From: David Honig <honig@sprynet.com>
Date: Tue, 27 Oct 1998 04:29:55 +0800
To: Mok-Kong Shen <mok-kong.shen@stud.uni-muenchen.de>
Subject: Re: PRNGs and testers.
In-Reply-To: <19981021151835.A26267@krdl.org.sg>
Message-ID: <3.0.5.32.19981026110933.007ba9a0@m7.sprynet.com>
MIME-Version: 1.0
Content-Type: text/plain



At 09:47 AM 10/26/98 +0100, Mok-Kong Shen wrote:
>David Honig wrote:
>> Now run Maurer's test; I've posted a version for blocksize = 16.
>> The cipher-PRNG output will not have the entropy expected for randomness.
>> The physical-random file will.
>>
>
>I don't see you have answered my question of whether a test has to
>take into consideration how a sequence of numbers has been obtained.

A test measures only what it measures.

You never know if this is sufficient.

A sample of a non-random process may pass your tests
if they don't measure the right thing.

>Also what you wrote above seems to be less than clear. Do you suggest
>that Mauerer's test is extremely good in deciding whether a sequence 
>is TRULY random? (I think Maurer's test is good for investigating 

I only find it interesting that the test can distinguish between
the two data samples.  To the eye, or ear, or Diehard, they appear the same.

>What if some PRNGs pass Maurer's test?
>
>M. K. Shen

I'd find it surprising if any did, given what I described.

DH