Attention and Prevention

Dear Linux Pro Reader,

The Linux Foundation launched the Core Infrastructure Initiative (CII) [1] as a bold stroke in 2014. The foundation, which stands astride the FOSS world and mediates between the realm of business and the hacker culture, started the CII as a reaction to the infamous Heartbleed bug, which shocked the open source faithful and left doubts about the security of FOSS technologies. The original goal of the CII was to "fund and support critical elements of the global information infrastructure," which sounded like a good idea. I didn't have high hopes for them doing much besides giving out money, but money is always good. In the business world, where the Linux Foundation keeps one foot, if you can't make a problem go away by denying it, the next best thing is to pounce on it dramatically and say, "We've got this under control!"

The plan seemed to be to accumulate a big bank roll through donations from corporate partners and then dole out the money judiciously to help core infrastructure projects in need of attention. But which projects? Thousands of Free Software projects exist today. Some are huge corporate operations; others are tiny utilities written years ago by a single developer. Do you chase after the biggest projects with the most lines of code, or do you hunt for hidden weak links, as an intruder would do, searching for that tiny mousehole no one is watching?

Who decides where to put the money? A grant committee? A task force? A team of hired consultants? A funding czar with a personal vision?

To address the problem of assigning its resources, the CII just launched the Census program [2], which offers a set of objective criteria to identify projects that might need help. A point system identifies projects that might need attention by assigning a score based on the predefined parameters, including:

• Website – If the project doesn't have a website, it receives 1 point.
• CVEs – Warnings and announcements from the "Common Vulnerabilities and Exposures" project, which tracks security problems and assigns the familiar "CVE numbers" when a new vulnerability is discovered. 1 point for 1 CVE; 2 points for 2 or 3; 3 points for 4 or more.
• Contributors – If the project has had no contributors over the past 12 months, it gets 5 points. If it has only had 1 to 3 contributors, 4 points; 2 points if it isn't clear how many contributors are working on the project.
• Popularity – An extra point for any project in the 10% most popular Debian packages.
• Network Exposure – If the application operates directly on the Internet, it gets 2 points. If it is used to process data from the Internet, it gets 1 point. If it runs as root, it gets 1 point.
• Application Data – 3 points are taken away if the project isn't really an application but is more like what the census takers call "application data."

The CII reports that interested onlookers have offered some tweaks for the algorithm, including metrics for the number of dependencies, number of patches, and crash statistics. The announcement refers to the Census whitepaper [3] for "a full list of parameters that were considered by this and related projects before this particular set of parameters was chosen for the first implementation of the Census."

The concept of measuring the health of a software project through objective parameters is not new. The whitepaper lists a series of previous efforts, including the Black Duck Open Hub, the FLOSS Metrics project, and other initiatives. And, of course, the list does not identify projects that are the most vulnerable. It simply finds projects that give reason for concern – that would be a higher priority for additional study and effort.

Topping the current list are tcpd, whois, ftp, and netcat-traditional with 11 points each. According to the announcement, these projects were all scored with a CVE count of 0, which means they must have topped out all the other parameters. (The authors point out that no CVEs does not necessarily mean no security problems; it could be that no one filled out the paperwork to file a report.)

The Census project is a work in progress, and the CII is currently seeking input on new parameters or methodologies for evaluating Free Software. The announcement invites users to "sound off on the mailing list or fork your own version from the project's github repository [4] to try out your ideas."

The Census project testifies to the agility and ingenuity of the Free Software community. Rather than doling out funds based on who wrote the best grant or who bought the best dinner for a member of the selection committee, the Census tackles a vast problem with a systematic approach that might not be perfect, but at least offers a starting point for deciding which FOSS projects have the most need for attention.