Buckets and Promises

Dear Linux Magazine Reader,

Security is always big news in IT. The talk today is that the Hollywood Presbyterian Medical Center, in Hollywood, California, has just suffered a crippling ransomware attack. Most of the computers at the hospital are compromised with what appears to be a variant of the CryptoWall ransomware tool.

In case you're new to this topic, ransomware is a fiendishly nasty kind of malware that encrypts all the data on your computer so you can't access it and then charges you a ransom to get it back.

Locking up all the computers in a hospital seems really cold blooded. It means no access to patient records, no information on medications, no test results … along with all the surrounding problems you could possibly imagine. The total ransom required to bring all the systems back online is said to be around $3.6 million. The hospital is relying on the fax machine and old-fashioned telephone calls to muddle through the crisis. Hospital officials have said the effect of the attack on patients will be "limited," which might sound reassuring to some, but I would read it as "not as bad as it could have been but worse than if this hadn't happened."

Since I work for Linux Magazine, you're probably expecting I will use this news to say "You should have been using Linux." Actually, though, Linux and Unix systems, like Mac  OS, are not as immune to such things as people used to think. The Linux ransomware tool Linux.Encoder.1, which is similar to CryptoWall, appeared in the wild in 2015. I seriously doubt that all the computers in this hospital were running Linux (knowing what I know about institutional computer use in the US), but in this case, it doesn't really matter, because I'm not really here to talk about Linux.

The episode at the Hollywood Presbyterian Medical Center highlights the real problem with computer security as we know it and discuss it today. When something like this comes up, all the experts weigh in on the lack of security, but the real problem is the presumption of security. We are invited to consider that there is some clear and attainable standard for how secure a system or a network should be or would be if it were well managed and performing as designed, and if an attack is successful, we're invited to infer that the institution somehow fell short of that standard. But the reality is, no such standard exists. The whole meaning of zero-day vulnerabilities, which seem to pop up almost every day now, is that we don't really know how secure our systems really are.

"The guy who sold me my network sounded so confident. He didn't tell me the system was so porous that someone in a remote location could take hold of the system and extract tribute money from a hospital."

At some level, the people who sell and support computer systems rely on the public's limited understanding of what the product really is. When you buy a front door, you don't expect it to fall off its hinges. When you buy a bucket, you expect it will hold water and not leak. You don't feel you have to ask the clerk at the hardware store "Does this bucket leak?" because you have an implicit conception of a bucket as something that doesn't leak.

When we buy a computer system, we think we're buying something like a rake, or a front door, or a bucket that exhibits simple and logical behavior, but actually, our computer systems leak – a lot! If you want to say Linux leaks less, that's fine, but no system is truly secure. And if anyone had a recipe for how to make the systems less leaky, so we never see another zero-day vulnerability, it already would have happened.

So maybe the best way to promote better security is to quit telling people we're selling them something like a rake or a bucket and just admit "we don't really know what this is, but it does work sometimes, except when it doesn't."

Joe Casad, Editor in Chief