and has written up an amusing article noting some of the completely useless and trivial stats you can now compare different books over. They really seem to like the "words per dollar" feature, for instance. "But in its pure form, Text Stats is a triumph of trivialization.... Now you too can sound like a literary insider at Washington cocktail parties. You can throw around statistics and make clever conversation about the hard history books, the long-winded novels, even those thick, heavy, make-you-think philosophy tomes that contain really, really long words. And the beauty of it is, with Amazon's "Search Inside" Text Stats and other features, you won't even have to read them."

4 Comments | Leave a Comment..


If you liked this post, you may also be interested in...
 

Reader Comments (rss)

(Flattened / Threaded)

  1.  

    But can it do Type I Nested F-tests?

    identicon
    dorpus, Aug 30th, 2005 @ 2:10pm

    You can have covariates that appear insignificant on their own, but appear significant in the presence of another covariate. Or the opposite can occur. We have condition indexes, variance inflation factors, and type I F-tests to evaluate these phenomena.

    It might be fun to perform a principal components analysis (PCA) on a book, to find eigenvectors of words that describe a typical page. What if every page in a book is merely a linear combination of eigenvectors? It would probably work really well for Techdirt, with its predictable anti-recording-industry postings, free market dogma, and anti-dorpus rants.


    reply to this | link to this | view in thread ]

  2.  

    SIPs and search

    identicon
    stochastix, Aug 31st, 2005 @ 3:00am

    SIPs are useful for certain types of searches (especially technical stuff). A SIP is very rare in the universe of books... so if you can find a small set of books where the phrase occurs several times then there is a strong chance that these books are relevant to the topic of the SIP.

    reply to this | link to this | view in thread ]

  3.  

    how ironic

    identicon
    Anonymous Coward, Aug 31st, 2005 @ 9:10am

    What phrase is so improbable as "statistically improbably"?

    reply to this | link to this | view in thread ]

  4.  

    Re: how ironic

    identicon
    malhombre, Aug 31st, 2005 @ 9:25am

    Or "anti-dorpus rant"

    reply to this | link to this | view in thread ]


Add Your Comment

Have a Techdirt Account? Sign in now. Want one? Register here
Get Techdirt’s Daily Email
Save me a cookie
  • Note: A CRLF will be replaced by a break tag (<br>), all other allowable HTML will remain intact
  • Allowed HTML Tags: <b> <i> <a> <em> <br> <strong> <blockquote> <hr> <tt>


A word from our Sponsors...
Follow Techdirt
Flattr rss rss
From the Techdirt Archive...
A word from our Sponsors...

Close

Email This