Re: A unique number for every "person" - can it be done?

From: TGOS (tgos_at_invalid.invalid)
Date: 03/01/05 

Date: Tue, 01 Mar 2005 23:23:42 +0100

On 28 Feb 2005 10:29:01 -0800 john_ramsden@sagitta-ps.com wrote in
comp.programming:

> I'm curious to know why you used no-archive, as it's an interesting
> problem (considered in abstract - as a practical matter it's fairly
> pointless and impossible given all your constraints), but anyway ..

Because I usually post to groups that appreciate it and that dislike
Google. And I don't want to reconfigure my client all the time.

> It's easy for individuals, apart from maybe Siamese twins: Just use
> their exact GMT (no daylight saving) birth date/time, in ISO format
> [Y..YY]YYYYMMDDhhmmss (with the field delimiters compressed out),
> followed by the exact 10-digit zero-padded coordinates of their
> place of birth, to the nearest meter, followed by its exact distance
> from the centre of the Earth at the moment of birth.

How *huge* must a database be, to hold the coordinates of every place on
earth up to only a few meters and the distance of such a place to the
middle of earth? And you know the exact second of your birth? You are
sure the hospital clock was really accurate up to the second? And you
are sure the doctor really wrote down the exact time? And when is birth
over? Once the head is out, once the last toe is out? Once the umbilical
cord is cut?

Surely this would work, but such a database would be too huge and people
would otherwise fail to determine the coordinates and the distance to
earth middle. And the times would never be really accurate.

You can expect people to know the day of birth, but not the exact time,
not even up to the minute. And you can expect people to know the
hospital of their birth, but not coordinates of the room of birth up to
the meter.

> A central "Domesday Serial Key" generator would certainly have been
> the simplest solution!

But central control is exactly what I try to prevent.
 
> You'd have to assume a person knows their precise birth data, which
> leaves abandoned babies, e.g. found on church steps, at somewhat of
> a loss (unless their discovery time and place counts as their "birth"
> for the purpose of constructing their Domesday Id).

Well, what I like would be a formular of the kind:

ID = d1 + d2 + d3 + d4 + d5 ... + dn

And + does not mean it must be an addition, it's a placeholder for
whatever operation you'd like to use and the operation may even be
different for two consecutive + signs.

Of course, some people may not know d1 or d3 or d4 or maybe not know two
of them. But walking around that limitation is part of the game :)

Either they can choose whatever they like for the missing data, causing
a lot of randomness and avoids collision, but also reduces
reproducibility (people may simply forget what they chose). Or they must
use fixed, generic, placeholder data, so the rest of the data, that is
not missing, must be good enough to still make collisions highly
unlikely.

> I reckon this makes the task practically impossible, unless one simply
> plays with words and says that a person's Id should be their _mother's_
> exact birth time and place (or for abandoned orphans, their
> discoverer's birth details), with an extra index to uniquely identify
> each of her children.

Well, I wait for a wonder formula.

I mean, I could go and say: Let's make a String containing first/last
name of parents, your name, your birthday, your blood type, your gender,
your eye color, and so on; and finally hash it all to 512 bit hash.

How big are 2^512? More than the universe has atoms? Should be enough in
any case. But that calls the cryptographers to the table, explaining how
likely a collision would be over a several thousand years and which hash
algorithm could be used in the first place? The likeliness is tiny, if
the algorithm is good. But which 512 hash is really good? And can it be
proven, that it is good?

A scheme that can be proven to be collision free would be better.

> Seems like you're trying to turn into something practical a problem
> which in its full generality is almost open-ended, unless you assume
> peoples' bodily limitations will remain moreorless what they are today
> (which, if new scientific developments continue at their present rate,
> is highly unlikely).

How does the fact that people could live forever touch the problem in
any way? Or do you speak about people living without bodies? Still the
same problem.

>> [...]
>>
>> - Blood Type (consider companies have no blood type)
>> - Gender (consider companies)
>> - Eye color (should be constant, consider companies)
>
> All those, and many more (including no doubt fingerprints), will
> probably be capable of being changed at the drop of a hat (or pill
> anyway) well within one or two hundred years let alone billions !

Maybe, but you are only born with one blood type, one gender and one eye
color. If you change it later on, that's your choice, but still you had
these unique identifiers after birth.

Maybe not anymore in a billion of years, but than the ID generation
formula could be slightly adjusted and some data is replaced with some
other data. As long as this doesn't cause conflicts with old IDs. 

-- 
TGOS

Relevant Pages

  • Re: GB2RS NEWS Sunday 13 April 2008
    ... I don't know your exact date of birth, ... at the time you took the RAE, and also mentioning the month and year.. ... not an exercise ICBA to do. ...
    (uk.radio.amateur)
  • Re: A unique number for every "person" - can it be done?
    ... > on earth, that is guaranteed to be unique till the end of time. ... their exact GMT birth date/time, ... place of birth, to the nearest meter, followed by its exact distance ... The snag with this approach is that individual Ids have variable ...
    (comp.programming)
  • Re: A unique number for every "person" - can it be done?
    ... > on earth, that is guaranteed to be unique till the end of time. ... their exact GMT birth date/time, ... place of birth, to the nearest meter, followed by its exact distance ... The snag with this approach is that individual Ids have variable ...
    (sci.math)
  • Re: A unique number for every "person" - can it be done?
    ... > on earth, that is guaranteed to be unique till the end of time. ... their exact GMT birth date/time, ... place of birth, to the nearest meter, followed by its exact distance ... The snag with this approach is that individual Ids have variable ...
    (sci.crypt)
  • Re: A unique number for every "person" - can it be done?
    ... > from the centre of the Earth at the moment of birth. ... And you know the exact second of your birth? ... and so on; and finally hash it all to 512 bit hash. ... likely a collision would be over a several thousand years and which hash ...
    (sci.crypt)
Loading