Re: TSA Response serialNumber Field

[Paul Koning <pkoning@xxxxxxxxx>]

>>>>> "Peter" == Peter Sylvester <Peter.Sylvester@EdelWeb.fr> writes:

 >>  1) Time Stamp Token ordering can be done in most cases (see
 >> above) by only using GenTime and accuracy.

 Peter> A litte flame about accuracy: A user that wants to get a time
 Peter> stamp in order not to miss a deadline goes to a TSA: The TSA
 Peter> tells, well it is 23h59'59" but maybe not, since we are not
 Peter> sure whether we have the right time, it may already be well 2
 Peter> seconds later or earlier, well, that's what we think, or
 Peter> almost.

 Peter> What kind of decision can you make based on that?  What would
 Peter> be the harm if the TSA would simply adjust the time to the
 Peter> lower end of the interval, and generate no accuracy.  If there
 Peter> is an application that requires to get the earlimost stamp
 Peter> just after midnight, and you are too early, you just repeat.

 Peter> Comparison between two time stamps of the same tsa can be done
 Peter> without using the accuracy; if the genTime is different, then
 Peter> there they are ordered by authority of the TSA.

I can't see the point of this.

The interpretation of a timestamp with accuracy is simple: the TSA is
making an assertion that the time when it generated the stamp is in
the range [ts - accuracy .. ts + accuracy].  (In other words, it
promises that UTC will be contained in that interval.)

So if you want to know the lower end of the interval, fine, you can
compute that trivially.  It is not necessary to throw away information
in the protocol to give you that option.

Other applications may want to know the upper end of the interval.
And a number will want to know the interval itself.

For example, to order timestamps from two different TSAs, you have to
have the intervals.  The order is a partial order: if the intervals
don't overlap, the answer is obvious, and if they do overlap, the
order is indeterminate.  This in fact is the reason for having the
accuracy field.


>>>>> "Michael" == Michael Zolotarev <mzolotarev@baltimore.com> writes:

 Michael> 3. a stamp issued by another TSA (AND the precision
 Michael> diapazons of the two stamps overlap). The only purpose of
 Michael> comparing is to figure out which stamp was issued
 Michael> earlier. We have two cases to consider here:

 Michael> Case1: Precision_TSA1 is equal to Precision_TSA2.  Here the
 Michael> rule would be to treat the precision as the span of the
 Michael> probability graph, with the probability being the highest at
 Michael> the middle point (base time value), and 0 at the ends of the
 Michael> 'hill'. Therefore, the court (my court :) would simply opt
 Michael> to ignore the precision field at all, and look at just the
 Michael> base values.  I would consequtively reject any claims that
 Michael> *your* probability graph is not a 'hill-shaped' curve.

 Michael> Case2: Precision_TSA1 is smaller than Precision_TSA2.  As a
 Michael> judge I fully trust the TSA1 'cause it's time is more
 Michael> precise. So I again look at the base time value. If TSA1
 Michael> stamp says it was generated a second earlier than TSA2 stamp
 Michael> - that's the truth. If TSA1 stamp says it was generated a
 Michael> second later than TSA2 stamp - that's the truth as well.

I don't understand this reasoning at all.  Whether TSA1 has an
accuracy smaller than that of TSA2 isn't interesting, as far as I can
see.  I don't understand why you would use it to make a binary
decision of merit (i.e., trusting only the TSA that has the smaller
accuracy). 

If you ignore the accuracy and compare only midpoints, you will
definitely be making an incorrect decision when the intervals
overlap.  ("Incorrect" in the sense of "not supported by the
evidence".)   This is true whether the accuracy values are the same or
different.  (For example, in your case 2, if TSA1 says 1:00:00 +/- 1s,
and TSA2 says 1:00:01 +/- 3s, it sounds like you're saying that the
TSA1 stamp is generated a second before the TSA2 stamp.  Not so -- the
TSA2 stamp may have been generated as much as 3 seconds before the
TSA1 stamp.)

     paul