| Comment Number: | 531096-00122 |
| Received: | 8/24/2007 3:33:15 AM |
| Organization: | |
| Commenter: | Jefferson Ogata |
| State: | MD |
| Agency: | Federal Trade Commission |
| Rule: | Private Sector Use of SSNs |
| No Attachments |
Comments:
In 1883, Auguste Kerckhoffs published his well-known essays on military cryptography, which set on paper some principles used widely today for strong authentication methods in computer systems. Regarding keys--i.e. the secrets used in a cryptographic system--he wrote that keys should be "memorable without notes and should be easily changed." The latter part of this is extremely important: if a password is compromised it should require little effort to change it so that no one else can use it to authenticate. All good password-based authentication systems make it easy to change one's password for exactly this reason. Unfortunately, financial institutions in this country have decided to use the SSN as a key. They did this because it required no effort on their part; they could pretend that the SSN was a secret--which it isn't--and thus pretend they were authenticating people without having to build a real authentication system. The SSN made this easy because a difficult problem in authenticating people is initial establishment of a key; establishing a shared secret is a chicken-and-egg problem. Consider PINs, for example--to establish a PIN for use of an ATM card, banks typically mail you a letter. If someone intercepts that letter, they know the PIN. Since the SSN is already known to the customer, the bank can ask the customer to use his SSN as a key, and the bank simply passes the name and SSN to the Social Security Administration for authentication. In a way, this method coopts the SSA for authenticating people to commercial financial institutions. There are two obvious problems with this: 1. Throughout one's life, many people have access to one's SSN: employers, tax preparers, relatives, banks, lenders, landlords, etc. Thus knowledge of an individual's SSN actually doesn't prove very much. 2. The SSN is extremely difficult to change, making it unsuitable as a key, since once it's compromised, it's compromised permanently. Would you be willing to share your checking account number and PIN with hundreds of people? What if these numbers were set in stone shortly after you were born and could never be changed? It's clear to people that work with computer security that the use of SSNs as authenticators is a very bad idea and should be stopped as soon as possible. In addition, because the SSN has value as an authenticator, criminal organizations have worked hard to collect SSNs of individuals, and therefore a given individual hasn't the faintest idea how many people know his SSN. So we need to go beyond the cessation of using SSNs as authenticators, and render them actually useless as authenticators. The most efficient way to do this is for the SSA to publish all SSNs with their account holders' names. Once this is done, no financial institution could reasonably claim that a transaction authenticated by SSN was meaningful, and thus consumers would no longer be at risk of fraud based on SSN compromise. Financial institutions that wish to authenticate individuals will have to go to the trouble of building systems for this purpose, which is how it should have been in the first place.