My current, disorganisedish and somedeal unoriginal, thoughts.

* Someone should've double-checked the transcription by now, if we're taking
this seriously -- the very first thing to do is to have a reliable (or as
reliable as possible) transcript to work from, nonne?  

Thankfully one travis on noisebridge has done a second transcription.   
 
https://www.noisebridge.net/pipermail/noisebridge-discuss/2010-June/014630.html
Put in the same language as the first one using
  $ tr a-z1 'VFQPABDWXIMTEGJON$CRYKSZLUH'
this is
VFQPAVAVFQBADAWXAIVWAIMTAEIVG
FQGJOGAINAI$FFQGCFQFQGJG
QIRGITFIFQYANEJKSVZQOFQA
QMTFEFZPFTFQKEFEVFQGDAVQAI
SVEFQCGAWQOWEANJ$JFQYANEJ
FIANOPJFFVAWENGFTJFVFQAVGAFOA
FQYANEIFQOAFQCIFQFLEKQFAOCEA
DFKVGTIFQOQUAIJGSEVDNAJF$I
DAOSETVIWQAJFGIHQFQSVEJFIF
FQGWENJGIUAEFQGOHZZAIFQ$AI
FQYNEJFQOASGFOEFIFZWTOFQNSFV
VFIFWQAAUAITFIFQPQJVJBJFQQUJFV
FQGJGZAIFVFQGOXNQAIGIWIFZEJG
FQOASFAGFGFOODETWIXEAK
STAEJNAFQGPFNXJWQAFDEFEJF
FQFGIAIFQOAJATSVEIFQSAJFGYFSQ
VFQVLFEFOFQGWIAJVHYVZFFQSVFAE
FQBAOWFIAGIJCGAWSVFAEFQSVQJ
VFQOATF$IFQQJDOPQSQFQODAT
QKSOVZQOFQAJBJFQQUJFQGSOVTJ
FQFIVFECVTFI
FQGVRVWJVFQYANEJFQOAFIFFSOTTJF
OPAXATJVPOEFG
FQDASGPGSHQNFIFQGLEVWA
FQEPAG
TDGFQQIFQEPGI

There are several single-character deviations, mostly some Js in the first
are Is in the second, and some Bs $s, and some Ms Ss.  On line 11, a Q is an
O.  Line 13, a Q is a Z.  Line 17, a Z is a Y.  Line 18, a Q is a G.  And
maybe others I didn't see (wish diff(1) worked character-by-character, or at
least could refrain from displaying multiple adjacent differing lines in
chunks.)

* That said, I think left-to-right English should be discarded posthaste as
a hypothesis.  Decrypto, which Sai linked at some point, falls flat on its
face, not even finding plausible word breaks; it has no trouble with normal
English messages five times shorter than this.  Plus, if it were, the list
would've gotten a way in by now.  In any event

* the fact that exactly 26 symbols occur ***isn't particularly good
evidence*** for it using the basic Roman alphabet.  It's 630 characters.  If
it were English you'd expect only 0.4 of a Z and 0.6 of a Q and so on in a
text of that length.  My gut feeling (I haven't worked the numbers) is that
if only 25 distinct characters occurred that would be better evidence for it
being English, and 26 I'd put weakly in favour of Arabic.  

[Did you know that frequency cryptanalysis was invented in Arabic? 
Al-Kindi:, 9th c.]

* 630 characters is still too short to hope to solve by just lining up the
frequencies and substuting top for top, second for second, etc, except maybe
as something to stare at and get an inspiration.  

* I don't know enough Arabic to do much in that direction.  The suggestion
that FQ is alif-lam is worth evaluation by someone who does know better. 
The four FQQ would have to be word boundaries, I guess, unless he's written
shadda.

* It might indeed be that some of the characters are nonletters but are
punctuation or the like.  I'm not really very good at those.  

On Tue, 15 Jun 2010 17:48:41 -0700, Gary Shannon <[log in to unmask]> wrote:
>Given the left-justified original Arabic seems less likely than
>English as the original source.

Maybe marginally marginally so.  Not significantly.  Someone who knows
Arabic and English both will have noticed this difference and it'll be one
of the first things they'd play with, in my estimation.  

In answer to Lee, I'm inclined against the image being flipped top-to-bottom
given where the signature-looking block is.  I'd think flipped left-to-right
unlikely too unless Arabic conventions are the left-to-right reflections of
ours as regard signatures.  

Hm.  If that is actually a signature, can it be matched in any plausible way
against the guy's name, or the name of anything he represents, or ...?  Can
the short line before it be any traditional closing formula?

On Tue, 15 Jun 2010 18:41:28 -0400, Karen Badham <[log in to unmask]> wrote:
>I'm skeptical about it being used for English and Arabic interchangeably.
>The phonology is pretty different,

Jen had it right: nearly everyone on earth is entirely naive about
phonology; no reason to think this guy is not among them.  

My best guess at how he's pulled off the "universality" is by rendering
English into Arabic phonology, via a method which could be more or less ad
hoc (and without a doubt collapses any pair of phones that doesn't sound
distinct to an Arab), and then writing that.  

On general principle it would fit nicely as an illustration if it were
_multilingual_, but I think the uniformity of things like the frequency of
FQ is against that.  Still, might be worth more investigation.

Alex