Re: [httpRange-14] What do HTTP URIs Identify?

On Friday, August 2, 2002, 8:26:57 PM, Paul wrote:


PP> But detecting and dealing with two people using the same identifier to
PP> mean different things is hard, because you can never differentiate it
PP> from a mere contradiction. "Michael Jackson is white." "No, Michael
PP> Jackson is black." Are we disagreeing on Michael's pigment du jour or
PP> talking about different Michaels? The standard way to decide is to use
PP> an addressing/naming system that we agree is globally unambiguous. "I'm
PP> talking about the Michael Jackson with this social security number. Who
PP> are you talking about?"

You deal with the case where the SSN shows they are different people,
but not the case where it shows they are the same person.

PP> If, after using a global identifier, we could *still* be talking about
PP> two different guys then the global identifier is useless and should be
PP> discarded.

But if the two SSNs match and you are talking about two different
periods in that individuals life, and supposing the details of that
lifer were not so well known, then the two people might still disagree
that they were talking about the same person, because their
invariability assumptions would have been challenged.

PP> If we can't depend upon that then you have no way to detect
PP> contradictions because you must constantly remain open to the
PP> possibility that they are really ambiguities.

You can't depend on it, as you sday, and it requires a human to give
an estimate of whether it is more likely to be an error or an
ambiguity. Different humans will give different answers.

To pursue the time-based difference further - based on two point
samples of skin coloration in the example above, there was scope for
either ambiguity, or for error (which would be given a ghigher
probability if the two people conducting the conversation believed that
skin pigmentation is invariant).

To bring this back to the Web - two people discussing
http://www.w3.org/TR/ might agree they are talking about the same
thing, over a long period of time. Then again, as soon as someone
asserts that the fourth letter of the second word in the fifth
paragraph is an "e", then they disagree based on the dates of the
respective point samples that they have chosen to use.

PP> Another analogy: the credit card companies routinely deal with
PP> contradictions: "I bought that. I didn't buy that." But they define away
PP> ambiguity. By definition, two credit cards with the same number are the
PP> same credit card. If they detect two people using the "same" credit card
PP> (number) in conflicting ways they just invalidate the card: they do not
PP> split the card into two logical cards and try to deal with the
PP> ambiguity.

Which is fine until some company in some country starts issuing
'spouse' cards with the same number.

PP> They don't tell merchants: "this credit card number is being used
PP> ambiguously. Please figure out which of the two people you are talking
PP> to." They say: "This credit card number is generating an unreasonable
PP> number of contradictory statements. Please ignore that card from now
PP> on."

That doesn't make it a true statement though, just the one they are
currently betting on based on their experience.

-- 
 Chris                            mailto:chris@w3.org

Received on Saturday, 3 August 2002 15:59:32 UTC