Slashdot Mirror

A county district attorney in Pennsylvania has threatened to file felony child pornography charges against three teenage girls for pictures that they took of themselves, even though the girls' lawyers say the pictures are clearly not sexually explicit and do not meet the legal definition of child porn. The American Civil Liberties Union has countered by asking a federal judge to block District Attorney George Skumanick from filing charges.

Skumanick won't show the pictures to anyone, including the girls' lawyers, but according to the reported descriptions, one picture shows two of the girls flashing the peace sign in their bras, and the other picture shows a girl wrapped in a towel with her breasts exposed after stepping out of the shower. Unless there's something very significant being deliberately left out of those descriptions, it sounds pretty obvious that the pictures do not meet the definition of child pornography, which requires sexual explicitness, not just nudity.

Skumanick may even sound like a buffoon for threatening to prosecute the girls over those pictures, but his overreaching is probably an example of the "context syndrome" that I referred to in writing about a Wikipedia article about a CD showing a naked underage girl on the cover. In that article, I wrote:

Suppose you read a news article about a man who was arrested for possession of child pornography, and you happened to see a sample of the images (never mind how) that he was arrested for. And suppose the Virgin Killer album cover photo had been mixed in with those images. Would it have jumped out at you as an obvious case of over-reaching by the police?

In other words, even an obviously legal photo might seem illegal when it's mixed in with a group of photos that constitute actual child porn. According to the AP, Skumanick's office first found the photos in question after confiscating students' cell phones and rounding up 20 students accused of making or distributing the images found on the phones. Some of those other photos were presumably racy enough to meet the definition of child pornography, and Skumanick probably just lumped in the bra and towel pictures into that category without thinking too much about it. Giving him credit, if someone had come to his office and shown him the picture of the towel girl by itself and asked him to prosecute the girl for creating child pornography, he might have said that it didn't meet the legal definition.

But the "context syndrome" only excuses the initial mistake, and only partly. By now, he's had time to think about those particular pictures, and he knows that non-sexually-explicit photos do not constitute child pornography, so what is he doing? He claims that the girls in their bras were posed "provocatively", but that's not the same as sexual explicitness, and he hasn't even made that claim about the towel picture, so unless there's some bombshell piece of information about the photos that he's still keeping secret (and why would he?), there's no excuse for him not to drop the threats of prosecution right away.

But could even the initial mistake have been avoided? I think it could have, if you designed a scientific procedure for deciding, objectively, whether an image meets the legal definition of "child pornography", by borrowing some of the principles used in police lineups.

Now, obviously one big difference between deciding if the right suspect has been identified in a lineup, and deciding whether an image constitutes child pornography, is that the question of a suspect's identity in a lineup is a question about objective reality, while the question of whether an image is "child pornography" is a matter of opinion and consensus about an imprecisely defined English phrase, so it may sound odd to try and find a "scientifically objective" answer. But by "objective", I mean that the procedure should eliminate the influence of factors that are not relevant to the legal definition of child pornography (for example, if asking someone to decide if they think a picture meets the definition, don't tell them whether the photo was found in a pedophile's basement or in a parent's photo album, because under the strict legal definition, that shouldn't matter). And by "scientific", I mean that the Yes/No answers returned by the procedure should be repeatable as far as possible, so that different defendants aren't being tried under wildly different standards, where Bob is convicted of possessing an innocuous photo while Alice is acquitted even though she possessed a racier one.

A naive solution, from a scientific point of view, would be to poll a random sample of lawyers or other professionals in a police go-to database, and ask them to evaluate whether the picture is child pornography, without any information about where the picture came from. These results would be objective (if the respondents didn't know the source of the picture), and would generally be repeatable, if the sample size is large enough. The problem with this method is that while all defendants would be held to the same standard, all citizens would not be. Suppose the lawyers in the go-to list start to decide, as many of them probably would, that anybody who is being prosecuted for possessing a picture of a topless underage girl is probably a pedophile creep anyway, and would start voting "child pornography" for all but the most obviously legal pictures. The prosecutor would realize this, and would know that they could threaten to ruin people's lives by charging them with possession of child pornography because of pictures found in their possession -- even while other members of society possessed similar pictures without ever being charged.

Here's where the analogy to a police lineup comes in. Police lineups are supposed to include "known innocent" candidates in order to test the credibility of the eyewitness; if the eyewitness selects a candidate who could not have possibly committed the crime (because, for example, they were in jail), then the police know the eyewitness is not reliable. (This was one guideline notoriously violated by District Attorney Mike Nifong in the Duke lacrosse team rape trial; he assembled a lineup consisting only of lacrosse team members from the party, so that whomever the eyewitness identified was guaranteed to fall under a cloud of suspicion.) In the same vein, the lawyers or other experts being consulted by the police could be shown a "lineup" of photos, consisting of several photos that were determined in advance to be legal (either because of a prior court ruling, or perhaps just because the D.A. had declined to prosecute the photos on previous occasions), along with the photo whose legality was in question. Ask the experts to pick which photo they think is closest to the definition of child pornography. Unless most of them pick the photo that's on trial, then that photo can't be said to be worse than any of the other photos that had already been deemed legal.

This is closer to a fair solution, but there's still a big loophole. When police assemble candidates for a lineup, they are supposed to pick candidates who match the general physical description given by the eyewitness. If the eyewitness said they were attacked by a redhead, the police can't fill out the lineup with one redheaded suspect that they want to railroad, and 10 blondes. Because attributes like "Caucasian" and "redhead" are pretty straightforward, if the rules for lineups are being enforced properly, the police don't have a lot of wiggle room to fill out the lineup with candidates who blatantly don't match the description. Unfortunately, it would be a lot easier to cheat when creating a "lineup" of photos to compare against a photo whose owner was on trial for possessing child pornography. If the photo at issue is probably legal but still provocative, then the police could fill out the rest of the lineup with completely non-sexual but perhaps eyebrow-raising photos, like a naked teenage girl watering some houseplants. Then when the police ask, "Which of these does not belong?", everybody would pick the provocative one, and the police would take that as "vindication".

The only way I can think of to guard against this, would be to let the defense counsel pick the other photos in the lineup, and then they could pick the most "provocative" ones that were still legal! For any photos that have been declared legal in the past, the defense ought to be able to argue that if an independent panel of experts doesn't think their client's pictures are any worse than those, then their client should not be prosecuted either. (If the defense lawyer decided their client was a child molester and wanted to throw them to the wolves, they could deliberately pick non-sexual photos for the lineup, so that their client's photo gets pegged as the odd one out -- but when the defense lawyer decides to railroad their own client, it's almost impossible for the system to guard against that anyway. Also, it's probably not a good idea to make this an option for child pornography defendants who decide to represent themselves, so that they can rifle through thousands of photographs of naked children, even legal ones, to find the pictures that they think are the "sexiest" to use for their defense.)

Perhaps someone can think of a better method that is still roughly scientific, in the sense of trying everyone according to the same standard and giving repeatable results. The irony is that despite the potential of child pornography charges to destroy a person's life, it is in possible in principle to try child pornography cases more objectively than almost any other type of crime, because you can separate out the alleged criminal act from everything else about the defendant, and let people examine the evidence of criminality in isolation. If someone shoots a person and claims it was self-defense, it's hard to imagine how you could distill out only the relevant facts of the case, and pass along just those facts to some third-party observer who then renders a judgment without knowing anything else. Half the courtroom battle is over what facts are "relevant" in the first place. But in the case of a child pornography charge, you can give the photo -- and no other information -- to an expert, and ask them to make a judgment.

I know, I know. The police and prosecutors are not actually doing to do this. But that in itself says something. Even if it's not possible to try most crimes in a truly objective fashion, why don't the courts and the police do this when it is possible? Many first-year psychology students that have an intuitive grasp of the principles of sound double-blind testing, could probably come up with a procedure better than the one I've described. When you've spent long enough thinking about how to design experiments objectively, you can't even hear about lawyers arguing over whether a photo constitutes child pornography, without the thought popping into your head: "Have a group of experts look at the photo and rate it, independently of each other. Compare the results to a 'control' result where the experts look at a photo that is not child pornography." And so on. Why don't those suggestions ever come from within the legal profession itself?

And on the flip side, what about using scientific methods to examine facts about the legal system? When considering that judges are tasked with evaluating parties' claims in an objective and fair manner, one could ask: Are they really being objective? What are different ways that we could test this? Perhaps by having two actors in different courtrooms on the same day, charged with exactly the same crime under the same circumstances, except one is black and the other is white, and repeat the experiment many times to see if they receive different average sentences. For a scientist, the idea is the most natural thing in the world. Forget the fact that the legal system doesn't do this -- why is virtually nobody in the legal profession even suggesting it?

Probably because most people who think in terms of objective experimental design are drawn towards the hard sciences, not toward law. That's probably a good thing; such people can likely do more good as physicists and research psychologists than they could as lawyers and policemen. But they can still speak out for the principles of science to be applied wherever possible, in any area where objectivity is important -- especially the law.

All true scientists at heart should keep telling the world that "science" is not just a label that encompasses nerd subjects like biology, physics, and chemistry, with other subjects like art and law being "outside the domain of science". While the statements made within the framework of those subjects are not scientific ("This painting is pretty", "The court finds the defendant not liable", etc.), science can make statements about the people in those professions and the patterns in the conclusions that they reach. If art experts are evaluating paintings differently depending on whether they think the paintings come from an art gallery or a 4-year-old's kitchen table, you could find that out through a scientific experiment. If judges are giving an easier time to lawyers than they are to parties who represent themselves, even when they make exactly identical arguments, you could test that hypothesis with an experiment, too. And scientific principles could be used to draw up procedures for trying cases more objectively, as in the procedure for deciding the legality of sexting photographs. We just need to get over the idea that "scientists" should limit themselves to the forensic CSI stuff and then stay away from the legal arena because that's a "separate domain". Science could tell us quite a lot about how fairly justice is dispensed in the courtroom, and sometimes even how to fix the problems.