On Monday, I gave a presentation about incarceration and data. It was, by far, the worst presentation I’ve given, and it upset and offended audience members. I’m truly sorry, and this is an attempt to outline the general feedback that I received and lessons that I’ve learned.

I am a white male from Flint, Michigan. This is important to point out because my talk (and its blind spots) came from this vantage point. While I have worked hard to learn about the systemic injustices that make my perspective privileged, I will never experience racism or sexism the way others do. The problems in my presentation were a product of this privilege.

Mass incarceration has long been a blight in the United States and it has been especially harmful to communities of color. My intent with this project is to use data to elucidate new ways of looking at an intractable problem. My hope is to pull people into the topic who are otherwise indifferent or paralyzed by its scale. The work-in-progress that I presented fell short of that goal. I will do my best to learn from this feedback and build a piece of work that leads to positive impact.

System Bias

The goal of my projects is to affect the discourse on complex topics. I want to believe that using data brings a level of objectivity beyond individual anecdotes, a force that makes people who are otherwise removed from a topic lean in and care.

Data is important, but I was wrong in viewing objectivity and subjectivity as mutually exclusive, especially as it relates to social issues. In the talk, I shared statistics that could have benefitted from historical and social context, such as longer terms served and disproportionate arrest rates faced by people of color. I intended to shed light on injustice with data, but what manifested was a conflation of the bias that people of color face as it relates to incarceration without suitable context and nuance. Data can be misleading, and I have an obligation to ensure all data has the appropriate context. The next version needs an appropriate level of analysis so that readers know how to interpret the data. This has to come from experts who understand the context beyond what I can research on my own. Dropping data and letting the reader play detective can cause more harm than good.

Personal information

Federal and state governments release personal information about prisoners. The most public version of this is Megan’s Law for sex offenders, but some states go as far as publishing photos, names, and incarceration histories. This is, at the end of the day, for the benefit of people who haven’t been convicted of crimes, and it does nothing to decrease recidivism. Perfect data on a national level for the incarceration system is nonexistent. I saw these state datasets as a way to visualize recidivism, sentence length, and other issues about the prison system that are hidden from society. I also hoped that using it might show the human impact of a system that is often inhumane. While this data is public, we have a duty to protect the privacy of people who very well might be innocent, forced into a plea, or just trying to live their lives. Celebrating or building tools off of this data is unethical. In the talk, I treated this topic far too lightly by exhibiting prisoner data sourced from states that follow this practice and presenting example records. There are many ethical imperatives here, and it requires greater due diligence for any project that involves personal information. The next version will absolutely obscure personally identifying information. If there’s a need to humanize “dots on a screen,” I’ll rigorously vet the final product with people who’ve accomplished this appropriately in the past.

Tone

Overall, many parts of the presentation were tone deaf. I’ve been living in this data for 5 months, and have lost sight of how sensitive it can be for a fresh audience. The talk should have had a far more serious tone, and audience members were understandably taken aback by this insensitivity. For future versions, I need to talk to people who have been victims of this unjust system, revealing blind spots that I’ll otherwise miss.


After the election, an internal monologue for me (and I think others?) was, “I wish I had data viz’d the shit out of this rather than making that, in retrospect, other trivial thing.” The feeling was regret…that voters read the prediction models and completely misunderstood their complexity. It felt like a visualization problem — one that some one like me could try to work on. I want to lean into more meaningful topics, but I’m learning that also comes with far greater responsibility.

I want to express gratitude to the audience members who didn’t have any obligation to give me feedback, but went out of their way to do it (which, upon reflection, is next-level patience and maturity). Topics as important as criminal justice reform are new territory for me, and I have a lot to learn. I’m taking with me a far greater sense of my responsibility around due diligence, feedback, and meticulous consulting. And again, to those I have offended or hurt, I’m hoping I can earn back your trust with the next version.