[futurebasic] RE: [FB] Arithmetic Encoding Compression

Message: < previous - next > : Reply : Subscribe : Cleanse
Home   : October 2001 : Group Archive : Group : All Groups

From: "Edwards, Waverly" <Waverly.Edwards@...>
Date: Fri, 26 Oct 2001 11:25:15 -0400
There are small algorithms if you dont want to spend half you life making
foolproof code.  If your concerned with the average person, they'll probably
try little more than an XOR to unencrypt the code.  If you're concerned with
hackers, you'd be better off using a well known (but complex) algorithm

Here is a small one.

http://www.ecst.csuchico.edu/~atman/Crypto/code/tea-encryption.alg.html


W.

-----Original Message-----
From: jonathan [mailto:jonnnathan@...]
Sent: Friday, October 26, 2001 6:23 AM
To: futurebasic@...
Subject: Re: [FB] Arithmetic Encoding Compression


le 2001/10/26 10:16, Heather Donahue à heatherd@... a écrit :

> The Applied Crypto book has all the major algorithms but I wonder if there
are
> any open-source ones on the internet?

most valuable algorithms tend to be open source as peer review is the best
way of ensuring the degree of 'unbreakability' of a crypto solution.

the most basic rule is: don't rely on an algorithm being secret to ensure
its strength and inviolability. the more others can pick and hack at it, the
stronger it will be... you only have to look at how all the 'attempts' to
keep so-called 'customer-management' developments around digital music
'secret' leads to weak crypto. and making reverse engineering, or research
into cracking these 'locks' illegal in the us [CDMA] will not make the
problem go away either.

the other factor that is often overlooked in crypto is 'time'. if your
crypted communication only needs live for a very short time, you can use
weaker crypto provided that:
- current methods cannot break it in real time [or in the time that it will
live]
- current methods do not allow the creation of a spoof method in real time

and provided that you survey 'current methods' very closely and increase
strength as techniques [and machines] improve [which is what the french
credit card cartel didn't do, thus allowing one level of crypto on the cards
to be cracked a while back*].

most 'brute strength' methods [ie, script-kiddie level] assume that there is
plaintext encoded. the 'cracked' text is then analysed [quickly] to see if
the letter frequency distribution corresponds to the supposed target
language [ie, more 'e's than 'z's in english - think scrabble!]. i remember
reading of a cryptographer cracking a code but not realising that this was
so as the plaintext was in turkish! thus a scrambling algorithm prior to
encrypting removes the easy letter-frequency method of determining that the
cracking was successful. remember also that the shorter the text, the harder
it is to crack, and, as a corollary to the before, the fewer 'obvious'
words/letters in a crypted stream [ie, even if it is crypted you can
probably recognise 'AZGS-SCHS-SHCY-HHYS' as credit card quads], the harder
it will be to crack.

´:-j

* the guy who cracked this was a bit of a nutter, but that doesn't matter.
he contacted the group of card issuers telling them he had cracked it, but
they didn't believe him; they said 'send us proof'. so he carried out a
transaction, in the presence of a public notary as an independant witness.
they got him jailed on this basis of this! and then hummed and hawed about
the cost of upgrading the system. meanwhiles others have discovered his
crack, and the system won't be upgraded entirely before next year... however
the courts have changed their jurisprudence following this and now if you
contest a transaction, the onus is on the banks to prove that that it was
not fraudulent, rather than before when the card holder had to prove her
innocence...



--
To unsubscribe, send ANY message to <futurebasic-unsubscribe@...>