u500k.erinye.com - the AOL data analyzed

Looking for Registered Sex Offenders

Registered sex offenders are a special case because their data is very public in the US. For example, in a comment on TechCrunch, one person was mentioned who appears to live in Bradenton, Florida.

Why Sex Offenders are Screwed (Pardon the Pun)

In case you are reading this from outside the USA, know that in the US, there is a set of laws usually referred to as Megan's Law, requiring states to collect and make available information about convicted sex offenders. This information may include web pages complete with high resolution portraits, birth date and history. While you can have the honour of being a registered sex offender because you did things as harmless as showing your bare behind in public, you probably won't search the internet for sex crimes then.

The User from Bradenton

The user quoted above, however, apparently searched for terms that make it sound very likely that he is, in fact, a sex offender of the less harmless kind. Among her searches, one can find the following topics: This is his online activity profile (number of searches by hour of the day):

hourly search activity of one person

Since this user almost exclusively searched for one certain topic, it seems very clear that she needs help. Either because she is genuinely interested in child porn, or because she believes she is looking for child porn so she can report it, or because she is a cop on child porn reporting duty who doesn't know how to use the internet. I can see how the blogosphere is going to focus on the first possibility.

The searches of the Bradenton person (I'll call her Brad for now) do not include any personal data other than those related to Bradenton, FL. You can search registered sex offenders in Bradenton online. Entering Bradenton as the city of residence and Manatee county as the county of residence yields a list of 211 registered offenders. All of them have so-called flyers with personal information, including the following:

How to Reduce the Possibilities

There is a number of constraints we can use to reduce the set of possible identities for this person. Keep in mind though that there is no certaincy that Brad actually is among the registered sex offenders of Bradenton, FL. It's just likely. If she is, there are some groups of offenders she is probably not part of:
Young Adult Offenders with Older Victims under 18 Years
There is a number of offenders on the list whose victims are given as minors. This includes the stereotypical 22 year old with a 17 year old girlfriend. It seems very clear that this description does not fit Brad.
Offenders with Adult Victims
Apparently, most of the 211 offenders listed had minor victims, though. This is in itself devastating because Bradenton only has a population of about 50.000, meaning that roughly half a percent of its inhabitants are of the type you would not want in your neighbourhood as a parent. Bradenton covers an area of about 12 square miles. This means that, statistically, if you live in Bradenton, in an area of one square mile around your location, there are 16 convicted sex offenders.
These and some other constraints limit the possibilities to less than, but still close to 200. At this point, it would be fairly easy to just call all of them and ask. Since their data is public, this is something anyone can do and something 200 different volunteers could do in a day. Certainly, one dedicated researcher could do it alone.

While I am not going to further lengths to identify Brad here, because of the extreme likelihood that she is in fact in that database, I consider her identity compromised by the release of the search data. However, the impact is still low, because all of her data was already online in the first place. Locating any convicted sex offenders via the official databases seems to be much easier and yield much more results than looking through the AOL data.

Other Searches Related to Sex Offenders

Out of the roughly 658.000 users whose queries were released, 4665 searched for the term "offender". 1683 of those 4665 users only entered the term exactly once. There are two users who occur more than 140 times with that term, but this includes them hitting the "next page" button. I'll call them 145 and 140. Let's look at 145 first. 145 searched, among others, for these topics: There is no single search pointing to anything incriminating by this person, except a couple of questions about what happens to soldiers that go AWOL. This person is probably not a sex offender. Let's look at 140 next. Well, actually don't. Turns out this 140 is identical to the user ID that has the most queries recorded in the data set. Read more about it here. It is highly unlikely these searches can be connected to a specific person.

Out of the 658.000 users, 3347 searched for the term "sex offender". Note that the user ID with the most searches for this term is also identical to the one mentioned in the previous paragraph. The user ID with the second most searches appears 77 times connected to the term "sex offender" and 419 times in total. This is her profile:

hourly search activity of one person

Looking at her searches reveals:

This person is obviously no sex offender either.

The person with the third most searches appears 74 times and 439 times respectively. This is the profile:

hourly search activity of one person

Among her searches, there again is nothing incriminating. Some searches on sex offenders in Pike County, Alabama and specifically in Troy. Only 14 offenders are registered in Pike County, and it is unlikely that this user is one of them. She also searches for used mobiles in Dothan, Alabama, so it is likely that she lives there. She is also searching for a couple of names and one screenname and a yahoo e-mail address containing that name, also for a real name similar to the screen name. While it is not certain (and in fact not even particularily likely) that this is this user's own screen name and e-mail address, if it is, she can be identified using the phone directory for Dothan, Alabama. Furthermore, one could contact the e-mail address and just ask.

If you do that, remember that there's nothing incriminating about this user's searches. Not even remotely. Not even more personal details. No particular habits. Nothing. So, while the user ID might be identified, nothing new will be learned from that except that the person behind it in fact used AOL during March and May.