Prev: Question About Cryptographically Hashing a Hash (SHA-512), ThenHashing That Hash, Etc.
Next: Fallout 3 crypto?
From: Matthias on 8 Jun 2010 17:15
I am looking for a 'wallet vault' that allows me to carry the pin codes
of my diverse credit and ATM cards safely with me on a slip of paper and
decipher them without the aid of computers or pocket calculators in a
simple paper and pencil operation in a matter of minutes in case I forgot
one of them.

This is my idea:

Assume that the pin code of my first credit card is 56 78 (of course, it
is not). I convert it two digit wise into hexadecimal format, yielding 38
4E.

Now a simple to remember passphrase comes in, assume it to be
THEMADHATTER. To encipher the first pin code I use the first two letters
of my passphrase, or rather their hexadecimal ASCII representations, i.e.
54 48 (You need not to learn the ASCII table by heart, you need to know
that A is 41h and count on - or forget about ASCII and choose any other
suitable numerical representation of the alphabet.)

Enciphering is done by XORing both values wordwise.

38 4E (the pin code) XOR
54 48 (the first two letters of the passphrase) yields
-- --
6C 06 (that's what I write on my 'remember me' slip)

For the next pin code the next two letters of the passphrase 'EM' will be
used, and so on.

Deciphering is done in the same fashion, ciphertext wordwise XOR
passphrase letters.

It does not look very safe at a first glance. It is not a theoretically
safe one time pad at all, since the 'pad' is far from even looking
random. On the other hand a dictionary attack will fail for the obvious
reason that most words or expressions will yield possibly valid pin
codes, i.e. four digit numbers in the (decimal) range 0 to 9999.

A statistical attack might be promising if the opponent had a large
enough sample of ciphertext to work with, but all he has is a slip of
paper with typically < 20 Bytes of information.

Do you see any feasible attack under the given assumptions?


Matthias


--
With or without religion, good people can behave well and bad people can
do evil; but for good people to do evil — that takes religion. [Steven
Weinberg, Nobel Laureate in Physics]
From: Paul Rubin on 8 Jun 2010 17:50
Matthias <arnd-matthias.langner(a)arcor.de> writes:
> I am looking for a 'wallet vault' that allows me to carry the pin codes ...
> I convert it two digit wise into hexadecimal format, yielding 38 4E.

Seems way too complicated. Who are you trying to protect these numbers
against? What kinds of attacks do you think they can use? The usual
situation with an ATM card is if you guess the wrong PIN three times in
a row, the machine confiscates or invalidates the card. So it's likely
enough to do something very simple and then keep the method to yourself.
Main thing is to not identify what the numbers are.
From: Maaartin on 8 Jun 2010 18:08
On Jun 8, 11:50 pm, Paul Rubin <no.em...(a)nospam.invalid> wrote:
> Matthias <arnd-matthias.lang...(a)arcor.de> writes:
> > I am looking for a 'wallet vault' that allows me to carry the pin codes ...
> > I convert it two digit wise into hexadecimal format, yielding 38 4E.
>
> Seems way too complicated.  Who are you trying to protect these numbers
> against?  What kinds of attacks do you think they can use?  The usual
> situation with an ATM card is if you guess the wrong PIN three times in
> a row, the machine confiscates or invalidates the card.  So it's likely
> enough to do something very simple and then keep the method to yourself.
> Main thing is to not identify what the numbers are.

Agreed. First, I'd avoid the hexadecimal and switch to decimal with
addition/subtraction mod 10000, where addition gets used for
decryption. Second, I'd switch to digit-wise addition since it's
easier and as good as the normal one here. Third, I'd use a numerical
key instead, so I need no conversion.

N-th, I'd do something completely different, what I don't reveal here,
as in this case the algorithm is part of the key. I was using it and
maybe I'll again, and I wrote the numbers on the cards themselves, so
I couldn't loose it, and the attacker would have had something to
think about. :D
From: Matthias on 8 Jun 2010 18:27
On Tue, 08 Jun 2010 14:50:34 -0700, Paul Rubin wrote:

> Matthias <arnd-matthias.langner(a)arcor.de> writes:
>> I am looking for a 'wallet vault' that allows me to carry the pin codes
>> ... I convert it two digit wise into hexadecimal format, yielding 38
>> 4E.
>
> Seems way too complicated. Who are you trying to protect these numbers
> against? What kinds of attacks do you think they can use? The usual
> situation with an ATM card is if you guess the wrong PIN three times in
> a row, the machine confiscates or invalidates the card.
My imagined opponent is a clever thief. I read the story of an elderly
lady in my newspaper, who wrote down her pin codes as fictitious phone
numbers in her notebook. It took the thief only several phone calls to
get a clue ....


>So it's likely enough to do something very simple and then keep the
>method to yourself.

Security by obscurity is not always a good idea. My assumption was: The
thief has my wallet with all the credit and ATM cards along with the slip
of paper with the enciphered pin codes. He has an idea which information
this slip of paper might contain and which algorithm I may have used, and
he has - as you mentioned - three tries per card.

Of course it is obvious that trivial passwords (first names or birth
dates of wife or kids ...) are insecure in this scenario. I assume a not
too easy to guess passphrase.

I accept 'way to complicated' if it implies reasonably secure, but this
is a logical non sequitur. Complicated is not necessarily secure. This
was the point of my question.


Matthias



--
With or without religion, good people can behave well and bad people can
do evil; but for good people to do evil — that takes religion. [Steven
Weinberg, Nobel Laureate in Physics]
From: David Eather on 8 Jun 2010 18:32
On 9/06/2010 7:15 AM, Matthias wrote:
> I am looking for a 'wallet vault' that allows me to carry the pin codes
> of my diverse credit and ATM cards safely with me on a slip of paper and
> decipher them without the aid of computers or pocket calculators in a
> simple paper and pencil operation in a matter of minutes in case I forgot
> one of them.
>
> This is my idea:
>
> Assume that the pin code of my first credit card is 56 78 (of course, it
> is not). I convert it two digit wise into hexadecimal format, yielding 38
> 4E.
>
> Now a simple to remember passphrase comes in, assume it to be
> THEMADHATTER. To encipher the first pin code I use the first two letters
> of my passphrase, or rather their hexadecimal ASCII representations, i.e.
> 54 48 (You need not to learn the ASCII table by heart, you need to know
> that A is 41h and count on - or forget about ASCII and choose any other
> suitable numerical representation of the alphabet.)
>
> Enciphering is done by XORing both values wordwise.
>
> 38 4E (the pin code) XOR
> 54 48 (the first two letters of the passphrase) yields
> -- --
> 6C 06 (that's what I write on my 'remember me' slip)
>
> For the next pin code the next two letters of the passphrase 'EM' will be
> used, and so on.
>
> Deciphering is done in the same fashion, ciphertext wordwise XOR
> passphrase letters.
>
> It does not look very safe at a first glance. It is not a theoretically
> safe one time pad at all, since the 'pad' is far from even looking
> random. On the other hand a dictionary attack will fail for the obvious
> reason that most words or expressions will yield possibly valid pin
> codes, i.e. four digit numbers in the (decimal) range 0 to 9999.
>
> A statistical attack might be promising if the opponent had a large
> enough sample of ciphertext to work with, but all he has is a slip of
> paper with typically< 20 Bytes of information.
>
> Do you see any feasible attack under the given assumptions?
>
>
> Matthias
>
>
Create a business card size list of random characters and use your mind
to link it in some way to the passwords / pins you need. Alternatively
work on fixing your (biological) memory or use a small personal data
organiser with a pin you won't forget
 |  Next  |  Last
Pages: 1 2 3 4
Prev: Question About Cryptographically Hashing a Hash (SHA-512), ThenHashing That Hash, Etc.
Next: Fallout 3 crypto?