The 2013 report from the Coverity Scan service shows that open source software has significantly fewer defects per thousand lines of code than proprietary software. Coverity’s scan service, which is sponsored by the US Department of Homeland Security, provides free software testing services so developers can look for critical quality and security defects in their C, C++, and Java code. The scan service has been gaining popularity and now supports more than 1,500 projects. 

The 2013 report compares the compares the Defect Density (errors per thousand lines of code) for open source versus proprietary software. According to the report, Coverity tested 741 open source projects, totalling 252 million lines of code, and found a Defect Density of 0.59. The service studied 493 proprietary projects, totalling 684 million lines of code, and found a defect density of 0.72. 

In past years, open source projects of up to a million lines of code had fewer defects than their proprietary counterparts, but projects with more than a million lines did not perform as well as their closed-source equivalents. In 2013, however, open source performed better at all levels – including larger projects. Coverity believes the reduced Defect Density for larger projects results from increased commitment and dedication from large projects such as NetBSD, FreeBSD, LibreOffice, and the Linux kernel. 

The superior performance of open source at all sizes of projects clearly debunks the common FUD myth that concealing the source code improves security. Although FOSS coders might see this as a symbolic victory, it could have a real effect on the way the US government spends money on future software contracts.