Re: [Geopriv] draft-ietf-geopriv-uncertainty-00

["Marc Linsner (mlinsner)" <[e-mail]>]

I will guarantee 100% I am in the state of Florida, USA as I type this.
Zero uncertainty!

-Marc-

-----Original Message-----
From: Martin Thomson <[e-mail]>
Date: Thursday, April 17, 2014 at 1:16 PM
To: Marc Linsner <[e-mail]>
Cc: "[e-mail]" <[e-mail]>
Subject: Re: [Geopriv] draft-ietf-geopriv-uncertainty-00

>I see where this is coming from now.  I disagree with your conclusion.
>
>On 17 April 2014 10:05, Marc Linsner (mlinsner) <[e-mail]>
>wrote:
>> Uncertainty wrt to a measured location, expressed as 95% confidence,
>> equates to a 5% chance the target is OUTSIDE the given polygon.
>
>To be precise, it means that there is a *less than* 5% chance that the
>target is somewhere else.
>
>> There is ZERO CHANCE the target is outside the supplied elements.
>
>There is *never* a zero chance that the information is wrong.  Because
>that would require an infallible determination of the supplied
>elements.  100% confidence is basically impossible [1].
>
>Civic address information is still prone to provisioning errors.
>You'd think that with systems like MSAG, the error rate would be much
>lower than 5% and you are probably right.  Feel free to indicate
>99.999% confidence.  Though I think that I'd want you to back that
>claim with some fairly extensive testing.  But don't ever think that a
>mistake can't happen.
>
>Then there are the cases where the input information has been tweaked
>to affect the outcome.  And the outright lies.
>
>--Martin
>
>[1] Which is one reason that climate science has such a hard time in
>the court of popular opinion.