u500k.erinye.com - the AOL data analyzed

Top Child Porn Destinations in the United States

Postulating the following:
  1. people will search for their location or events near their location and
  2. people will search for their hobbies or interests,
it should be possible to correlate searches for locations and searches for specific interests.

I chose child porn as an interest because, taking privacy considerations into account, this group deserves protection least. Also, many of the "scary" user IDs posted all over the blogosphere are related to this topic.

Difficulties

It is not trivial to detect searches for specific interests because language is not exact. In order to create these diagrams, I chose several terms related to child porn from the overwhelming total number of queries that appear to refer to child porn. Terms chosen were common in a sample of the recorded query histories that was examined. Examples of such terms are "nude girls", "preteen sex" and "lolitas".

Degree of Obsession

From the query histories of the users that did search for the keywords I chose enough to qualify, a measure of obsession with child porn can be calculated. This measure of obsession equals the degree of relevancy of their queries given my keywords. The following figure shows the absolute numbers of unique recorded users that are above a given degree of obsession. Note that this means that the number of users above any degree of obsession is also included in the numbers for all lower degrees.

absolute ranking

To account for users searching for child porn related news and for unrelated topics where the same keywords would apply, only users with a certain degree of relevancy in their queries were chosen. This means that users who searched a lot for little girls in the nude would be considered while users who only searched that once would not. In relation to the degree of obsession figure, only users above a degree of obsession of 500 were selected.

Note that it is evident from the previous image that more than 40.000 of the 658.000 recorded users, or about 6 percent, did at least once search for a child porn related topic.

Absolute Figures

These absolute figures are the results of the filtering process mentioned before. This means that the 141 unique users from Florida is really the number of unique users who searched for Florida and for child porn and did that often enough to qualify and used the same keywords I was looking for often enough to qualify.

absolute ranking

Normalized Extrapolated Figures

Florida leading the pack seems to prove common prejudice, but this turns out to be false when the results are normalized. The following diagram is the result of normalizing the number of unique users by dividing by the number of inhabitants of each state. These normalized figures were then used to extrapolate a percentage of inhabitants of each state from the data by multiplying with 427, because the 658.000 AOL users recorded represent about one 427th of the total population of the United States (which is roughly 280 million). This conversion is not to be trusted because it assumes an equal distribution of recorded AOL users in the US. There is no way to tell, but likely this isn't the case.

normalized ranking

Conclusions

With these results in mind, if you are a mom with young children, the best place for you to live is either
  1. the East Coast metropolitan areas, because there seem to be no AOL users that are into child porn or
  2. the other states, because East Coast child porn addicts are too clever to use AOL search for this purpose.
Decide for yourself.