![]() | This is an archive of past discussions. Do not edit the contents of this page. If you wish to start a new discussion or revive an old one, please do so on the current talk page. |
Archive 1 | Archive 2 | Archive 3 | → | Archive 5 |
This question enters into the general frame established by the mathematician Claude Shannon, in his paper "Communication Theory of Secrecy Systems". Cl. Shannon has established a general formula giving the "Unicity Distance". If the length of a text is shorter than the "Unicity Distance", the text cannot be deciphered, because there is more than one solution. If it is longer, the text may be deciphered when one starts from correct starting hypotheses concerning the language, the code used, etc. As another mathematician, J. Faucounau, wrote, the main factor in this formula is, beside the code, what Cl. Shannon has called "Entropy", which depends upon the language and the text itself. I quote this author : "Le facteur que Cl. Shannon a appelé "Entropy", qui depend du langage et du texte lui-même, est beaucoup plus décisif que la longueur du texte elle-même. Il est ainsi possible de déchiffrer un "texte" d'une longueur aussi courte que 5 ou 6 signes, écrit dans un système de simple substitution (= système du disque de Phaistos), lorsque l'on part, bien entendu, comme le suppose Cl. Shannon, des "bonnes hypothèses". En voici un exemple : Texte : 5-14/11-6-14-11. "Bonnes hypothèses" : a)- un chiffre = une lettre (et réciproquement) b)- le langage est du français c)- il s'agit d'un objet d)- le signe / est un séparateur de mots." If one accepts for the Phaistos Disk a "reasonable figure" for its entropy, the Shannon's Formula leads to a figure of c. 225 signs for the "Unicity Distance", inferior to the real length of the Disk (241 signs). The text of the disk can therefore be deciphered, when one starts from the "good hypotheses". (User 80.90.57.154).
What kind of a linguist are you, for confusing "phonetic evolution" with "mixture" !?.. I have written about "the expected intermediary" between "Common Greek", i.e. the ancestor of all the Greek dialects (= a language spoken probably c.3500 BC), and the "Classical Ionic", known from IXth Century BC inscriptions... It is your comparison which is absurd !... What one can say is that Old French, a language spoken in France during the Middle Age, is "the expected intermediary" between Latin and Modern French. And even if Old French is often closer to Spanish than Modern French is (for instance : Latin <cappa>, Old French <chapa>, French <chape> v./ Spanish <capa>), Spanish has nothing to do with the phonetic evolution which has led from Latin to Modern French, Old French being the intermediary !... (User 80.90.57.154).
"Archaeological Decipherment" by E.J. Barber, page 204, with a correct calculation of the "Unicity Distance", but a slightly inaccurate conclusion : it is said that the solution would be "highly suspicious". In fact, it has just to be proven, in a way or in another (Application of the solution to a second text, or verification of all its consequences), because the length of the text is too close from the "Unicity Distance". In other words : the length is big enough for finding a unique solution, but not big enough for being absolutely sure that this unique solution is the good one. Satisfied now ? (User 80.90.57.154).
It is necessary to know a language perfectly to use (the Shannon's formula). No !.. I am surprised that you don't understand that the Shannon's Formula gives the minimum length for a text to be deciphered if (a big if !) the decipherer has found - by chance or by any other means- the "good hypotheses". You don't need to know the language for making an estimate of the Unicity Distance. You have just to suppose "reasonable hypotheses" concerning the "Entropy". (Otherwise, you will find "exceptional figures", like for instance the 6-signs long text J.Faucounau has given as an example. See hereabove). All searchers (Cl. Shannon, J.Faucounau, E.Barber, Y.Duhoux, etc.) agree that, in the case of the Phaistos Disk, the "reasonable Unicity Distance" is c. 225 signs, i.e. inferior to the real figure. My theory, as you call it, is therefore accepted by all mathematicians. It doesn't say that the text's length is such that anyone can decipher the disk's text. It just say that the text's length is not a problem, as many people - who are not mathematicians- believe, including yourself. The real problem is in the need of checking the solution found (User 80.90.57.154).
a)- I didn't change my position concerning the language. I have always talked about the necessity to start from "the good hypotheses" to reach the correct solution, whatever the value of the Unicity Distance !.. Language is one of them. It can be known (as in the case of Proto-Ionic), or not (as in the case of Etruscan). b)- It depends of what you call "superficially check". So, I can tell that the deciphering attempts supposing the Phaistos Disk to be read from the center outwards are surely wrong. No need for a more complete examination !.. c)- No !.. The time passed in the search for proofs is significative of the difficulty of this indispensable step in the case of the Phaistos Disk. Convincing people is another problem, and History of Science has shown that it took sometimes 30 years or more for a correct theory to be adopted... So, I maintain my redaction. ( 80.90.39.25 12:21, 25 October 2005 (UTC))
It's really tiring to discuss with Kadmos who is unable to understand Shannon's work !.. The more as Gbrunner is taking his interventions to censor the Point of View he doesn't like ... ironically in the name of NPOV !!! The nonsense is yours. I quote the text you cited (emphasis is mine) : Shannon defined the Unicity Distance ...as an approximation of the amount of ciphertext such that the sum of real information (entropy) in the corresponding plaintext plus the entropy of the encryption key equals the number of ciphertext bits used. In other words, Shannon's Entropy depends a)- first and mainly of the Code b)- secondly of the language c)- thirdly of the plaintext itself. The more the ciphertext will be close to a randomly series, the greater will be the Unicity Distance. This is why a code which randomizes the ciphertext better than another will be more secure, and why a "Simple Substitution Code" (case of the Phaistos Disk and of my example) is the worse in regard to secrecy, Shannon's main concern, because it keeps the peculiarities of the language and of the text. In the example I've given, it's clear that the 6 signs are not taken at random : two are occurring twice, what allows the decipherment. ( 80.90.39.25, Dec. 2, 2005)
Sorry if my answers to Kadmos seem to you as personal attacks! I will try to lower their tone, but one must understand my irritation in front of Kadmos' attitude, using words like "Nonsense!" or "Absurd!" when it's he who is wrong. He was wrong, for instance, saying that Shannon's Formula is of no use in the case of the Phaistos Disk's decipherment, when it has been used by several scholars ; he was wrong saying that the Unicity Distance is not the point , or misunderstanding the question of Unicity of the Solution.
BTW, you seem to have the same problem concerning this question, so let me be more explicit about it : Shannon was dealing with cryptography. So, he was implicitely supposing that the encryption was corresponding to an encrypted text, coming from a plaintext in a known language, and he could talk about a unique solution. If you abandon this implicit condition, the notion of a unique solution vanishes, or better said, has to be modified. For instance, during a war, a coded message is decrypted as Order to attack tomorrow. Well, a fool can also say : "It is not a message. It's the numbers the ennemy intends to play at the roulette in Las Vegas". And there is no more a unique solution because the fool's solution is theorically possible... So, when one deals with something which may not be a text, and of which the chosen language may not be the good one, like this is the case with the Phaistos Disk, the notion of unique solution has to be understood : unique solution if the used hypotheses are the good ones, and one has to verify the solution discovered , for instance, like in the (pretty stupid) example I gave hereabove it can be done : By verifying the consequences of the decipherment (verifying that the ennemy prepares an attack, or even waiting until tomorrow, to be sure that the decipherment was good). To go back to your remark, one has to understand : From the Shannon's Formula, the length is sufficient if one has to deal here with some kind of syllabic text, with a Simple Substitution Code concerning the phonetic values, and with a not-exceptional Entropy (= Reasonable assumptions). As a consequence, if the other starting hypotheses (in particular concerning the language) are '"the good ones", the solution can be found (= Unicity of the Solution). But the text's length is not sufficient to warranty that the chosen hypotheses were surely the good ones. Therefore, a verification is needed by examining the consequences of the solution found. I hope to have clarify the things, this time... ( 80.90.39.25 Dec. 5, 2005)
It's really discouraging to discuss with you ! a)- I didn't change my position. Your criticism is nonsense. As this text can be deciphered (because of its abnormal entropy), it means that the Unicity Distance is 6 in this particular case (a thing that you don't seem to understand !!!) b)- the need for verification : of course, it has to do with the Shannon's Unicity Distance ! (What you don't seem to understand !!!) c)- Use of the Shannon's Formula . Why would it be of no use as you pretend ??? What hinders anyone to suppose (rightly or wrongly, it doesn't matter!) the language of the Phaistos Disk to be a known language, and apply then the Shannon's Formula??? d)-the example I gave hereabove : Of course, there are several solutions ! And for the reason you give, concerning the Unicity Distance in the case one supposes it's a date ! Because the Unicity Distance = 6 ONLY IF the "good hypotheses" that I mentioned (It's a text, written in French, etc.) are taken in consideration. Supposing that it's a date doesn't respect this condition ! (a thing you seem unable to understand!!!) e)-How will you decide whether a solution is valid or not : please read my preceding paragraph (with an example!) in this ridiculous discussion : by its consequences ! As simple as that !.. ( 80.90.39.25, 20.15, Dec. 5, 2005)
![]() | This is an archive of past discussions. Do not edit the contents of this page. If you wish to start a new discussion or revive an old one, please do so on the current talk page. |
Archive 1 | Archive 2 | Archive 3 | → | Archive 5 |
This question enters into the general frame established by the mathematician Claude Shannon, in his paper "Communication Theory of Secrecy Systems". Cl. Shannon has established a general formula giving the "Unicity Distance". If the length of a text is shorter than the "Unicity Distance", the text cannot be deciphered, because there is more than one solution. If it is longer, the text may be deciphered when one starts from correct starting hypotheses concerning the language, the code used, etc. As another mathematician, J. Faucounau, wrote, the main factor in this formula is, beside the code, what Cl. Shannon has called "Entropy", which depends upon the language and the text itself. I quote this author : "Le facteur que Cl. Shannon a appelé "Entropy", qui depend du langage et du texte lui-même, est beaucoup plus décisif que la longueur du texte elle-même. Il est ainsi possible de déchiffrer un "texte" d'une longueur aussi courte que 5 ou 6 signes, écrit dans un système de simple substitution (= système du disque de Phaistos), lorsque l'on part, bien entendu, comme le suppose Cl. Shannon, des "bonnes hypothèses". En voici un exemple : Texte : 5-14/11-6-14-11. "Bonnes hypothèses" : a)- un chiffre = une lettre (et réciproquement) b)- le langage est du français c)- il s'agit d'un objet d)- le signe / est un séparateur de mots." If one accepts for the Phaistos Disk a "reasonable figure" for its entropy, the Shannon's Formula leads to a figure of c. 225 signs for the "Unicity Distance", inferior to the real length of the Disk (241 signs). The text of the disk can therefore be deciphered, when one starts from the "good hypotheses". (User 80.90.57.154).
What kind of a linguist are you, for confusing "phonetic evolution" with "mixture" !?.. I have written about "the expected intermediary" between "Common Greek", i.e. the ancestor of all the Greek dialects (= a language spoken probably c.3500 BC), and the "Classical Ionic", known from IXth Century BC inscriptions... It is your comparison which is absurd !... What one can say is that Old French, a language spoken in France during the Middle Age, is "the expected intermediary" between Latin and Modern French. And even if Old French is often closer to Spanish than Modern French is (for instance : Latin <cappa>, Old French <chapa>, French <chape> v./ Spanish <capa>), Spanish has nothing to do with the phonetic evolution which has led from Latin to Modern French, Old French being the intermediary !... (User 80.90.57.154).
"Archaeological Decipherment" by E.J. Barber, page 204, with a correct calculation of the "Unicity Distance", but a slightly inaccurate conclusion : it is said that the solution would be "highly suspicious". In fact, it has just to be proven, in a way or in another (Application of the solution to a second text, or verification of all its consequences), because the length of the text is too close from the "Unicity Distance". In other words : the length is big enough for finding a unique solution, but not big enough for being absolutely sure that this unique solution is the good one. Satisfied now ? (User 80.90.57.154).
It is necessary to know a language perfectly to use (the Shannon's formula). No !.. I am surprised that you don't understand that the Shannon's Formula gives the minimum length for a text to be deciphered if (a big if !) the decipherer has found - by chance or by any other means- the "good hypotheses". You don't need to know the language for making an estimate of the Unicity Distance. You have just to suppose "reasonable hypotheses" concerning the "Entropy". (Otherwise, you will find "exceptional figures", like for instance the 6-signs long text J.Faucounau has given as an example. See hereabove). All searchers (Cl. Shannon, J.Faucounau, E.Barber, Y.Duhoux, etc.) agree that, in the case of the Phaistos Disk, the "reasonable Unicity Distance" is c. 225 signs, i.e. inferior to the real figure. My theory, as you call it, is therefore accepted by all mathematicians. It doesn't say that the text's length is such that anyone can decipher the disk's text. It just say that the text's length is not a problem, as many people - who are not mathematicians- believe, including yourself. The real problem is in the need of checking the solution found (User 80.90.57.154).
a)- I didn't change my position concerning the language. I have always talked about the necessity to start from "the good hypotheses" to reach the correct solution, whatever the value of the Unicity Distance !.. Language is one of them. It can be known (as in the case of Proto-Ionic), or not (as in the case of Etruscan). b)- It depends of what you call "superficially check". So, I can tell that the deciphering attempts supposing the Phaistos Disk to be read from the center outwards are surely wrong. No need for a more complete examination !.. c)- No !.. The time passed in the search for proofs is significative of the difficulty of this indispensable step in the case of the Phaistos Disk. Convincing people is another problem, and History of Science has shown that it took sometimes 30 years or more for a correct theory to be adopted... So, I maintain my redaction. ( 80.90.39.25 12:21, 25 October 2005 (UTC))
It's really tiring to discuss with Kadmos who is unable to understand Shannon's work !.. The more as Gbrunner is taking his interventions to censor the Point of View he doesn't like ... ironically in the name of NPOV !!! The nonsense is yours. I quote the text you cited (emphasis is mine) : Shannon defined the Unicity Distance ...as an approximation of the amount of ciphertext such that the sum of real information (entropy) in the corresponding plaintext plus the entropy of the encryption key equals the number of ciphertext bits used. In other words, Shannon's Entropy depends a)- first and mainly of the Code b)- secondly of the language c)- thirdly of the plaintext itself. The more the ciphertext will be close to a randomly series, the greater will be the Unicity Distance. This is why a code which randomizes the ciphertext better than another will be more secure, and why a "Simple Substitution Code" (case of the Phaistos Disk and of my example) is the worse in regard to secrecy, Shannon's main concern, because it keeps the peculiarities of the language and of the text. In the example I've given, it's clear that the 6 signs are not taken at random : two are occurring twice, what allows the decipherment. ( 80.90.39.25, Dec. 2, 2005)
Sorry if my answers to Kadmos seem to you as personal attacks! I will try to lower their tone, but one must understand my irritation in front of Kadmos' attitude, using words like "Nonsense!" or "Absurd!" when it's he who is wrong. He was wrong, for instance, saying that Shannon's Formula is of no use in the case of the Phaistos Disk's decipherment, when it has been used by several scholars ; he was wrong saying that the Unicity Distance is not the point , or misunderstanding the question of Unicity of the Solution.
BTW, you seem to have the same problem concerning this question, so let me be more explicit about it : Shannon was dealing with cryptography. So, he was implicitely supposing that the encryption was corresponding to an encrypted text, coming from a plaintext in a known language, and he could talk about a unique solution. If you abandon this implicit condition, the notion of a unique solution vanishes, or better said, has to be modified. For instance, during a war, a coded message is decrypted as Order to attack tomorrow. Well, a fool can also say : "It is not a message. It's the numbers the ennemy intends to play at the roulette in Las Vegas". And there is no more a unique solution because the fool's solution is theorically possible... So, when one deals with something which may not be a text, and of which the chosen language may not be the good one, like this is the case with the Phaistos Disk, the notion of unique solution has to be understood : unique solution if the used hypotheses are the good ones, and one has to verify the solution discovered , for instance, like in the (pretty stupid) example I gave hereabove it can be done : By verifying the consequences of the decipherment (verifying that the ennemy prepares an attack, or even waiting until tomorrow, to be sure that the decipherment was good). To go back to your remark, one has to understand : From the Shannon's Formula, the length is sufficient if one has to deal here with some kind of syllabic text, with a Simple Substitution Code concerning the phonetic values, and with a not-exceptional Entropy (= Reasonable assumptions). As a consequence, if the other starting hypotheses (in particular concerning the language) are '"the good ones", the solution can be found (= Unicity of the Solution). But the text's length is not sufficient to warranty that the chosen hypotheses were surely the good ones. Therefore, a verification is needed by examining the consequences of the solution found. I hope to have clarify the things, this time... ( 80.90.39.25 Dec. 5, 2005)
It's really discouraging to discuss with you ! a)- I didn't change my position. Your criticism is nonsense. As this text can be deciphered (because of its abnormal entropy), it means that the Unicity Distance is 6 in this particular case (a thing that you don't seem to understand !!!) b)- the need for verification : of course, it has to do with the Shannon's Unicity Distance ! (What you don't seem to understand !!!) c)- Use of the Shannon's Formula . Why would it be of no use as you pretend ??? What hinders anyone to suppose (rightly or wrongly, it doesn't matter!) the language of the Phaistos Disk to be a known language, and apply then the Shannon's Formula??? d)-the example I gave hereabove : Of course, there are several solutions ! And for the reason you give, concerning the Unicity Distance in the case one supposes it's a date ! Because the Unicity Distance = 6 ONLY IF the "good hypotheses" that I mentioned (It's a text, written in French, etc.) are taken in consideration. Supposing that it's a date doesn't respect this condition ! (a thing you seem unable to understand!!!) e)-How will you decide whether a solution is valid or not : please read my preceding paragraph (with an example!) in this ridiculous discussion : by its consequences ! As simple as that !.. ( 80.90.39.25, 20.15, Dec. 5, 2005)