A manual system for paedophile detection
As a technologist there is always a lot of pressure to find a smart automated way to solve technical problems… but sometimes the answers lie in simple manual solutions. For example:
Facebook could easily make a condition of membership that the user supply a landline or cell number as a condition of the users account use.
If suspicious activity is reported against an account a security officer could talk to the suspect user and ask various questions to verify the persons identity.
If the contact telephone number is not answered within a reasonable number of attempts then the account is placed on hold until the account owner is verified. This may mean that the next time the user comes online they would see a message telling them that they cannot send messages until their identity is verified by a security officer.
If the account says that the user is a 16 year old boy, then the security officer can ask the user for the school they go to or the telephone number of their parents to verify their age etc. Various security verification techniques can be used. For example if the user has a New York cell phone number but says they go to school in LA, there is reason to question the persons identity and query further.
If the user fails to supply enough information to pass the security officers questions, the account is tagged as “unverified”.
Another interesting twist that may be useful is the concept of referencing. I have been working on this concept for some time as the basis for a whole social network infrastructure. But in simple terms, as it relates to Facebook, the idea is to tag any inappropriate behaviour to a specific account and then use a weighted system of references to allow the user to get back their verified state. For example Facebook friends could vouch for a fellow Facebook user to verify that they know the person and their age etc. Of course they would become legally bound to have ascertained the truth should an investigation show that the person under investigation was indeed lying about their age and turned out to be a paedophile.
A network wide system of reputation can help facebook security officers determine the credibility of each vouch.
A vouch questionnaire could be like this:
“Dear user, if you would like to help keep Facebook safe, please help us by verifying the identity of <username>”
- Have you met this user in-person? Y/N
- What is the age of <user>? x yrs
- Is this an approximate age or do you know the persons birth date? approx or birthdate.
- How long have you known this person? Xyears
- What is their current cell phone number? <verify>
- Please list two of your fiends on Facebook that may be able to help verify the identity of user? user1, user 2
An interesting side benefit of the referencing approach is that even if paedophile's do get other users to play along with them, over time anyone who vouches for a sexual predator (even by being tricked) will be discovered and dealt with making the online community more aware of their responsibility to protect children.
This article is not meant to be a comprehensive solution outline but more a hint at a couple of directions that haven’t been explored yet.