Navigating Software Vulnerabilities: Eighteen Years of Evidence from Medium and Large U.S. Organizations
How prevalent are severe software vulnerabilities, how fast do software users respond to the availability of secure versions, and what determines the variance in the installation distribution? Using the largest dataset ever assembled on user updates, tracking server software updates by over 150,000 medium and large U.S. organizations between 2000 and 2018, this study finds widespread usage of server software with known vulnerabilities, with 57% of organizations using software with severe security vulnerabilities even when secure versions were available. The study estimates several different reduced-form models to examine which organization characteristics correlate with higher vulnerability prevalence and which update characteristics causally explain higher responsiveness to the releases of secure versions. The disclosure of severe vulnerability fixes in software updates does not jolt all organizations into installing them. Factors related to the cost of updating, such as whether the software is hosted on a cloud-based platform and whether the update is an incremental change or a major overhaul, play an important role. Observables cannot easily explain much variation. These findings suggest that there could be high returns to incorporating organizations' relative (in)attentiveness to act on software update releases into the design of cybersecurity policies.