Re: dbdebunk 'Quote of Week' comment

From: JGE <>
Date: Fri, 19 Aug 2005 23:04:05 -0000
Message-ID: <>

On 2005-08-19, -CELKO- <> wrote:
>>> Here in Canada, a SIN. There are many situations where one needs to store information about people where one has no right to ask for and the person has no obligation to deliver, their SSN. <<
> We have some of the same privacy laws here, but you still need it to
> pay taxes.

Well obviously. How does that justify (as an example) Radio Shack asking for a social security number? Because it is a natural key? (They don't of course because most people would tell them to bugger off. As such, they can't use the SSN as a primary key). My point here is actually a pet peeve of mine. The SIN (here in Canada) and the SSN are NOT intended for general usage. If you have a database that involves income, then you MAY have a right to ask for the SSN. Otherwise, I'm not going to tell you and you can't make me. And if I'm a customer in a customer database, you have to ask: Is the customer wrong?

>>> Also, is not the SSN simply an artificial key, generated by the government? As such, how does it differ from the lovely primary key or 'identity' that Access is so generous to provide? In principle, rather than in scale. <<
> TOTALLY different In principle! -- I have an article on how to
> validate and verify an SSN sitting at in the slush
> pile. The SSN has nothing to do with the internal state of a single
> piece of hardware. It is generated by a set of rules in a rigid fomat,
> verified by the Social Security Adminstration and issued numbers are
> posted once a month on the Internet so you can look them up. Certain
> SSN are never issued, certain ranges are available for assignment to
> persons without an SSN, etc. This is not just a serial number.

I'm not sure why the fact that the algorithm for generating a key is more complicated than x=x+1 changes the fact that an SSN is still a generated key that is intended to be an artificial and unique identifier of a member of a set and that identifier has nothing to do with the elements of the set. Yes it is more than a serial number. No, in principal, it is no different from that automatically generated key thing in Access. More complicated yes. Platonically different. No. I could be wrong here Joe. I'll leave this thread and the last word to you (if you're interested). Also, I'm not quite as well versed on the theory as you are.


JE Received on Sat Aug 20 2005 - 01:04:05 CEST

Original text of this message