Re: [wg-c] Non Destructive Testing (SNT)

["Mark C. Langston" <skritch@home.com>]

On 18 November 1999, "Robert F. Connelly" <rconnell@psi-japan.com> wrote:


>Dear Colleagues:
>
>Permit me to post a little redirect to Milt's comments on Dr. Bray's 
>homily, "One test is worth three expert opinions."
>
>There are some pretty tight limits on what one is permitted to test.  There 
>are stories about the testing performed on humans during the dark ages of 
>WWII.  Today, we have tight limits on genetic research.  For reasons often 
>good, occasionally ill conceived, society puts limits on what kinds of 
>tests are proper to take.
>
>Years ago, I spoke before chapters of the Society for Non Destructive 
>Testing.  SNT, as it called, includes fluorescent penetrant, x-ray 
>inspection and a raft of procedures used in the aerospace industry.
>
>Dr. Bray had a name for tests which destroyed the sample, "It's sorta like 
>the possum, he ain't much good after you've skinned him once or twice!"
>
>That analogy could be applied to Milt's proposal to try one or more of 
>every combination and permutation of models posted here (all except the 
>delegation of no gTLDs at all, which was a proposal to do nothing.)
>
>Once delegated, innocent third parties will register names, put up web 
>sites, link them to other web sites and be linked from other web sites by 
>innocent fourth parties.  If and when it becomes clear that a particular 
>option was ill advised, it will be very upsetting to a lot of folk, 
>innocent parties of the fifth degree of separation.  They will be upset, 
>not at wg-c, of which they are totally unaware, but at a substantial part 
>of this Internet of which we are so dear.
>
>Let's test, but be selective about what we test.
>


Bob, I'm sympathetic to non-destructive testing.  But non-destructive
testing and testing that can't abide any failure whatsoever are two
different things.  The former's a matter of methodology;  the latter's
not a valid test.

How can we perform a real test if none of the testing candidates can
fail in any way?  (and let's avoid semantics like, 'well, we only
test candidates that will show varying degrees of success,' because
one person's catastrophic failure is another person's significantly
poor success rate.)

-- 
Mark C. Langston
mark@bitshift.org
Systems Admin
San Jose, CA