Re: I-D Action: draft-ietf-6man-default-iids-11.txt

[Brian E Carpenter <[e-mail]>]

A few years ago I found that the academic literature on hash functions seems
extremely weak. Actually a very good theoretician of randomness that I know
referred me to a paper from 1951*. Most of the "literature" appears
to be experimental results ("this algorithm seems to work better than
that algorithm").

http://www.isthe.com/chongo/tech/comp/fnv/index.html for example.

Regards
   Brian

* J. von Neumann, Various techniques used in connection with random digits, National
Bureau of Standards Applied Math Series 12, 36-38, 1951.


On 25/05/2016 14:58, weigengyu wrote:
> Hi Fernando,
> 
> Would you please provide the published refereces of Alfred Hoenes -- IIRC, a mathematician.
> IIRC, he noted that a hash function is simply one possible way of a PRF.
> 
> It is interested for the proof of Hash function to be a PRF.
> 
> Regards,
> 
> Gengyu WEI
> Network Technology Center
> School of Computer
> Beijing University of Posts and Telecommunications
> -----原始邮件----- From: Fernando Gont
> Sent: Wednesday, May 18, 2016 10:19 AM
> To: Brian E Carpenter ; weigengyu
> Cc: [e-mail]
> Subject: Re: I-D Action: draft-ietf-6man-default-iids-11.txt
> 
> On 05/16/2016 04:08 PM, Brian E Carpenter wrote:
>> Hi Gengyu,
>>
>>>> RID is not random or random identifier according to the context above.
>>>>
>>>> F() is an inreverse function, not a pseudorandom function.
>>
>> I agree that this is incorrect use of terminology; an irreversible hash is
>> not mathematically equivalent to a pseudo-random number generator. But the
>> effect is the same: an attacker has no way to predict the next F(x) or to
>> obtain the value of x. As far as practical engineering goes, F() can be treated
>> as a random number. So I don't see why this invalidates the use of RFC 7217.
>>
>> You could submit an erratum for RFC 7217 at http://www.rfc-editor.org/errata.php#reportnew
>> so that the terminology can be corrected when the document is updated.
> 
> FWIW, the term "PRF" (as *opposed*) to PRNG was suggested by Alfred
> Hoenes -- IIRC, a mathematician. IIRC, he noted that a hash function is
> simply one possible way of a PRF.
> 
>