Monday, June 30, 2008

Googling Morals

From the New York Times:

In the trial of a pornographic Web site operator, the defense plans to show that residents of Pensacola are more likely to use Google to search for terms like “orgy” than for “apple pie” or “watermelon.” The publicly accessible data is vague in that it does not specify how many people are searching for the terms, just their relative popularity over time. But the defense lawyer, Lawrence Walters, is arguing that the evidence is sufficient to demonstrate that interest in the sexual subjects exceeds that of more mainstream topics — and that by extension, the sexual material distributed by his client is not outside the norm.
By way of background, the defense would take such an approach because under current US law,
[t]he question of what constitutes obscenity relies on a three-part test established in a 1973 decision by the Supreme Court. Essential to the test has been whether the material in question is patently offensive or appeals to a prurient interest in sex — definitions that are based on “contemporary community standards.”
Defining "contemporary community standards" is no easy task, and I would think that hit rates on the googles would be a perfectly valid way to define what's really inside and outside the community's standards since it would produce metrics on what people actually do and prioritize as opposed to what people represent about what they do and prioritize. In principle, it holds the same promise as Nielsen ratings, which measure what people actually watch on tee-vee, whether or not it matches what they say they watch.

Web scold Andrew Keen doesn't like the idea:
If one accepts this argument, if enough people in a community enter "group sex" into Google then this establishes the media distribution of group sex acts as an acceptable communal moral standard. Instead of the wisdom of the crowd determining knowledge, what we have here is the morality of the crowd determining ethical standards. Google's artificial algorithm, then, becomes both the judge and the jury in establishing what is good and what is evil.
If large numbers of people are entering "group sex" into google, it indicates they have a strong interest in the topic. It doesn't necessarily mean they approve of it, but it does indicate curiosity, and we know from the first pages of Genesis where that leads.

Keen's sneering evocation of "Google's algorithm" doesn't suffice to demonstrate that it's a faulty or misleading algorithm. That is, I think, a question worth exploring -- do google's metrics produce a statistically defensible representation of the community's web browsing habits, or do they skew toward a few googlers with very strong interests?

But to Keen's larger point: to whatever extent anyone here is arguing that "the morality of the crowd" is "determining ethical standards," it's the law, not the counsel for the defense, and certainly not google's algorithms. It's the law that suggests a relationship between "community standards" and obscenity; google is just a measurement instrument, and I say a promising one.


Laura said...

First of all, we're talking about Pensacola, which may not be a hotbed of intellectual or moral maturity. But still, if residents of Pensacola are more likely to use the Google to search for terms like “orgy” than for “apple pie” or “watermelon," it could just mean there's a higher percentage of pubescent boys, college students, or very immature people with a lot of time on their hands repeatedly searching. Or it could mean there are a lot of Baptists (read: people whose natural sexual instincts have been so shamed, punished, and denied that they see this as their only outlet). Or it could mean they already know enough about apple pie and watermelon.

So whether "google's metrics produce a statistically defensible representation of the community's web browsing habits, or do they skew toward a few googlers with very strong interests" shouldn't dictate the law, should it? There are still plenty of people with and without the internets who, even if they do find orgies more interesting than apple pie or watermelon (and who doesn't), they wouldn't want porn popping up every time they searched for "organs" or "porgies."

Laura said...

Oops! Hit that "publish" button too soon. Sorry, no offense to you, Pensacola.

Dale said...

Laura, I think there are genuine questions of methodology to address. Are the google results representative of the community's web-browsing habits, or are they biased in favor of the heaviest users? If google results are representative of the community's web habits in this way, are web browsers representative of the wider community?


The law supplies the premise that seems to bother you (and I'm not terribly happy with it either): obscenity is a matter of "community standards." Under such a standard, what's obscene in Pensacola might not be obscene in Portland. For what it's worth, it would seem that very little is obscene in Portland. ;-)

If you accept the premise that "community standards" tell us something important about obscenity -- as the law does -- then it's reasonable to cast about looking for ways to guage community standards. Checking google metrics seems like a promising means of doing so. This checking has to work out all the methodological questions. And as you say, it has to somehow draw conclusions about why people are searching for certain terms.

It sounds like a huge mess to me -- exactly the kind of mess I think is inevitable when government intervenes on question of "good/wholesome thoughts" versus "forbidden/obscene thoughts."

hardiansyah said...

nice info. thanks