When Security Gets in the WayColumn written for Interactions. © CACM. This is the author's version of the work. It is posted here by permission of ACM for your personal use. It may be redistributed for non-commercial use only, provided this paragraph is included. The definitive version is published in Interactions, volume 16, issue 6: Norman, D. A. (2010). When security gets in the way.
The situation with security is similar to the situation once faced by the human-centered design community. In the early days of software development, programmers and engineers devised the systems, sometimes acceding to feature lists from the marketing community. After they had finished, they would ask the usability and the technical writing communities to make it usable and understandable (and the designers to make it pretty). It is only when practices change so that all of these groups can work together as team members from the start of the project that improvements result. So too with security and privacy, except in this case, it is the security and privacy processionals who are the outcasts. It is time to make them first-class citizens working with the product team throughout the entire development cycle to produce cohesive systems that are understandable and usable, functional and safe, secure and private.
If this endeavor is to be successful, we need more understanding of the issues, better toolkits to deliver to developers, and a comprehensive set of tools, scripts, and templates to provide the administrative support staffs around the world so that the rules and polices they develop will be consistent both with one another and with best practices of the security and privacy community. Today we lack a deep understanding of critical things, including people's conceptual models of security and privacy, what new models might be effective, and how to present the conceptual models so that they are both effective and also non-intrusive. The programming community needs better tools to defeat attacks. More research needs to be done and the purpose of the meeting at the National Academies was to develop a research plan.
Usability or Security: does it really have to be a choice? At times the two seem immutably bound. Make it more secure, goes the belief, and as night follows day, things become harder to use. It is a never-ending challenge, with security experts pitting themselves against usability experts, and both of these fighting with the engineers and marketing representatives, all convinced that their view is most important, each convinced that attention to the others defeats their goal.
Does added security make things more difficult to use? Will people always resent the extra steps? The answer to both questions is the same: Not necessarily. Consider the physical world of doors and locks. Locks on houses, cars, and private records get in the way of easy access, but are tolerated because they seem necessary and the amount of effort required to open them usually seems reasonable. Note the two different components: the understanding of the necessity for protection and the reasonableness of the effort required. Both are design issues. And both require at their base, a coherent, understandable conceptual model of both the need for security or privacy and the workings of the mechanisms that enforce them.
Note that the terms "security" and "privacy" themselves oversimplify a complex set of issues. Thus, we often confuse the problem of identification (who is this person?) with that of authorization (is the person or system using the system authorized to do so in the way requested?). In many cases, authorization can be established without identifying people or systems, just as identifying the person or system need not establish authorization.
In the real world, identification and authorization are confounded, requiring needless identification checks which impair privacy, yet overlooking the important issue of authorization. Medical systems sometimes over-enforce privacy concerns, in part because of over-interpretation of government-regulated privacy policies, impairing the very services they are designed to support.
Note that different groups are involved, each requiring a different form of design assistance. System developers provide the underlying mechanisms, but the Information Technology (IT) administrators at the various sites determine just how those policies shall be enforced. The IT staff is under considerable pressure from their own administration to reduce security and privacy concerns, but to do so, they must be well-versed in technology, in the law, in the needs of the user community, and in the psychology of both the legitimate and illegitimate users. What the community needs is a set of standardized scripts, templates, and system tools that allows them to implement best practices in ways that are both effective and efficient, standardizing interactions across systems in order to simplify the life of users, but still tailoring the requirements to any special needs of the organization. These tools do not exist today.
In the absence of standard guidelines and adequate tools, different systems implement the same policies with very different philosophies and requirements, complicating life for users who must use multiple systems. Developers who lack an understanding of real human behavior tend to impose logical rules and requirements upon a bewildered, overwhelmed audience. The audience, either not understanding the rationale or simply disagreeing with the necessity for the procedures imposed upon them, see these as impediments to accomplishing their jobs. Moreover, the systems people may lack understanding of the clever rouses and social engineering skills of the illegitimate users, who break into systems the easy way: by lying, stealing, and deceiving. The strongest locks in the world do not deter the clever social engineer.
Only amateurs attack machines; professionals target people. Bruce Schneier
Does more security necessarily translate into less usability? No, a more secure lock is no more difficult to operate than a cheap, insecure one. With appropriate technology we can make some of these systems even easier to use while simultaneously enhancing security. Thus, biometric identifiers and Radio Frequency Identification tags (RFID) have the potential to make identification easier while simultaneously offering more security.
Want a classic example of a failure? Passwords. There are several myths in the world about security, but the most pervasive of these has to do with password security. Look at Northwestern University's password requirements: an over-reaction to the problem of password discovery through brute-force attacks. Breaches can occur, especially when people use simple, common passwords or the names of their family or pets. There are easy to find lists of common passwords and personal information is readily obtained, sometimes supplied by the victims themselves on the social sites they belong to. Systems do need to be engineered to prevent rapid, prolonged attempts at finding passwords, but there is no need to go to onerous requirements, just long enough, non-word combinations that sufficiently slow up the guessers. Lacking sophisticated understanding of the nature of password theft, system administrators often take the extreme measures exemplified by Northwestern's IT department, even though security professionals continuously point out the harm such schemes produce. This is another example of where usability experts can collaborate with security professionals to develop secure, yet humane solutions.
Although there is much emphasis on passwords security most break-ins occur through other means. How do thieves break into systems? They usually don't use brute force. They phish, luring unsuspecting but helpful people to tell them their login name and password. Or they install sniffers on keyboards and record everything that was typed. The strength of passwords is irrelevant if the thief has already discovered it.
Passwords illustrate several of the difficulties with current methods of enforcing security. Passwords are the least expensive mechanism known for securing systems. But onerous password requirements reduce security and increase costs. They reduce security because many people either use trivial algorithms to generate memorable passwords or write down their passwords and store them in easily discoverable places. They increase costs because help desks must be staffed to handle the large number of users who have forgotten their passwords and can no longer log in. These requirements keep out the good guys without deterring the bad guys.
Another problem is scale. Most of us have tens or even hundreds of passwords. If a single password exceeds the limits of normal memory skills, what about hundreds of them, each with different kinds of requirements?
What do security professionals do? I asked attendees at the security conferences. Many of the security experts said they do "what everybody does: have two passwords." They have a simple, easy one for everyday use (all those internet sites that require logins for no apparent reason) and a complex one for places they care about. "Only two?" I asked, surprised. "I can't get by with two, many places have obscure conflicting requirements about length, or how to start or end, and requirements for - or prohibitions against - weird characters. Or they make me change my password many times a year, never permitting use of old ones." "Oh, yeah," they admitted, "we write those down." Is this a wise policy? No, but how else can one cope?
Bruce Schneier, security professional, has it right: passwords are not where the action is among thieves and spies. It is like the front door of your home. Professional thieves don't even bother. They know the front door is solid, with expensive locks and bolts, but the rear and side doors are often left unlocked or equipped with cheap locks. While we carefully lock the doors, they get in through the windows, or by finding the key hidden above the threshold or under the potted plant. The easiest way to break into a security system is to lie and spy, or to use their technical terms, "social engineering," or "key logging" or "sniffing" and "shoulder-surfing." Send out phishing attacks, install key loggers, peek over people's shoulders at public terminals (or install a video camera to do it for you).
Think about it: the big losses occur when someone accidentally emails important information, or loses an unencrypted hard drive or personal computer. One notable example of the "good news -- bad news" category illustrates the problem. The bad news is that someone lost a hard drive with confidential information. The good news is that it was encrypted. The bad news is that because the password was so difficult to remember, it was taped to the drive in plain sight. Here is a case where usability should not have triumphed.
Both security and privacy are difficult problems. We need systems that are easy to use for their intended purposes or by the intended people, but difficult for non-authorized people or uses. For these purposes we need components not normally considered in simple product design: means of authenticating identities or authority, needs and permissions. Some of this will require physical tokens, biometric identifiers, and privately known information. Some of this requires rules and policies, sometimes editable by the user of the system, sometimes only editable by authorized administrators, sometimes buried in the code and unchangeable without significant development costs.
Finally, we need more humane ways of interacting with the systems to establish eligibility, ways that the people who use them regard as appropriate and reasonable. This means that the systems must be accompanied by a clear and understandable conceptual model. If people could understand why they were required to do these things, they would be more willing to pay a reasonable penalty. We all willingly adapt to the inconvenience of locks that seem reasonable for protection, but not those that get in the way, as the propped-open door at the security conference indicates.
We have a wonderful design challenge before us. It is time to make systems that are more secure, that enhance privacy, and that are still eminently usable. We need systems that are effective at performing their tasks, while providing high quality of user experience at reasonable cost. The solution is going to require sensible analyses, the development of appropriate technologies probably including automation, enhanced interaction protocols and interfaces with better feedback, and the development and continual communication to support the development of an appropriate conceptual models. The only way this will happen is if all parties work together as a team from the start. With notable exceptions, the security and privacy concerns have been addressed by the security and privacy experts, coupled with the arbitrary rules and policies of system administrators, where these concerns have been tacked on to existing systems as afterthoughts.
Usable security and privacy: It's a matter of design.
this posting has received a reasonable amount of commentary. Richard Lewis, of the PA Consulting Group in London sent a thoughtful email abut the problems faced by the security administrators. With his permission, I post his email here.
"One factor you don't touch on - and it's a sensitive one for the security community - is the motivation of the security officer responsible for delivering security to a company.
"Security officers are in a difficult position - especially those responsible for delivering electronic (as opposed to physical) security where consequential damage of any breach can be huge. Unfortunately, total responsibility for security is commonly delegated to the security officer and when things go wrong it's his neck on the block.
"So the first line of thinking for a security officer in this position is going to be "How do I protect my job and make sure that I'm being seen to deliver security" rather than one of delivering the best possible security - which might appear to others to be "weak". This drives sub-optimal solutions: extreme lockdowns, severe penalties for breaking rules, etc.
"For me, policy - the backstay for many security professionals - is very weak as a security measure unless people actually buy into it as a group. Draconian and unreasonable rules are ignored, especially as humans are more worried about the risk of getting caught than the potential penalty. But for the security officer, policy is a "get out of jail free" card. He/she can simply say "Yes, but so-and-so broke corporate rules, so it's not my fault."
"We've seen this defence recently in other areas here in the UK eg, social work where there have been high-profile child deaths which could have been prevented by better social worker action. In some instances the excuse was "They didn't follow procedure" as if "procedure" absolutely 100% guarantees success - which it clearly can't.
"So the social issues of security don't stop at finding workable solutions. They have to include a true understanding of delivering security which penetrates upwards into an organisation right through to board level. The board needs to recognise that the benefits of doing electronic business (for example) entail risks which can't be entirely mitigated and that security events will occur. And, of course, security events occur all the time, just that the only ones that get reported are when something goes wrong!
"Security needs to become less about blame and more about how best to deliver it and how to handle incidents when they occur."
Useful Reading Matter (Sidebar)
Adams, A., & Sasse, M. A. (1999). Users are not the enemy. Communications of the ACM, 42(12), 40-46.
Anderson, R. J. (2007). Security engineering: A guide to building dependable distributed systems. New York: Wiley.
Florêncio, D., Herley, C., & Coskun, B. (2007). Do strong web passwords accomplish anything? Paper presented at the Proceedings of the 2nd USENIX workshop on Hot topics in security, Boston, MA. http://www.usenix.org/event/hotsec07/tech/full_papers/florencio/florencio.pdf and also http://research.microsoft.com/pubs/74162/hotsec07.pdf
Schneier, B. (2000). Secrets and lies: digital security in a networked world. New York: Wiley.
Don Norman wears many hats, including co-founder of the Nielsen Norman group, Professor at Northwestern University, and author, his latest book being The Design of Future Things. He lives at jnd.org.
- All Books
- The Design of Everyday Things, Revised and Expanded Edition
- Living with complexity
- The Design of Future Things
- Emotional Design: Why we love (or hate) everyday things
- The invisible computer
- Things That Make us Smart: Defending Human Attributes in the Age of the Machine
- Turn Signals Are the Facial Expressions of Automobiles
- The Design of Everyday Things