More on fortifying Lotus Notes

[Duncan Campbell]

4 June 99

The issue of the NSA trapdoor in the International Edition of Lotus Notes 4 
is attracting a number of argumentative strands in different places :

http://www.heise.de/tp/english/inhalt/te/2898/1.html

http://www.heise.de/bin/tp/forum/get/telepolis/2115.html

Its also been one of the more commented on issues arising from the new 
European Parliament report on Echelon.

http://www.iptvreports.mcmail.com/ic2kreport.htm#_Toc448565572

One of the features there and in private correspondence I've had is that 
IBM/Lotus folk feel aggrieved because their crypto system is better than MS 
Mail and other US competitors, so I'm being unfair in pointing out how it 
has been tailored to suit NSA surveillance.    That position is 
understandable.

MS Mail (and Netscape, etc)  are completely crypto-crippled, while Lotus 
pretends not to be, by having an NSA trapdoor instead.  Choose (a) MS (b) 
Lotus or (c) something not made in the US (or other UKUSA nation) and not 
required to be NSA surveillance - friendly.

Now that the crypto barriers are coming down completely within the EU, 
there can be no justification for EU customers buying export-controlled 
US-licensed software for any communications or information security 
application.  This is the nightmare that US manufacturers warned the US 
goverment about.  Now they have to face the consequences.

Duncan Campbell




>Paul Crowley <paul@hedonism.demon.co.uk>
>
> > * What legal hurdles stand in the way of (a) using a bunch of tools to
> > search the binary files that come with Notes to find the embedded
> > public key, (b) publishing the key, and (c) writing a program to find
> > the key and scramble it?
>
>The tools are already here.
>
>  od will show you the content
>
>  dd if=INPUT_FILE of=df bs=1 count=3 seek=10374 conv=notrunc
>  I've just writen 3 'A's into a binary of 'df'.
>002882: 64 20 41 76 41 41 41 61 62 6c 65 20 43 61 70 61 d AvAAAable Capa
>
>Writing a real binary editor is not that hard either.
>
>
>* What should be done to the key once it's found?  Is it sufficient to
> > replace most of it with random noise, or is it important that it be
> > replaced with a real key?
>
>Experiment ought to find that out.  It would be fairly easy for them to
>have some built-in check at encryption time, but they may not have
>bothered.   Not much is really worth doing in a model where someone can
>make arbitrary changes to the binaries you ship.