[arin-ppml] Recommended Draft Policy ARIN-2014-12: Anti-hijack Policy

[John Santos]

On Fri, 6 Jun 2014, Leif Sawyer wrote:

> On 6/6/14, 11:04 , David Farmer wrote:
> > [...]Given the "should" is immediately followed by a conditional "unless"
> > the intent seems sufficiently clear, the intent is to create a special-case
> > exception, and "should" seems appropriate.  Furthermore, "must" or "shall"
> > followed by "unless" seemed an awkward way to create such an exception.
> > 
> > Staff generally agrees that in most cases for policy "must" is preferred
> > and it is best to avoid "should" in most cases.  However, in the sentence
> > above the intent seem clear enough and "should" seems appropriate in that
> > particular case.
> 
> Unfortunately, that still has indirect parsing issues.

I disagree with your parsing of both examples.

"Should" implies advice, rather than a requirement.

> 
> 1. You should eat an ice-cream cone, unless you ate a taco.
>   [and then you shouldn't...but you still could]

This statement does NOT imply that you should not eat an ice cream cone
if you ate a taco.  It says all non-taco eaters are advised to eat an
ice-cream cone.  It actually says nothing about taco eaters.  They
can choose whether or not to have an ice-cream cone on their own.

> 
> 2. You must eat an ice-cream cone, unless you ate a taco.
>    [ sorry, no ice-cream for you, taco-eater.  You get a churro instead. ]
> 

Taco eaters can have a cone if they want.  It is only non-taco eaters
who have a requirement that they eat an ice-cream cone.  This
statement, like the first. imposes no requirements nor recommendations
on taco eaters, one way or the other.

> 
> It's tri-state versus dual-state.  If the objective is to refine
> intent, then we should be most clear about our intent and diminish 
> the grey areas where possible.

This I agree with.  If we're going to go all medieval about language,
we must (or should?) do it right.

I think all the shoulds should be musts, and also the phrase "experimental
documentation" should be replaced by "request" (i.e. "their request must
clearly describe and justify why this is required.")

However, if the choice is between fixing the bug that causes real world
problems (the overlapping allocations of experimental space with allocated
space) immediately and fixing all the bugs in 6 months, including those
that haven't actually been manifest, I would vote for the immediate fix
and would support fixing the rest of the wording later (to answer
David's question in a later email.)


-- 
John Santos
Evans Griffiths & Hart, Inc.
781-861-0670 ext 539