Wednesday, March 23, 2005

More on GoogleNews: Sources, Bias, Purpose

The concern over content selection at GoogleNews continues on Buzzmachine. I've had various reactions, which can be found throughout the comment threads, and in my earlier post on transparency. Allow me to highlight three more issues:

Sources
This issue first arose when Jeff Jarvis questioned the inclusion of Nazi news sources in GoogleNews, and called for his readers to list other "questionable" news sources in order to "judge GoogleNews' judgment."

Reader submissions included Infoshop (appears to be an Anarchist news site), antiwar.com, The Center for American Progress (described by a commenter simply as "lefty"), The World Socialist Website, Executive Intelligence Review (described by a commenter as "that outright lunatic Lyndon LaRouche's publication"), Jihad Unspun (which appears to be a very anti-American site), and Prensa Latina (Cuba's state-run media).

I'm not here to defend or support the actual content of these various news sites, but I question the point in collecting a list of "questionable" news sources. Jarvis states "I'm not suggesting that there should be an orthodoxy of news or certification of news," but it certainly does sound like he's suggesting exactly that. Calling for transparency is one thing, but gathering lists of "questionable sites scraped by Google" sounds like little else than a desire to create a regime of news certification.

The whole point of GoogleNews is that you have a wide variety of sources. You can read, filter, process, absorb what you want, and ignore the rest. A plurality of voices, perspectives and, yes, even biases is a positive feature of web-based news aggregation. Readers might actually learn something about the world (and themselves!) by reading about how people they don't agree with (including anarchists, socialists, leftys, communists, etc) see an issue.

[To his credit, Jeff later asks for suggestions of what other sources should be included in GoogleNews which are not - a much more valuable endeavor.]

Bias
Many commenters suggested bias at GoogleNews in not only the selection of news sources, but also which articles are listed from a search query. Such things are difficult to prove, but bias in computer systems do exist. (see B. Friedman and H. Nissenbaum, "Bias in Computer Systems" (pdf), ACM Transactions on Information Systems July 1996, 330-347.) Three kinds of biases could exist here:

Technical bias, which would arise from technical constraints, such as only certain types of news sites are available to Google's crawlers, an error in Google's algorithms allowing certain sites to appear to be more relevant, etc.

There also could be preexisting bias, having its roots in existing social institutions, practices and attitudes. Perhaps Google only lists certain news sites and not others because existing cultural norms preference one type organization over another. Perhaps the writers of the code themselves had biases in the type of news they would like to have appear on GoogleNews' pages.

The third bias is emergent bias, which is when a bias emerges only in a context of use. Perhaps only after a sharp increase use of (reliance on?) GoogleNews for providing news information has a bias in news sources emerged.

In all accounts, biased computer systems are instruments of injustice, though admittedly, to varying degrees. In Google's case, bias would certainly violate their unofficial "don't do evil" motto and destroy public trust in the company. For such reasons, freedom from bias must be counted among the set of criteria by which the design of technical systems are evaluated. (For more on this last point, see our Values in Design website)

Purpose
Finally, the actual purpose of GoogleNews is at issue here. How do we (how should we) define GoogleNews: do they "report" news (in the sense of how the NYT or FOXN report), or do they simply provide "access" to what others report (merely a gateway).

Put another way, should GoogleNews be a place one can find for only certain news reporting, or a place where one can find all news reporting. Are they a portal to all that's out there, or a gateway to only the "good" stuff? How we answer these questions is vital to what we can expect (demand?) from GoogleNews.

(Interestingly, this debate mirrors many of the concerns throughout the history of dictionaries and encyclopedias: should a dictionary be a source for each and every word you come come across (including slang, profanity, improper words), or a source for only proper English words)
UPDATE: Jeff Jarvis responds to my discussion of "Source" above, countering my argument for plurality in news sources with the sentiment: "Is it worth knowing what nazis think is news? No, it is not. In no universe." I disagree. And so does this commenter at his site:
Actually, on the premise that exposure to all knowledge is to be desired, I disagree that Nazi news should be excluded. I try to read news from those who disagree with me and those like me, so that I won't cut myself off from those points of view.

If more of us read the hatred and exaggerations of Nazis, as we do read them from al-Qaeda and other hate-filled organizations, we might be more aware of the negatives involved. And act more sensibly in the interests of reason and the reasonable.

Technorati tag: