Google as Trustworthiness Filter, and What it Means for Librarianship

A few months ago data scientists at Google developed a proposal for ranking search results by the trustworthiness of the sources. This would be an alternative to the current approach, which ranks sources by popularity (ie, the number of other web sites that link to the sources you see in your results feed).

As Google's team put it, "We propose a new approach that relies on endogenous signals, namely, the correctness of factual information provided by the source. A source that has few false facts is considered to be trustworthy...The facts are automatically extracted from each source by information extraction methods commonly used to construct knowledge bases. We propose a way to distinguish errors made in the extraction process from factual errors in the web source per se, by using joint inference in a novel multi-layer probabilistic model." (Bold mine, for reasons to be argued momentarily.)

In a smart Slate commentary, David Weinberger and Dan Gillmor offer a qualified endorsement of this proposal. They note that it is attractive compared to the rudimentary popularity ranking that currently constitutes the Google method. Ranking by trustworthiness has value, as long as Google never becomes the arbiter of what is true. Determining truth claims in various fields should be left to the practitioners of those fields, in the process of scholarship and exploration that has long defined the search for knowledge. In other words: Google should remain a secondary rather than primary source, no matter how sophisticated its search algorithms become. As Weinberger and Gillmor note, "Google is not smarter than the experts in science and other disciplines who are engaged in continuous, well-founded, evidence-based arguments."

That all makes sense, even though we can quibble about how much ground to cede to "experts" in our democratic, Wikipedia-article-producing era. But that's a topic for another post.

For now, back to the bolded section: "The facts are automatically extracted from each source by information extraction methods commonly used to construct knowledge bases. We propose a way to distinguish errors made in the extraction process from factual errors in the web source per se, by using joint inference in a novel multi-layer probabilistic model."

One of the long-prized skills for librarians is the ability to guide people to trustworthy sources. This can happen in multiple ways--either a direct and straightforward referral to a particular source, or (hopefully) via an instructional session that provides people with tools for evaluating the trustworthiness of sources they find on their own.

In either case, the librarian is the filter for trustworthiness.

If the Google team's proposal goes forward, there would be less need--perhaps eventually no need--for librarians to serve as such a filter. Assuming that most information seeking begins with a Web search, Google's routing to reliable sources would serve effectively to get people started. (Many sources will not be available to all readers, unless they are open access—academic librarians will still need to integrate their holdings into Google searches for the benefit of their communities). Of course, the identification of sources is just the start of a research project. People still need to evaluate the credibility of the sources for themselves, and then synthesize their understanding into new knowledge.

In general, these phases of evaluation and synthesis have not been the province of librarians. We have been concerned with selecting, organizing and making available quality sources. If this phase becomes more ubiquitously handled elsewhere, whither librarianship?

I argue for two actions in a “Google trustworthy sources” era: concentrate on helping people synthesize and evaluate the content they locate, moving into a more pedagogical vein; and intensify our focus on collecting, curating and preserving the unique content of our own institutions. The first action would be a worthwhile stretch, the second would allow us to apply familiar and unique skills. Both approaches would demonstrate the continued vitality of librarians in the digital age.