Google's Distribution Advantage Has Its Limits

Andrew Savikas
April 13, 2009 | Permalink | Comments (1) | Listen

Scott Karp has an insightful (and provocatively titled) piece over on the Publishing 2.0 blog about just how deeply Google has inserted itself in Web distribution of content. While much of the piece is about linking, one paragraph in particular is worth calling out for traditional publishers (emphasis added):

If media companies want to compete with Google, they need to look at the source of its power — judging good content, which enables Google to be the most efficient and effective distributor of content. They also need to look at Google’s fundamental limitation — its judgment is dependent on OTHER people expressing their judgment of content in the form of links. Above all, they need to look at sources of content judgment that Google currently can’t access, because they are not yet expressed as links on the web.

"Content judgment" is a neat way to put it, reinforcing that when there's already more than 1 trillion web pages in Google's index value is shifting away from more content toward better filtering and curating of what's already there. (Or as Clay Shirky says, it's not information overload, it's filter failure.) While many publishers fret about customers no longer paying for content, they may miss the boat by not realizing that customers will pay for packaging and convenience (which often means judgment and filtering). For example, at the same time the market for our printed reference books has declined, our Safari online subscription service has steadily grown at a double-digit pace, in part because those subscribers value the implicit filtering of the library.

"Bite-Size Edits" from BookOven

Andrew Savikas
April 2, 2009 | Permalink | Comments (12) | Listen

Hugh McGuire's startup BookOven has opened up an alpha version of a project they're calling the Gutenberg Rally, an attempt to harness collective intelligence Mechanical-Turk style to proofread Project Gutenberg texts for typos and OCR (Optical Character Recognition) errors. In "divide and conquer" style, the system presents just one small snippet of text at a time (with some surrounding context), effectively breaking down a mountain of a task into easily managed molehills:

I had a nice chat with Hugh on Wednesday morning, and what he told me about what's to come from BookOven was quite exciting (though apparently still very much in development).

This isn't the first attempt to harness eyeballs for finding and fixing OCR errors (see ReCaptcha), but reviewing the text in context is a much more satisfying experience, and left me wanting to read more of several of the books I was seeing only in snippet form.