Re: Good enough for crypto?
From: Scott Wilber (swilber_at_comscire.com)
Date: 12/10/03
- Next message: David Wagner: "Re: Voice encryption (Stream vs CBC mode)"
- Previous message: Mario Fischer: "Remarks: Re: MIX-Network Playback attack"
- In reply to: Tom St Denis: "Re: Good enough for crypto?"
- Next in thread: Tom St Denis: "Re: Good enough for crypto?"
- Reply: Tom St Denis: "Re: Good enough for crypto?"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]
Date: 9 Dec 2003 15:21:35 -0800
"Tom St Denis" <tomstdenis@iahu.ca> wrote in message news:<yqpBb.23165$r%u1.4443@twister01.bloor.is.net.cable.rogers.com>...
> "R3769" <r3769@aol.com> wrote in message
> news:20031209134709.08176.00000520@mb-m16.aol.com...
> > MKS wrote:
> >
> > > So how is one to be convinced that your
> > >package leads to better decision on the quality of
> > >sequences being tested than the other well-known
> > >packages?
Try testing a 10Mbit file with the NIST tests.
> > However, the claim that most of the generators that pass Diehard fail the
> > Comscire tests would be much more meaningful if only files of equal sizes
> where
> > tested. How many datasets of 11mb are there that pass Diehard and yet
> fail
> > RNGmeter? Has even one been exhibited? I call this the "Lizard" test for
> > randomness testers.
>
> If the dude wants to make sales he should back it up. E.g. if a PRNG fails
> his test that means his test can predict the outputs with a non-uniform
> probability [since afterall that's all you really need for "randomness"].
>
> Show a PRNG which fails then show how to use that to recover the output.
>
> Tom
The RNGmeter is simply a tool for comparing the statistical quality of
various generators, partly as an aid to designing better ones. I
don't believe we have claimed anything else, except that it is a good
tool. Also, you may note that the RNGmeter is free.
If it could automatically use a failure in statistics to "recover the
output," then I think the US government would not allow it to be given
away on the internet.
Scott
- Next message: David Wagner: "Re: Voice encryption (Stream vs CBC mode)"
- Previous message: Mario Fischer: "Remarks: Re: MIX-Network Playback attack"
- In reply to: Tom St Denis: "Re: Good enough for crypto?"
- Next in thread: Tom St Denis: "Re: Good enough for crypto?"
- Reply: Tom St Denis: "Re: Good enough for crypto?"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]
Relevant Pages
|