Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

Re: a plea for rationale appendix sections in the draft



Landon,

I agree that the standard would greatly benefit from a rationale  
section.

David

On Jan 4, 2006, at 10:39 AM, Landon Noll wrote:

> I'd like pause for a for the moment and say to those find
> themselves saying:
>
> 	"We have gone over this issue before!  Do we have to go
> 	 over it again??"
>
> I believe the answer must be YES!  I believe the reason why it
> keeps coming up by new members of the group is that the draft
> lacks what in P1003 we called "rationale" sections.
>
> 	Rationale sections are parallel sections to the draft that
> 	describe "the why", and that give some additional insight into
> 	the decision process of the group.
>
> The answer is yes, we must go over that ground again because the
> consensus was not sufficiently documented in the draft.  Fortunately
> this time it should be easier and quicker for those who know
> the group history to document their response in the form of
> rationale text.
>
> Let me give an example.  Say we have a "Section 1.3: Limitations".
> And for example, assume that in section 1.3 we state:
>
> 	A given LRW key should not be used to encrypt more than
> 	2^64 16-byte blocks.
>
> Then in an appendix we would have a parallel "Section R1.3:
> Limitations rationale" where we could state:
>
> 	(I'm hand waving here to serve as an example.  I'm not stating
> 	 the text below as fact.)
>
> 	The limitation of encrypting 2^64 16-byte was recommend
> 	because of the increased possibility of leaking very small
> 	amounts of plaintext.  The working group believed that the
> 	2^64 16-byte block was not a serious limitation in the year
> 	2006.  Transmitting that much data over a 128GBit bus would
> 	take over 544 years.
>
> So HAD the committee documented this consensus, THEN you could
> would great effect say:
>
> 	Are you sure you understand section 1.3 and R1.3?  If you
> 	do, then is there some text that we could add to that section
> 	that will make the point more clear?  Or is there a point that
> 	we missed that we need to take into account?
>
> If you go to IEEE ballot without these recurring / "we already
> discussed this issue!" matters being fully addressed, then you
> run the risk of IEEE balloters voting without the benefit of
> knowing the working group's experience and consensus.
>
> I believe that adding rationale sections to the draft will
> improve the chances of a successful IEEE ballot and it will
> increase the value the ballot process.
>
> chongo () /\oo/\