In my last blog, I broke down security pain points into four categories. Today I want to take a closer look at identity. Most people will tell you things are pretty bad today, but things have improved quite a lot since nobles dispatched their missives authenticated with rings impressed into wax. One needed a ring, a unique seal, and a load of wax. Not to mention recipients sufficiently well-trained to distinguish your seal from fakes. So today, with our richer toolkits - instantaneous communications, unlimited computing power in the cloud, and even ginormous touch screens in every purse and jacket pocket. So what the heck is going on? With amazing technology advances of the last decades, how does identity remain a problem?
Scale might be the most important complication. Today about 6 billion devices are internet connected, and as the so-called internet of things scales, some project 50 billion devices to connect by 2020. Even if we wanted to issue all the human users unique rings and hot wax, we’d still be left with the challenge of identifying the remaining billions of devices which lack the prehensile abilities required to hold and impress ring into wax. Today large public suppliers authenticate more users in seconds than the royal seal examiners did in lifetimes. And the risks have scaled accordingly. In a world that runs on automated authentication decisions, adversaries have become better at exploiting gaps and weaknesses. Estimates place the cost of cybercrime to firms globally as high as $1 trillion. So if your users complain you’re being paranoid, let them know you have just cause.
So let’s look at it from our users’ perspective for a moment. They’re continually consumed trying to remember usernames and password for a multitude of accounts – both business and personal. Names of first grade teachers, first pets, first cars. Text messages with one time passwords. And still we lack confidence and suffer risk of fraud as sophisticated malware can hijack the most closely guarded user-provided credential. And then we complicate things with byzantine rules and guidelines for creating and changing passwords. It all would be comical – if our guidelines weren’t making it impossible for users to get their work done! So now we’re all miserable, if united in our misery.
What if it wasn’t like this? What if we could simultaneously simplify life for users while at the same time increasing confidence in authentication decisions? What would it take to build authentication systems less vulnerable to compromise when an end-user was tricked into revealing a password?
There’s no mystery around the framework for solutions. It’s quite possible to design architectures for robust solutions that rely on all three of the traditional elements of authentication (something you know, something you have, something you are). At the September 2012 Intel Developer Forum, Intel’s Chief Technology Officer, Justin Rattner, demonstrated the ability to augment conventional passwords and tokens with the more robust schemes. Justin demonstrated a user walking up to a device and using biometric sensors to locally authenticate a user, then providing attestation to a service provider that user has successfully authenticated.
I think it would be great if these types of solutions were broadly available, delivered in solutions that were reliable, difficult to compromise, straightforward to manage and scale for large user populations, and – most importantly - simple for users! While some companies have delivered pieces of the puzzle, it’s still too tough for relying parties to stitch together a high quality fabric for authentication and decision making that’s easy for users. It may not be rings and wax seals, but we have some distance to go.
It would be great to hear your thoughts on these approaches. What are the biggest challenges you’re facing in balancing the robust and simple? Where can the industry do a better job solving your needs? (Hey and vendors, please, no blatant product promos).