Membership of this group has increased significantly, without a corresponding
rise in activity. This message outlines my own thoughts, based on what others
have said where appropriate, which I hope will encourage some discussion. I’m
assuming that everyone has cracked at least stages 1 and 2, and propose some
thoughts to prompt discussion on stages 3-10. Apologies to anyone who hasn’t
cracked the first two stages for their omission here – post a message to the
mailing list for everyone outlining your own thoughts and progress, and I’m sure
people will offer some advice or tips.

Stage 3:
Not, as I had previously thought, composed entirely of digrams – it has been
strongly hinted at that the ciphertext comprises a straightforward
monoalphabetic cipher with “a small” number of homophones. It seems evident that
these homophones are themselves digrams: the very high frequency of  X in the
ciphertext suggests that this character is used to create one or more of the
digrams/homophones.

Stage 4:
The instructions for how to solve this contained in the book are quite explicit,
and a number of people solved (myself included) this before stage 3.

Stage 5:
The only unsolved stage, the consensus is that the numbers refer to the relative
positions of letters (or words) in an as-yet-unknown text. Given that these
numbers are the same in all translations of “The Code Book” it seems unlikely
that the Singh’s text itself is the key text. (One down, millions to go...). The
keyword in the previous stage may provide a clue...

Stage 6:
Since no hint is given in the book as to the method of encryption, I can only
hazard a guess that it might be some form of Playfair cipher – Double Playfair
or Double Transposition Playfair, but not “simple” – the resulting ciphertext
could never be obtained using such encryption.

Stage 7:
With the ciphertext using only 8 letters, this HAS to revolve around
combinations of ciphertext characters representing one plaintext character –
doesn’t it? Is this a version of the ADFGVX cipher?

Stage 8:
Enigma. With German headings for the scramblers, plugboard, etc. Might this mean
that the plaintext is German? It appears that if you can crack the plugboard
settings, the plaintext will follow. I haven’t tried this stage yet, but my
first line of attack will be to check out the ‘repeated key’ mentioned in the
book. If nothing else, it’ll break me in to the process involved.

Stage 9:
I can’t think of anything that this might be, but it’s obviously a modern
(computer-age) method.

Stage 10:
My guess is that this is RSA, and that the “shorter message” is a public key.
Presumably, factoring this key will be of some use to decoding the message!

General: it occurs to me that if each stage contains an easily identifiable
codeword, then each plaintext might contain some variation of the phrase “the
codeword is...”, “... is the secret word”, in whatever language the plaintext
might be in.

Simon Singh has updated his web page,
http://www.4thestate.co.uk/cipherchallenge/
with a couple of buttons at the bottom for an "update" and for
reviews.
--
	 Jim Gillogly
	 28 Blotmath S.R. 1999, 18:12
	 12.19.6.12.16, 4 Cib 4 Ceh, Fourth Lord of Night

Jeff Hill wrote:
>        What about the letters with the special accent marks?  Can the
> marks be ignored or will those letters have to be given special symbols in
> the frequency tables?

In general it's my experience that at a first approximation one can
ignore accented and otherwise modifed letters, because those are usually
the lowest frequency letters.  When doing Scandinavian cryptograms, for
example, I normally focus on the consonants and let those vowels with
slashes and umlauts and circles take care of themselves.  Umlauted German
letters are much less common than plain ones, the cedilla isn't that
common in French, the n~ isn't very common in Spanish, and the ch,
kh, sh, and consonantal u are much less common in Esperanto than their
unaccented counterparts.  In short, leave those for the second pass.
If your cryptogram appears to have a metric butt-load of them, you've
probably done the segmentation wrong.  Don't forget the possibility of
other kinds of characters: numbers, punctuation, spaces and so on.  Some
table-based ciphers may use digraphs to represent not only those characters,
but complete words or endings such as -tion or "stop" or "the".  There's
some of this going on in the Zendian problem, for example.

--
	 Jim Gillogly
	 Highday, 26 Blotmath S.R. 1999, 03:13
	 12.19.6.12.14, 2 Ix 2 Ceh, Second Lord of Night

From: Jim Gillogly <jim@...>
<<  Lots of informal language is available in various Usenet newsgroups,
and intermediate language is available as transcribed literature --
Gutenberg is a good choice.  I've found that for solving German, for
example, the Kommunist Manifesto (available on the Web) was close enough to
zero me in. >>

         What about the letters with the special accent marks?  Can the
marks be ignored or will those letters have to be given special symbols in
the frequency tables?   I'm working on Stage 7 and I see more symbols than
I really expected once I got past the initial layer of security.

         -- Jeff

Fair enough.  I agree that it can be a pain, and I certainly wasn't
advocating enciphering everything, just specific questions or answers
within a message.  My problem (and an I the only one?) is that I tend
to view email via previews or paging down through the messages, so I
don't read the subject line often till after I've 'scanned' the text.
So a warning there may not be noticed until it is too late.  I'll just
have to be more careful!

Regards,

Brendan.

======================================
Brendan Smith  <brendan.smith@
                            bigfoot.com>

Encryption keys on request.
======================================

Memo from dlsmith on 16 November 1999, 10:48  Tuesday
To: CipherChallenge
cc:    (bcc: Brendan L Smith)
Subject: Re: [CipherChallenge] Re: An idea

From: "Dave Smith" <dlsmith@...>

Re: Obscuring some text to avoid revealing it to folks who don't want
to see it.

Jim Gillogly wrote:

>I'm not quite sure what the group is for if not to exchange info,
>but I have no objection to this kind of thing if others want to
>go this way....

and Sandy wrote:

>...the problem that this is the solution to is that of unwittingly
>receiving more information than you want. It would be simpler to
>include a warning in the subject line of any such message. The
>"temptation" to read it hasn't changed, and no-one can complain
>if they ignore the warning and read it anyway.

I agree with Jim; the group is to exchange info - but I have no
objection to a
simple method for obscuring information. I also agree with Sandy that a
warning
on the subject line is probably the simplest solution, if a solution is
needed
at all. Having to decipher our "normal" chatter would be a nuisance; and
deciding what to encipher could lead to troublesome complications. So,
if we
decide to obscure some information, let's avoid the nuisance factor by
keeping
all our group exchanges in plain text.

I'm enjoying this group. Thanks, Sandy.

--------------------------- ONElist Sponsor
----------------------------

The Mental Health Practitioner's Instant Resource Library for $5.99!
This concise 4-book set is a time-saving aid for basic clinical tasks.
(a $139.35 value)  Order NOW at Behavioral Science Book Service.
<a href=" http://clickme.onelist.com/ad/dblselect1 ">Click Here</a>

------------------------------------------------------------------------
Community email addresses:
   Post message: CipherChallenge@onelist.com
   Subscribe:    CipherChallenge-subscribe@onelist.com
   Unsubscribe:  CipherChallenge-unsubscribe@onelist.com
   List owner:   CipherChallenge-owner@onelist.com

Shortcut URL to this page:
   http://www.onelist.com/community/CipherChallenge



To: CipherChallenge@onelist.com@Internet
cc:  (bcc: CN=Brendan L Smith/OU=MEL-VI/OU=AU/OU=JHMarsh/O=MMC)
From: dlsmith@...

Re: Obscuring some text to avoid revealing it to folks who don't want to see it.

Jim Gillogly wrote:

>I'm not quite sure what the group is for if not to exchange info,
>but I have no objection to this kind of thing if others want to
>go this way....

and Sandy wrote:

>...the problem that this is the solution to is that of unwittingly
>receiving more information than you want. It would be simpler to
>include a warning in the subject line of any such message. The
>"temptation" to read it hasn't changed, and no-one can complain
>if they ignore the warning and read it anyway.

I agree with Jim; the group is to exchange info - but I have no objection to a
simple method for obscuring information. I also agree with Sandy that a warning
on the subject line is probably the simplest solution, if a solution is needed
at all. Having to decipher our "normal" chatter would be a nuisance; and
deciding what to encipher could lead to troublesome complications. So, if we
decide to obscure some information, let's avoid the nuisance factor by keeping
all our group exchanges in plain text.

I'm enjoying this group. Thanks, Sandy.

At 18:40 1999-11-15 +0000, you wrote:
>OK, I stand corrected.  Does Stage 2 match?

Yes.

>Stage 3 (Eng) starts: IXDVM

Match - which means I indeed was fooled by Stage 1 here. Oh well - thanks,
now I can have a much better go at it.

All the other stages match too, so it seems only one stage was translated
along with the book. This information (I think) is important enough to put
on the CipherChallenge web page. I hardly believe I am the first to arrive
at Stage 3 and find myself puzzled as to what language the cleartext is in.

Thanks for your help!

___/
_/

Rickard writes:
> Here are the first from stage 1, Swedish book:
>
> NJ ÄBRMNÖ RBW B RMLLM

OK, I stand corrected.  Does Stage 2 match?
Stage 3 (Eng) starts: IXDVM
Stage 4 (Eng) starts: KQOWE
Stage 5 (Eng) starts: 109 182
Stage 6 (Eng) starts: OCOYF
Stage 7 (Eng) starts: MCCMM
Stage 8 (Eng) starts: KJQPW (not counting the diagram)
Stage 9 (Eng) starts: MM5P7
Stage 10 (Eng) starts: 10052, then M.4#)

> Please give me some urls to those frequency lists!

Try the Crypto Drop Box; one entry there is:
http://www.und.nodak.edu/org/crypto/crypto/lanaki.crypt.class/docs/misc/xenocryp\
ts

>> but the easiest way to make them is to count your own.
> True, but whatever source I could come up with is bound to be quite
> technical ;)

Then you're not looking hard enough.  Lots of informal language is available
in various Usenet newsgroups, and intermediate language is available as
transcribed literature -- Gutenberg is a good choice.  I've found that
for solving German, for example, the Kommunist Manifesto (available on
the Web) was close enough to zero me in.  When I've solved Scandinavian
ciphers I've found that German trigraph tables are good enough to find
a solution in many cases.  Romance languages are similar enough that some
can be cracked using tables from another.

The bottom line is... just try it.  It might be close enough.  E is the
most frequent language in English, French and German (though not Latin
or Esperanto), and is often a good starting point.  N is usually pretty
high no matter what language.

--
	 Jim Gillogly
	 Mersday, 25 Blotmath S.R. 1999, 18:31
	 12.19.6.12.13, 1 Ben 1 Ceh, First Lord of Night

At 18:18 1999-11-15 +0000, you wrote:
>I'd be very surprised -- nay, astonished -- if the ciphers are changed
>from one edition to the next.  As a check, here are the first few words
>of Stage 2 in the English text:

Here are the first from stage 1, Swedish book:

NJ ÄBRMNÖ RBW B RMLLM

(Yes, there are two Swedish characters in that ciphertext)

>If this is the same as in your edition, then you can assume they're
>all the same, and the language or languages used in the English version
>will be the same as that or those used in the Swedish version.

I don't think checking stage 2 only sufficed.

>Frequency lists for various languages are available in lots of
>places on the Web and in books (such as Gaines' "Elementary Cryptanalysis"),

Please give me some urls to those frequency lists! I've been searching for
quite some time now, and I've only come up with a few low-res gifs that
really don't tell me much.

>but the easiest way to make them is to count your own.  It's a simple
>program to write, and you can download text from the Web or suck up a
>few Usenet groups to get appropriate language to count.

True, but whatever source I could come up with is bound to be quite
technical ;)

regards,
___/
_/

Rickard writes:
> * Are the cleartexts used for the ciphers in the same language as
> the book or not?

I'd be very surprised -- nay, astonished -- if the ciphers are changed
from one edition to the next.  As a check, here are the first few words
of Stage 2 in the English text:

     MHILY LZA ZBHL XBPZXBL MVYABUHL

If this is the same as in your edition, then you can assume they're
all the same, and the language or languages used in the English version
will be the same as that or those used in the Swedish version.

> The Swedish translation of the book only contains a character frequency
> list for Swedish, so _if_ a stage contains a cleartext in lets say English,
> Swedes would have to go find such a list for that language.

Frequency lists for various languages are available in lots of
places on the Web and in books (such as Gaines' "Elementary Cryptanalysis"),
but the easiest way to make them is to count your own.  It's a simple
program to write, and you can download text from the Web or suck up a
few Usenet groups to get appropriate language to count.

--
	 Jim Gillogly
	 Mersday, 25 Blotmath S.R. 1999, 18:11
	 12.19.6.12.13, 1 Ben 1 Ceh, First Lord of Night

Hi,

I've bought the Swedish translation of The Code Book. I've also began
solving the different stages, and I have a question for someone who knows:

* Are the cleartexts used for the ciphers in the same language as the book
or not?

I've done stage 1 and 2, so I know about those, but I ask the question in a
more general sense. Can I _assume_ something about the language used or is
it a part of the challenge to make educated guesses regarding the content?

The Swedish translation of the book only contains a character frequency
list for Swedish, so _if_ a stage contains a cleartext in lets say English,
Swedes would have to go find such a list for that language. I don't know if
the author/translaters have thought this through or not :)

Please don't spoil anything in your reply - I'm not after the price, I'm
after the challenge itself.

regards,
___/
_/

PS: What's the current speed record for stage 1&2? I did #1 in 25 minutes
and #2 in 18 .. pen&paper only. I have no idea whether that's fast or not
though ;)

Jim Gillogly wrote:
"I'm not quite sure what the group is for if not to exchange info,
but I have no objection to this kind of thing if others want to
go this way.  I don't have PeekBoo -- is it available for Linux?
Wouldn't it be more convenient to use either a more standard
system like PGP or a simple one like CipherSaber?"

I don't have PeekBoo either, and similarly have no objection to this method if
others agree, but I can't help thinking that it's unneccessary: the problem that
this is the solution to is that of unwittingly receiving more information than
you want. It would be simpler to include a warning in the subject line of any
such message. The "temptation" to read it hasn't changed, and no-one can
complain if they ignore the warning and read it anyway. (A reminder to new
members - please don't post solutions to _any_ of the stages to this list.)

The archives for this mailing list are now available to members only, btw.

About the problem of giving away too much:  how about if we use
...[fanfare]... encryption?  The thing is, that quite often your email
program gives you the message (say in a preview) and your eyes have
taken it in before your brain has said 'don't read this'!!  Or is that
just me ;-)?

For example, choose a simple program or method, such as PeekBoo (which
is excellent, IMHO, and perfectly suited to this sort of thing).  I
want to post a question about S4, but I know that even in the way I
write the question I would be giving too many hints to those who may
not want it.  So we have a standard of encrypting questions with the
key 'question', and answers with the key 'answer'.  I write the
following email:

A question on Stage 4 and language:

bgaqaaimxlTmW8d2GWlt3gAnZTq6gaaaabaaaaSszUWJcUZq{S17R5xdLsbhg1LaE3RT
tHOey0AMVDd6f

If you have done S4, you might read the message and post the following
answer:

bgaqaa9DbfCxYOyn0oy3AoB84GaudaaaabaaaaOJaKCLu4iUrw{O{a8IUZVga

Now, if I hadn't completed S4, and I saw the above email, I could
choose whether or not I decrypted the question and risked seeing hints
on that stage.  If I was at my wits end, I might do so, and having read
the question I could decide whether or not I wanted the answer to that
question.

The only caveat I would add:  we should not post complete solutions to
problems!  Even encrypted!  After all, it's meant to be a *challenge*!

Any thoughts?  Is this workable, or too much faffing around?  I would
be happy to do this if it makes conversations in the list more opaque.

Regards,

Brendan.

======================================
Brendan Smith  <brendan.smith@
                            bigfoot.com>

Encryption keys on request.
======================================

Congratulations, Frode, you have the balance right!  I appreciate your tips,
which can help a novice like me look in the right direction without giving
the game away!  I'll let you know how I get on!

I've only been to Zurich once, but it's a beautiful place, Switzerland.  I
guess winter is closing in now, though?

Regards,

Brendan.

========================================
Brendan Smith  <brendan.smith@
                              bigfoot.com

Encryption keys on request.
========================================

----- Original Message -----
From: Frode Weierud <Frode.Weierud@...>
To: cipherchallenge <CipherChallenge@onelist.com>
Sent: Thursday, 11 November 1999 8:03
Subject: Re: [CipherChallenge] Stage 3 and solutions


> From: Frode Weierud <Frode.Weierud@...>
>
> On 11 Nov 1999, Brendan L Smith wrote:
>
> > As someone else did, I was looking at digrams etc, but was always
> > stumped by
> >  what I now suspect was an incorrect assumption regarding the '*'.
> > Back to the
> >  drawing board.
>
> Hi all,
>
> I have been a quiet listener on this list as I am one of those old hands
> as you talk about and I feel it would be wrong to give to many hints on
> how a solution can be found. I would advice you though not to look to
> closely at what attracts your attention most easily like the '*'. Make
> instead a good count of all the binomes and hunt for the most frequent and
> the last frequent and let these guide your attack. Think about what
> they mean, are their frequencies unusual for such a message. And to give
> you yet another hint. Go back and read the book carefully and note what
> Singh says about these extreme cases. The rest will follow easily as soon
> as you have identified the language which you can also done from the
> statistics.
>
> I don't know what others feel but I think that to give more help than this
> would be unfair and unjust and also it would ruin an excellent chance of
> getting a better feeling for how to attack `tricky' ciphers such as to
> prepare you for the more difficult ones which follow.
>
> Best wishes,
> Frode
>
>
>         Frode Weierud           Phone  : +41 22 7674794
>         CERN, SL,               Fax    : +41 22 7679185
>         CH-1211 Geneva 23,      E-mail : Frode.Weierud@...
>         Switzerland             WWW    : http://home.cern.ch/~frode
>
> > Community email addresses:
>   Post message: CipherChallenge@onelist.com
>   Subscribe:    CipherChallenge-subscribe@onelist.com
>   Unsubscribe:  CipherChallenge-unsubscribe@onelist.com
>   List owner:   CipherChallenge-owner@onelist.com
>
> Shortcut URL to this page:
>   http://www.onelist.com/community/CipherChallenge
>

On 11 Nov 1999, Brendan L Smith wrote:

> As someone else did, I was looking at digrams etc, but was always
> stumped by
>  what I now suspect was an incorrect assumption regarding the '*'.
> Back to the
>  drawing board.

Hi all,

I have been a quiet listener on this list as I am one of those old hands
as you talk about and I feel it would be wrong to give to many hints on
how a solution can be found. I would advice you though not to look to
closely at what attracts your attention most easily like the '*'. Make
instead a good count of all the binomes and hunt for the most frequent and
the last frequent and let these guide your attack. Think about what
they mean, are their frequencies unusual for such a message. And to give
you yet another hint. Go back and read the book carefully and note what
Singh says about these extreme cases. The rest will follow easily as soon
as you have identified the language which you can also done from the
statistics.

I don't know what others feel but I think that to give more help than this
would be unfair and unjust and also it would ruin an excellent chance of
getting a better feeling for how to attack `tricky' ciphers such as to
prepare you for the more difficult ones which follow.

Best wishes,
	 Frode


         Frode Weierud           Phone  : +41 22 7674794
         CERN, SL,               Fax    : +41 22 7679185
         CH-1211 Geneva 23,      E-mail : Frode.Weierud@...
         Switzerland             WWW    : http://home.cern.ch/~frode

Hello everyone.  I just found this list via sci.crypt.  I am a novice
at this
  sort of thing, but loved the book and set about the challenge with
enthusiasm!

Like several others here, I have solved 1,2, and 4 in fairly short
order.  But I
  am finding Stage 3 a pain, to be blunt!  I certainly don't think that
it is
  correctly in place in order of difficulty!  Of course, I am probably
missing
  something....

As someone else did, I was looking at digrams etc, but was always
stumped by
  what I now suspect was an incorrect assumption regarding the '*'.
Back to the
  drawing board.

I agree that solutions should not be posted here.  After all, the
challenge of
  solving the problems is the main reason for me doing it, as I always
assumed
  that someone with years of experience would beat me to the dosh!  But
this
  raises the question:  what is the aim of discussing it here, as it may
be a tip
  or inadvertant comment that leads you swiftly to the solution.  Whilst
it may
  be briefly good to get the solution, wouldn't I feel better to have
found any
  shortcut myself?

What do you all think?

Regards,

Brendan.

======================================
Brendan Smith  <brendan.smith@
                            bigfoot.com>

Encryption keys on request.
======================================

Dave Smith wrote:
“Sandy, don't listen to that inner child! Have fun and intellectual satisfaction
by solving as many stages as you can. You might even win 1000 pounds if you
psych-out Simon Singh's text and method for stage #5. Applying it should be
easy; discovering what he was thinking is the hard part. And have fun with the
Nova contest too ;-) Regards, Dave”

Don’t worry: I enjoy doing these things for their own sake, and am well aware
that my inner child is a spoiled brat who speaks with his mouth full. I have to
admit, though, that my satisfaction would be tempered somewhat if I knew that
the challenge had been completed before I got too far into it – a bit like doing
yesterday’s crossword. In which respect, I suppose, the positioning of the most
intractable ciphertext at stage 5 would be an excellent idea, if it were
intended.

I’m thinking of making the archives for this mailing list for members only: at
the moment they’re available for view to non-members at the Onelist site. I’m
thinking of doing this because I’m aware that the majority of mailing list
members (if that isn’t to grand a word for n<20) don’t contribute, and might be
put off by the fact that the archives are public. If I don’t receive any
objections by the end of the week, I’ll do it. Non-posters should not read this
as any sort of slight: you don’t have to post if you don’t want to.

Sandy

Sandy wrote:

>... I got in touch with my inner child about this, and he whined
>"That's not FAIR!" Might as well concentrate on the Nova Online
>contest then...


and Jim wrote:

>... the first person to get Stage Five will be in the lead, no
>matter how many other problems have been solved above it.  If
>s/he keeps her lips sealed about the key-text, there's a good
>change s/he can keep on top for at least the first-level prize
>(1000 pounds) and perhaps even the top.

which means:

Sandy, don't listen to that inner child! Have fun and intellectual
satisfaction by solving as many stages as you can. You might even
win 1000 pounds if you psych-out Simon Singh's text and method for
stage #5. Applying it should be easy; discovering what he was
thinking is the hard part.

And have fun with the Nova contest too ;-)

Regards,
Dave

> That would be the same Jim Gillogly who designed a "crib-free" attack on
> the Enigma cipher, I suppose? I got in touch with my inner child about
> this, and he whined "That's not FAIR!" Might as well concentrate on the
> Nova Online contest then...

At least one other solver has all but #5, #9 and #10.  However, the first
person to get Stage Five will be in the lead, no matter how many other
problems have been solved above it.  If s/he keeps her lips sealed about
the key-text, there's a good change s/he can keep on top for at least the
first-level prize (1000 pounds) and perhaps even the top.

Good luck with the Nova contest -- that program airs tonight.

--
	 Jim Gillogly
	 Highday, 19 Blotmath S.R. 1999, 17:34
	 12.19.6.12.7, 8 Manik 15 Zac, Fourth Lord of Night

Dave Smith wrote:
"When I spoke to Jim Gillogly a few weeks ago he had solved all but #5
and #10."

That would be the same Jim Gillogly who designed a "crib-free" attack on the
Enigma cipher, I suppose? I got in touch with my inner child about this, and he
whined "That's not FAIR!" Might as well concentrate on the Nova Online contest
then...

btw, thanks for explaining the flat frequency thing.


Sandy

>sandy@... wrote:

>>There is no requirement that the frequency counts be flat.
>
>Could you explain that? I don't know what that means.

The early discussion of homophonic ciphers dealt with assigning homophones in
proportion to the frequency of plaintext letters (13 homophones for e, 9 for t,
8 for a etc.) for a total of about 100 homophones to cover 26 English
characters. Frequency counts of the homophones would be "flat", meaning that
they would each have about the same number rather than show the normal peaks and
lows of normal English frequencies.
That discission also led to an assumption of digraphic assignments in order to
get 100 or more different homophones.

I attempted to argue that stage #3 was too early in the scheme of things for a
cipher of that complexity, especially given that #2 and #4 are significantly
easier to solve than a "flat" homophonic cipher would be.
And note that the frequency distribution for #3 is definitely not "flat".

Stage #3 is a pretty difficult cipher for a novice. It has some nice twists, and
solving them is satisfying. It is, indeed monoalphabetic. It does employ
homophones. But it does not employ digraphic means to identify lots of
homophones. And the internet search engines can help, if you're not fluent in
the plaintext language or the subject matter of the enciphered quotation.

>Regarding the perceived difficulty scale, I'd gave to agree with Jeffrey: it's
misleading, as it's only based on Singh's idea of what's difficult. If you're
incapable of recognising a foreign language when you see it, for example, then
you might find some early stages impossible.

Yeah, the difficulty scale is pretty subjective.

>Stages 6 to 9 solved already? Oh dear.

When I spoke to Jim Gillogly a few weeks ago he had solved all but #5 and #10.
Judging buy his questions in this group, he still hasn't found the key to #5. I
wonder if he'll get the last 97 characters of the KRYPTOS cipher before he finds
it ;-)

Regards,
Dave

dlsmith wrote:
>There is no requirement that the frequency counts be flat.

Could you explain that? I don't know what that means.

Regarding the perceived difficulty scale, I'd gave to agree with Jeffrey: it's
misleading, as it's only based on Singh's idea of what's difficult. If you're
incapable of recognising a foreign language when you see it, for example, then
you might find some early stages impossible.

Stages 6 to 9 solved already? Oh dear.

<< From: dlsmith@... >>
<< In the scheme of difficulty, #3 should be more difficult than #2 and
less difficult than #4, both of which use single-letter substitution
(monoalphabetic in #2 and polyalphabetic in #4).  So it's doubtful that #3
would employ digrams, because to do so would imply a level of difficulty
too great for its position on the scale of "stages". >>

     The "scale of stages" is misleading since Singh gives detailed
instructions on how to solve #4.   In fact, he makes the solution more work
than is really required in this particular case.   Also, #6 thru #9 have
all been solved, but #5 has not.   Deciphering #5 is simple once the
keytext is known, but finding the keytext is quite a challenge.

         -- JH

I too, did stages 1,2 & 4 before finishing stage 3.

Without giving away too much, let me steer you away from a rigid
interpretation of homophonic ciphers.

When Singh says #3 is a monoalphabetic cipher with homophones,
only two homophones would be required (because he uses the plural),
and there's no implication of the number of letters to which the
homophones apply.

Since the cipher has only 27 different characters, since it's
monoalphabetic, and since in text of this length there should
be only a few unused letters, the number of homophones should
be relatively small. There is no requirement that the frequency
counts be flat.

In the scheme of difficulty, #3 should be more difficult than #2
and less difficult than #4, both of which use single-letter
substitution (monoalphabetic in #2 and polyalphabetic in #4).
So it's doubtful that #3 would employ digrams, because to do so
would imply a level of difficulty too great for its position on
the scale of "stages".

Simon Birt wrote in sci.crypt
> Am I then correct in assuming that each of our n hundred symbols (be they
> digrams trigrams or whatever) should appear an equal number of times in
the
> ciphertext?

    If you go by what Singh says, then the number of homonyms for each
letter is proportional to the letter's frequency, so that if we decide to
have a hundred homonyms total for all letters, then 'E' would have 12 to 13
homonyms because it is usually 12% to 13% of English text (more if French,
etc.)    If the person who enciphers the cryptogram is careful, he can see
to that each of the homonyms is used about the same number of times, but he
might have to make checkmarks after each homonym to be sure he isn't using
one more than the others.  Ideally, this is how it would be done, but few
people care to be that precise, so some homonyms get used more than others.

<< sandy@xxxxxx.xx
<< 3. This is a competition, and if it was too hard, no one would bother,
and
then it wouldn't generate any positive publicity, and then more books
wouldn't be sold. I think we'll find that things aren't as hard as they
might have been if the cryptography in each stage was used to the limits
of its powers. >>

     Actually, it is a hard contest.  The ciphers start easy and become
progressively harder.   However, Singh's book doesn't cover each cipher in
enough depth to enable you solve them all.  You'll have to read other books
for that.  Singh's book is long on history and short on cryptanalysis.

<< From: "Simon Birt" <simon.birt@xxx.xx.xxx
<< I've now done stages 1,2 & 4, although I've been cheating horribly by
writing myself little Java utilities for frequency analysis etc. >>

    How is that cheating?  It sounds like the smart thing to do.

<< I am, however, stuck on stage 3. >>
<< I've tried a number of schemes, and the closest I've got to the 1st
requirement is by assuming that each ciphertext symbol is a digram - this
yields 198 discrete digrams. However, the distribution is not constant - it
varies between 1 occurrence (for a good many of the digrams) and 28
occurrences (for QX). Which leads me to believe that there is a more subtle
scheme than mere digrams involved. >>
<< Does anyone have any tips for this? >>

    Singh doesn't always practice what he preaches, so it pays to look for
alternatives.  The first thing to notice is that with 198 homonyms, this
cipher would be as hard to solve as the Beale ciphers (assuming they are
genuine).  The second thing to notice is that two people solved it within a
week of the book being published, so it can't be all that hard.   So,
there's a trick to it (several tricks in fact), but that's part of the
challenge.  The first thing you have to do is reduce the number of
homonyms.

         -- JH

<< From: Jim Gillogly <jim@...>

<< Just to add some content, I'd like to put together a list of
possible ways Stage 5 might work.  It's evidently a Beale-type
cipher, based on where it fits in the time-line and its obvious
structure.  >>

     So the time line for the stages is as valid as it appears?

<< 1) It's precisely Beale style, i.e. they're word numbers in a
particular key-text, and the plaintext is the first letter of
the numbered word.

    It's not very convenient to do it this way because you really have to
reach to get some of the letters you need (i.e. count lots of words).  You
might also have to use the Beale approach and take a few letters from
within words.  You need an 'x', but no words end in 'x', so you take the
'x' from the word 'sexes'.   Works okay, so long as it doesn't happen very
often, and at worst it looks like a typo.

<< 2) Uses full words instead of letters.

     Like the Amazon.com contest?  It's not very convenient to to do it this
way either if you have something very precise that you want to say.  The
Amazon 'messages' were rather lame.

<< 3) Counts a book or section in reverse order.

     Certainly possible.  Or number only every Nth word or letter.

<< 4) They're letter numbers in a shorter text -- i.e. letters are
numbered instead of complete words.

     In my experience, this would be the most convenient way to create a
Beale-type cipher, drawing from a source with just 200 to 300 letters.   Of
course, "convenience" might not have been foremost among Singh's
priorities.

<< If they're full words, then as we've seen with the Beale ciphers
it makes a lot of difference exactly how certain things such as
hyphens are handled.  A hyphenated word that's counted the wrong
way could throw off the count. >>

     What if the text contains numbers (expressed in digits, 0-9), are these
counted as words or ignored?   Example:  The 21 goats were led to pasture.

     Since being multi-linguistic is one of the unwritten requirements of
this contest,  I suppose we have to consider the possiblity that the
cleartext is in something other than English.

     -- JH

Looks to me as though the onelist architecture allows archiving
messages.  Is this something set up at list initialization time?

Just to add some content, I'd like to put together a list of
possible ways Stage 5 might work.  It's evidently a Beale-type
cipher, based on where it fits in the time-line and its obvious
structure.  Several possibilities:

1) It's precisely Beale style, i.e. they're word numbers in a
particular key-text, and the plaintext is the first letter of
the numbered word.

2) Uses full words instead of letters.

3) Counts a book or section in reverse order.

4) They're letter numbers in a shorter text -- i.e. letters are
numbered instead of complete words.

If they're full words, then as we've seen with the Beale ciphers
it makes a lot of difference exactly how certain things such as
hyphens are handled.  A hyphenated word that's counted the wrong
way could throw off the count.

Any other likely ideas?
--
	 Jim Gillogly
	 Trewesday, 16 Blotmath S.R. 1999, 01:10
	 12.19.6.12.4, 5 Kan 12 Zac, First Lord of Night

OK, just to kick things off in our wee community of 4 :-)

I've now done stages 1,2 & 4, although I've been cheating horribly by
writing myself little Java utilities for frequency analysis etc. I am,
however, stuck on stage 3.

From my understanding of the book and from the recent usenet discussion, it
seems that the ciphertext should contain a number of homophones such that
the total number of discrete homophones should be roughly a multiple of 100,
and that each homophone should appear roughly an equal number of times.

I've tried a number of schemes, and the closest I've got to the 1st
requirement is by assuming that each ciphertext symbol is a digram - this
yields 198 discrete digrams. However, the distribution is not constant - it
varies between 1 occurrence (for a good many of the digrams) and 28
occurrences (for QX). Which leads me to believe that there is a more subtle
scheme than mere digrams involved.

Does anyone have any tips for this?

Simon

Simon Birt wrote in sci.crypt
> Am I then correct in assuming that each of our n hundred symbols (be they
> digrams trigrams or whatever) should appear an equal number of times in the
> ciphertext?

I thought I'd better practice what I preach, so I'm using the list to reply to
the newsgroup item.

The answer is no. Because:

1. It is highly unlikely that the plaintext letter frequencies will match those
in the averaged-out frequency table; the plaintext may not even be in the same
language as the frequency table (if you see what I mean).

2. The distribution of each plaintext letter's various homophones is supposedly
random, which is not the same as "even"

and most importantly

3. This is a competition, and if it was too hard, no one would bother, and then
it wouldn't generate any positive publicity, and then more books wouldn't be
sold. I think we'll find that things aren't as hard as they might have been if
the cryptography in each stage was used to the limits of its powers.

For the record, I haven't completed this stage, but I have completed stages 1,2,
and 4. Also, I'd like to propose that we don't post solutions on this list.

This mailing list aims to provide a focus for discussions about the Cipher
Challenge. It is open to anyone (but preferably those who've read the book), and
anyone who feels that they're too "Newbie" to post to established newsgroups and
the like is more than welcome. There's no FAQ here, but that's only because
no-ones asked any questions yet...

  - Sandy