[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: VMs: Crypto timeline needed (+ some thoughts)

Hi Petr,

At 23:05 21/12/2003 +0100, Petr Kazil wrote: 
I see that we need to make a timeline for VMS related cryptography along the
lines of http://web.bham.ac.uk/G.Landini/evmt/vtimel.htm Th discussion is
getting a bit muddy, but the question is stil clear:

What was the common level of cryptography around the time we think the VMS
was written?

Which might be:
- between 1400 -1500 if straight encrytion is assumed (based on the
illustrations)
- between 1500 -1600 if we assume a hoax by Kelly

I guess you're talking about a technological timeline for cryptography here...

The common level of Quattrocento cryptography (ie, pre-Trithemius, and even arguably pre-Alberti [as his ideas did not circulate for a long time]) can be observed (for Northern Italian codes, at least) directly in the Tranchedino cipher ledger.

While I'd like to examine the 1440 cipher ledger from Urbino as mentioned by Sacco (but which is probably buried deep in a buste in Florence) for parallels with the VMs' cipherbet, I don't expect to see any technological differences from Tranchedino.

There is a cut-off point circa (say) 1475 when numbers / number codes began to "invade" ciphers in the Milanese ledger - and in fact, about the same time the date on each page starts to get written in Arabic numerals (rather than in Roman numerals). So, from 1475 I'd say that there's a strong Arabic number influence on Milanese ciphers.

By way of contrast, Renaissance cryptography (1500-1600) seems to have been characterised by three main things:
(1) the development of more secure (but less robust) techniques, like polyalphabeticity
(2) the development of less secure (but print-centric) techniques, like Cardan grilles
(3) the general avoidance of all those techniques by most practitioners

IMO, Renaissance cryptography arose more from linear, sequential "printer thinking" than from Quattrocento cryptography - that is, in the way that printers reduce a page of text into a linear sequence of movable type fragments, so did Renaissance "intelligencers" begin to deconstruct the idea of text into a sequence of letters or shapes, where the position in the sequence was reliable and reproducible and the position on the page was deterministic.

This conceptual revolution enabled polyalphabeticity and Cardan grilles (for example), as they rely on a consistent sequential cycle of positions and consistent lines with print-like regularity to the spacing (respectively).

My opinion on the VMs is this: that the construction of its alphabet was not informed by the idea of Arabic numerals (the o 4 8 9 characters seem to be derived from the handful of Tironian notae "fossils" still in use circa 1400-1450), and so fits the 1400-1475 era of cryptography. However, the VMs' complexity and subtle structure would seem to point to a later date than to an earlier date: and if the author of Tristano Sforza's (far simpler) personal cipher (from 1450) also designed the VMs' coding system, then we should look to a later date than 1450.

Therefore, from the technological timeline plus comparisons with Tristano Sforza's 1450 cipher, I would date the VMs to 1455-1475.

I was wondering how a Fourier Transform of the VMS would look like on a "per
character" basis. So a FFT of the VMS where <a = 1> and <every other
character = 0>. And then for every character or combination like <iin>. This
might show the difference between real characters and nulls. I've been
looking for FFT programs, but found only some that do a FFT of a .wav file.
An Excel macro would be ideal but I had no luck with finding that that
either.

But probably this has been tried already somewhere in the VMS archives?

FFTs would be an interesting choice of technique for looking at Renaissance codes and later, as any kind of periodicity (such as for polyalphabetic) ~should~ show up in some (though perhaps muted) form within an FFT. However, AIUI modern cryptologists instead use techniques like the Kasiski Method to determine periodicity in a text - these seem more statistically reliable.

It ~might~ be that the VMs uses a verbose cipher to hide its polyalphabeticity... here's an idea for an interesting cracking algorithm, I don't know what the result would be:-
(1) For all possible pairifications of the VMs (ie <qo> --> single symbol, etc)
(2) *** Use the Kasiski Method to evaluate each pairification's affinity with periodicity
(3) Globally replace the pair with the highest chance of periodic output
(4) Repeat until no replacements give an improved periodicity metric over the existing text.

Just a thought! :-)

Cheers, .....Nick Pelling..... 


______________________________________________________________________
To unsubscribe, send mail to majordomo@xxxxxxxxxxx with a body saying:
unsubscribe vms-list