An article from BBC News discussing a new paedophile (sic) victim database provides some details into the new database that will contain the faces of children who have shown up in pornographic images on the internet.

All told, the database (called ChildBase, but apparently not associated with any of the other things called ChildBase) contains 220,000 images and digitizations that are designed to perform matching functions on input images.

Thus, once a picture is received, it too can be digitized and the face can be compared against all of the faces in the database for a match to determine if this is a new victim.

What wasn't discussed much in the article was the sensitive nature of this data. Although the report tells that the database will be strictly limited to law enforcement personnel and international police agencies, it does bring into question whether it is a good idea to provide a one-stop shop for a list of people who have been abused.

Since the UK already likes national databases of personal information, it would almost be more interesting to have a complete database of all children who are in the UK (take them from their school photos as an example) and have that used for the purpose of matching faces. If you are willing to assume that the data won't be used incorrectly, you can make an identification of the face without knowing if the child has been involved as a victim in a previous incident.

This idea, of course, is also replete with potential abuse, and in a country that has been known to employ cameras and face identification software in order to recognize perpetrators, it certainly raises some interesting questions.

However, a large database, in this case, is nicer than a small one. If you have access to five million faces and are trying to match one in a photograph for purposes of pursuing a paedophile, you would need to perform a maximum of 5 million comparisons (hopefully a lot fewer if they are using a reasonable algorithm). But that same 5 million person database cannot reasonably be used to search train platforms if you have to enumerate the entire database. This could be done by keeping the victim status of the victim out of the main database and only making it available to a more limited group of people.

In either case, it looks like technology's double-edged sword is at it again, and the good of being able to identify victims quickly is outweighing the potential bad of a "victims roster" in the UK.

I just hope they don't have some kind leak on that data, as that could prove horrifying for the children who are in the database.