CTO Articles

Home > News > CTO Articles

IT World
E-Business in the Enterprise – July 15, 2003

Identity in the real world

By Sean Mc Grath

There are lots of technologies making the rounds right now that pertain to identity and identity management. A combination of factors are at play here. On one hand, there is the security situation which has breathed new impetus into the sector. On the other hand, there is B2B e-commerce which, some would argue is "on hold" pending resolution of the identity problem.

I have devoted more thinking time during working hours to this problem over the last year than I would like. After all, I do not consider myself to be a specialist in this area and security is a very specialized subject. My mental travels have re-enforced a sneaking suspicion that I have harbored for some time now. Namely, identity management is not a technology problem. No, scratch that. It doesn't sound right. Try this: if technology "solves" the identity problem then the identity problem - as envisaged in e-business - is different from the identity problem in normal business.

Let me explain. Various technologies are being put forward as ways to be absolutely* sure that someone is who they say they are. Biometrics, retina scans you name it. The word "absolute" is the one that I find most interesting here. After all, in the real world, we do not have any way of telling that someone absolutely is who they say they are.

What we do have, are a number of time honored ways of reducing the probability that someone is not who they say they are. Signatures, ID cards, passwords, casual questioning, history of successful transactions, referrals and the all important stare-in-the-eye-and-wait-for-blink method.

None of these are absolute methods. They are all statistical in nature. For any given transaction - from buying something with a credit card to getting a passport, we use a combination of these imperfect methods. We then subconsciously accumulate the probabilities. Given that I look like the person on the ID card (say 80%), have a passable signature likeness (say 70%) and can produce a utility bill with the right address (say 85%) - each of these probabilities is taken into account to yield a final probability that I am who I say I am.

Given the fuzziness of the real world in this regard, isn't it odd that we seem to be holding out for a perfect answer - an absolute method - in the electronic world?

To make matters worse, in business to business e-commerce, people (identities) do not act themselves. Machines talk to other machines and present credentials on behalf of identities. This raises the highly non-trivial question, what is the connection between the identity and the machine? If I write a Cobol program to invoke one of your web services on behalf of my employer, whose identity is presented to your web service?

All B2B is effectively "action at a distance" as far as identity management is concerned. We have a mechanism for handling this sort of action in the real world. It's called contract law and the trust model that goes along with it. It goes something like this. In order to do business we both sign a contract. Part of that contract is that I know where to find you and you know where to find me if things go sour. We both know that the contract is there so we "trust" each other not to mess around because of the penalties contained in the contract.

In the electronic world we seem to have got hung up on enabling a completely anonymous form of e-commerce. A form of business without contracts, without face time and without all the other probabilistic methods and risk mitigators we use in the real world.

Again I find this odd. It may turn out to be the most important thing about e-business and then again, maybe it will not. Whatever the future of completely anonymous e-commerce, it is beyond the state-of-the-art right now. Why then, don't we get up and running with plain vanilla contract-based, trust model enabled e-business on a global scale instead of waiting for a magic technological solution to identity to drop from the sky? Let's find ways of using a combination of evidence rather than one absolute method to check identity. Let's proceed to do business in the presence of risk. It's how we do it in the real world after all.

http://seanmcgrath.blogspot.com