Although adding the OceanGate five to the tally would be premature, the recent interrogation got me thinking — are there instances where actual loss of life has occurred due to Excel or other spreadsheet errors? For those unfamiliar with the situation, OceanGate’s submersible Titan imploded in June of 2023. Last week, the U.S. Coast Guard held hearings to investigate what went wrong — and it appears that at least part of the issues were related to spreadsheets. During the hearings, Antonella Wilby, a former contractor for OceanGate, testified that their process for navigating to the shipwreck “was to write down the latitude and longitude coordinates, then type them into an Excel spreadsheet, then import the spreadsheet into a map.” Wilby testified that when she brought up her concerns, she was told she “wasn’t being solutions oriented.” Yikes.
The financial cost of a “simple” miscalculation or typo has been well documented since Excel’s first release in 1987. It is not a modern problem and totals in the trillions of dollars. In fact, way back in 2000, a wildly cited paper by Dr. Raymond R. Panko for the Spreadsheet Risk Symposium reported that an estimated 91% of business spreadsheets contained errors based on audits. Amazingly, he went on to report that 51% of spreadsheets in lab studies contained errors — despite most experiments being conducted with only 25 to 50 cells. In my experience, Dr. Panko accurately describes the cause as developers being overconfident and policies being lax. Technochauvism at its finest. However, I would add that developers are overworked as a secondary cause due to management neglecting to prioritize and provide adequate resources for data quality initiatives and safeguards for the development process.
Horror stories are so common, the European Spreadsheet Risk Interest Group maintains a growing list tracing back 15 years. Although some errors likely caused physical harm, the link isn’t direct or usually clear. For example, in 2023 a spreadsheet error delayed training for needed anesthesiologists, potentially delaying necessary medical procedures in the future. Or this disastrous exposure of police force officers and staff, which equates to a gold mine for terrorists and poses a real safety risk to the named individuals. In fact, as of May 2024, individuals listed reportedly still navigate a genuine fear of threat to their lives.
In 2021, a spreadsheet error resulted in falsely reporting that 8% of people in Swindon, U.K., were hospitalized with heart issues. This information was wildly circulated to discourage COVID-19 vaccination around the world. Later this data was determined incorrect, and updated to 0.01%. In the United States, misinformation, including misinformation based on this spreadsheet error, decreased vaccination intent by 6.4%. In the U.K., misinformation resulted in a 6.2% decreased vaccination when vaccination has been shown to reduce the risk of death from the disease. It is unlikely we will ever know the full impact of this misreporting and the subsequent death toll.
Another spreadsheet error related to COVID-19 resulted in missing data for contact tracing. This error which was caused from running out of rows in a spreadsheet is the only example I could find with a study attributing the loss of life directly to spreadsheet error. Fetzer and Graeber conservatively estimated the death toll attributable to the spreadsheet error to be over 1,500 people. This is a stark reminder of the cost of poor data management and process. So, to the Antonella Wilbys of the world, keep advocating for better solutions. You never know, you might save someone’s life.
