February 09, 2006

Searching for Love

in Google book gets this result.

Looking at that page, I find no snippets on the first ten results. All of those are under what used to be called the "publisher" program (it now seems to be the "partner" program).

I also learn that there are not three ways books are displayed, as Google misleadingly claims on the main "partner program" page, but four. That is, sample pages are displayed either with or without login to Google, depending on the agreement with the particular publisher. The latter case means that Google knows exactly what people have been reading, with names attached.

But I can't find one case where Google has no agreement with the publisher and therefore displays only snippets in those first ten results. The same is true for the second set of ten.

This would seem to indicate that the snippet view is not really important to the top search results.

Maybe the reason was the search term. Who cares about love. Let's search for "Supreme Court Sony" instead.

Still the majority of results on the first page is under the "partner program". However, we find two cases of snippet view, both of them Senate Judiciary Committee hearings.

These show how exactly those snippets look. Google is not displaying an electronic snippet here, but a part of the scan, a picture. They do display two lines of electronic data on the results overview page. So to be exact, Google is displaying snippets in two different ways.

We also find a strange case here. The result for "United States Supreme Court Reports" displays absolutely nothing. No snippet of a scan (maybe because Google has the data only in electronic format in their database), and no two lines context in the search overview. Note that the publisher in question is Lexis. I don't know why Lexis gets this preferential treatment compared to all other publishers.

I have been disgressing somewhat. My main point here is that most research results on the top pages are from the "partner program". I have confirmed that theory also with a couple of other searches.

If true, that might be important for assessing the copyright situation. For example, the assertion of Google supporters that you just can't build a meaningful book search engine while respecting copyright would be contradicted by the fact that, yes, most or all of the top search results are licensed.


Posted by Karl-Friedrich Lenz at February 9, 2006 09:11 AM