Re: One to One relationships

In message <46e627da.0305070647.2c2b84a8_at_posting.google.com>, AK <ak_tiredofspam_at_yahoo.com> writes
>>
>> In general there isn't a good natural key to identify people, that's why
>> most databases use various surrogates (like SSN.) I've proposed a
>> natural key for people; the precise latitude, longitude, altitude and
>> time of their birth. I don't claim that it's a practical system.
>
>I think that's a good idea. But then what do you think of fingerprints?

There are two problems with fingerprints.

The first is that there I know of no reason why there shouldn't be two or more people with identical fingerprints. It's statistically improbable that any two people will have identical prints, but not impossible. There's no law of physics (that I'm aware of) that prevents it.

The second is that that we don't usually compare fingerprints. Instead we often use some sort of hashing routine to convert the fingerprint to a code and then compare the codes. Those hashing algorithms are imperfect and that means that it is even more likely that different people will generate matching codes.

Fingerprints and other biometric techniques have their uses, but they aren't perfect.

-- 
Bernard Peek
bap_at_shrdlu.com
www.diversebooks.com: SF & Computing book reviews and more.....

In search of cognoscenti