Alexandra Friedman is a Senior Designer at Fat Pencil Studio
When I tell people what I do for work, a frequent response I hear is excitement based on people’s enthusiasm for TV shows and podcasts exploring various crime stories. Shows like CSI, Forensic Files, Bones, and Dexter have popularized the idea of forensic science in particular - with varying levels of accuracy in their portrayals. I’ve long been curious about the topic myself and my work on legal cases at Fat Pencil has only launched me deeper down the forensics rabbit-hole.
My recent dive into questions surrounding forensic science was spurred by listening to an episode of the podcast Criminal a few years ago called The Sailor’s Teeth. The episode focused on the rise and fall of the field of forensic odontology (that is, bite mark evidence), which, from the late 1970s to its debunking in the 2010s, led to the conviction and incarceration of dozens if not hundreds of suspects, at least 35 of whom have since been exonerated. The guest telling the story was M. Chris Fabricant, the director of strategic litigation at the Innocence Project, a New York-based non-profit that has been using DNA evidence to overturn hundreds of wrongful convictions since its founding in 1992. Fabricant’s 2022 book Junk Science and the American Criminal Justice System explores not only forensic odontology but a number of other questionable forensic "science” disciplines and how they have infiltrated our legal system.
Forensic science is a very broad category, and includes a huge number of sub-disciplines.
The industry standard for visualizing all the forensic science disciplines, as outlined by the Forensic Science Standards Board.
As part of my research into the state of modern forensic science and how it overlaps with our work at Fat Pencil, I spoke with Janis Puracal, the Executive Director of the Forensic Justice Project. FJP "strives to exonerate the innocent and works to reform the justice system by changing the way forensics are used in criminal cases, regardless of guilt or innocence." During our conversation, Janis stressed the inherent fallibility of forensic science: all science is based on assumptions and testing those assumptions. If the assumptions are wrong, we end up with unreliable results.
The most obvious problems arise with analysis related to pattern matching. This includes things like:
- bite mark comparison (comparing possible tooth marks on victims with the teeth of suspects)
- hair microscopy (using high-powered microscopes to compare hairs from crime scenes to those of suspects)
- comparative analysis of bullet lead/firearms/toolmarks
- shoe print comparisons
- handwriting comparisons
- fingerprint examination
According to a report released by the National Academy of Forensic Science (NAS) in 2009 — Strengthening Forensic Science in the United States: A Path Forward — these forensic methods do not “have the capacity to consistently, and with a high degree of certainty, demonstrate a connection between evidence and a specific individual or source.”
These pattern-matching disciplines become problematic when promising that they can find a 1:1 match. In reality, the best these techniques can do is rule suspects out - they cannot definitively identify who left the evidence behind. This is not to say that pattern-matching tools can’t be useful when used in conjunction with other evidence, but they should not be the sole basis for conviction.
This 2009 report by the National Academy of Forensic Science (NAS) was a landmark publication that launched a national discussion about the necessity of forensic science reform.
DNA testing is now the gold standard for forensic evidence. As mentioned above, it has led to the exoneration of thousands of innocent criminal defendants and helped to solve over 500 cold cases and counting. However, even DNA testing is not infallible. In its early days, DNA analysts could only determine whether or not a suspect’s DNA was present if an obvious sample, like a visible blood or semen stain, was available. As the technology has evolved, scientists have gained the ability to generate DNA profiles from even trace samples. This is a double-edged sword since it allows for higher sensitivity but also a greater likelihood of finding DNA information from multiple individuals. In fact, of the 3,676 exonerations that the National Registry of Exonerations has tracked since 1989, the majority of convictions have been overturned because they were made on the basis of either debunked pattern matching or samples that contained DNA from multiple people. (For more information about the complexities involved in DNA testing, check out this article from the National Institutes of Science and Technology.)
Furthermore, the type of evidence that has historically proved the most convincing for jurors — eyewitness testimony — has often been found to be inaccurate based on scientific studies conducted over the last few decades. Even when test subjects are presented with a sample of similar-looking faces under ideal conditions, misidentification rates are remarkably high. Another important finding is the inexactitude of cross-racial identification. Regardless of their intentions, most people are much less capable of correctly identifying individuals of races other than their own.
What does this all mean?
With all of this scrutiny and scientific discovery, it might seem like forensics as a whole is being called into question — but the truth is, this is all a part of making further progress, as we cast a more critical eye on previous assumptions and seek to be more thorough and thoughtful in our pursuit of due process and justice for everyone. In general, these findings have led to a push for more education for experts, analysts, lawyers, and judges, so that they better understand the limitations of forensic science. As Janis Puracal explains, we need more oversight to prevent analysts from going too far in drawing their conclusions. There is also a growing awareness of Linear Sequential Unmasking, an approach meant to minimize bias and improve forensic decision-making by regulating the flow and order of information such that decisions are based only on the evidence and task-relevant information.
The orange boxes highlight the disciplines FPS deals with most directly.
Fat Pencil Studio staff have developed expertise in two areas of forensic science: Video Analysis and Crime Scene Reconstruction. There are many factors to consider with respect to admissibility of our work, but here are a few of the big ones:
- Enhanced Video. We try to educate our audience about the reliability of using “zoomed-in” images to form a conclusion. It it tempting to enlarge small areas of an image for ease of viewing, and popular TV shows such as CSI promote the myth of being able to magically see tons of detail when zooming deep into a digital image. The real story is that a computer cannot know details that were not captured by a camera. While magnified images (even blurry ones) can be useful for tracking the motion of objects over several frames of a video, they are unreliable for identifying small details such as facial features or clothing patterns. There are many computer algorithms used to “improve” blurry images, but when there is any question about identifying features, it’s best to stick with “nearest neighbor” interpolation, which gives a more accurate depiction of information that was actually captured by the camera sensor.
Comparing two common types of image interpolation (nearest neighbor and bicubic), with an AI tool for increasing the number of pixels. [Image: Wikimedia Commons]
- Scene Reconstruction. When creating digital 3D models of a scene, we are aware that every measurement has a margin of error. For example, distances measured in an aerial photo may have an accuracy of +/- 6 inches, while measurements derived from a laser scan are typically better than +/- 0.5 inches. It’s important to consider a model's accuracy in the context of the questions being posed. If the purpose of a model is to understand the general layout of a scene and how people moved around over time, lower accuracy is needed. For questions surrounding visibility or bullet trajectories, higher accuracy may be needed.
Beyond analysis of video or crime scene evidence, we also collaborate with other experts in areas such as accident reconstruction, biomechanical engineering, police use of force, and firearms/ballistics. In these situations we aim to create demonstrative exhibits to help them explain their findings and opinions. The admissibility of these exhibits is tied to their expert testimony since they are guiding us in how different subjects are depicted in the scene.
Finally, we can use digital 3D models to help eyewitnesses explain what they saw at a crime scene. While it has been scientifically proven that witnesses are unlikely to remember more information with the passage of time, we’ve found that they are often able to share their memories more effectively with the use of visual tools, particularly in cases where there is a language barrier or mental capacity issue.
A useful concept to keep in mind through any discussion about forensic science is the importance of educating a fact-finder (be it a juror or judge) about the assumptions and process so they can understand how an expert reached their conclusions and evaluate their credibility. Visual exhibits are an effective way to bridge the expertise gap and promote decisions based on sound science.