"I'm just amazed at the amount of failure that goes along here," said Bruce Landsberg, National Transportation and Safety Board's (NTSB) vice president, during a recent hearing about the fatal December 2017 Tacoma Amtrak derailment, according to the Seattle Times.
"We have five or six or seven different organizations that all say safety is their primary responsibility, and yet nobody seems to be responsible," Landsberg observed. "And it just flows all the way throughout the entire operation here, from the very top management down to the lower levels."
Let's change the word safety to security because in today's world, where security is everyone's responsibility, this report offers an opportunity to reflect on the similarities and differences between the highly regulated world of rail travel and the world of Internet technologies.
One crucial difference between cybersecurity and transportation is that there's a widely respected organization, the NTSB, that comes in after accidents and produces a report, and that report establishes facts. Despite many calls for such an organization in the technology world, we still do not have one. There are also important differences between a cyber investigation and a real-world accident involving trains, planes, automobiles, and other vehicles. For example:
None of these apply to cyber incidents, where, in contrast, the relevant systems may be virtual machines long since shut down, the logs aggregated, and the computers involved owned by many different parties, including individuals.
Time for a Cyber Safety Board?
The NTSB has issued a preliminary synopsis of a forthcoming report, and the 10 pages are both thought provoking and easy reading. I read the report because it was a local tragedy, and, like most NTSB reports, it doesn't have very much to do with cybersecurity. But as I read, I noticed a couple of things as I went through it.
First, the cause of the accident is established, as are contributing factors. There are technical, training, and process failures, and many of these are interesting to us in cybersecurity.
Perhaps most interesting are the training findings: "Amtrak did not provide sufficient training on all characteristics of the Charger locomotive," and "Engineers could better master the characteristics of a new locomotive with the use of simulators."
How many of us have gotten "sufficient training" on "all characteristics" of the software we use to get our jobs done? What would that even mean for a systems administrator? How long is sufficient RedHat system administration training? What does it mean to get sufficient training on an Amazon Web Services component, which is subject to change at any time? How many of us have ever used a simulator or range?
We are far more open about breaches than we were even a decade ago, but facts are often thin on the ground. We have a tremendous stream of speculation. We can look over at the transportation sector and see the value of impartial analysis. And that is value to us. It's time for our industry to figure out how we can get an impartial investigator in cybersecurity.