- cypherpunks-legacy - lists.cpunks.org

Re: Re[2]: Sound card as a random number source ??
by paul＠fatmans.demon.co.uk 17 Dec '03

17 Dec '03

> And what if I will use FM receiver, tuned on some channel where is the noise. > If the potentional attacker don't know the what channel was tuned, he is not > able to reproduce this IMHO. Of course, maybe it is possible to record the whole > spectrum to tapes, but I think that the noise will change when recorded and > played back from tape. It might possibly be random enough, especially if you use a tape recorder of a reasonably poor quality, thus inducing wow and flutter noise to the recording before it is sampled. anyone else care to comment? Datacomms Technologies web authoring and data security Paul Bradley, Paul(a)fatmans.demon.co.uk Paul(a)crypto.uk.eu.org, Paul(a)cryptography.uk.eu.org Http://www.cryptography.home.ml.org/ Email for PGP public key, ID: 5BBFAEB1 "Don`t forget to mount a scratch monkey"

1 0

Re: Announcement: Very Good Privacy
by Benjamin Grosman 17 Dec '03

17 Dec '03

Dear Sir, > I'm not sure how an encryption product that uses encryption > algorithms weaker than Pretty Good Privacy can be described > as being better than PGP. > > Especially when all the algorithms listed have known problems > of one kind, or another. << And yes, I know that the known > problems -- in some instances --- are entirely theoretical in > nature. >> What puzzles me is that he included two cyphers that are _extremely_ easy to break, the vignere cypher and the ascii cypher. Why include these? And what is his new permutation of RC4 and DES? Yours Sincerely, Benjamin Grosman

1 0

Strong-crypto smart cards in Singapore and Germany
by pgut001＠cs.auckland.ac.nz 17 Dec '03

17 Dec '03

Wednesday's "Straits Times" contains two front-page articles on the introduction of a CashCard which acts as an electronic wallet capable of storing from $10-$200 ("Dr Hu launches cash-in-a-card payment system"), and an identity card capable of Internet electronic transactions with (presumably RSA) 1024-bit encryption ("50,000 to take part in electronic ID trials"). The ID card can also be implemented as software on disk. It appears to be purely a form of storing an ID which is then transmitted in encrypted form. The CashCard, on the other hard, is an electronic wallet developed by a group of Singapore banks. There are no details on how it works, except that it doesn't have any sort of protection - it's up to you to make sure the card isn't stolen. The standardisation committee of the German banks have also produced an electronic wallet which should have 25 million (yes, 25M) users by January of next year. Again, this is a pure electronic wallet, with 2-key triple DES and 768-bit (to become 1024-bit) RSA encryption. The relevant standards are still in the process of being translated, but should be available Real Soon Now (the complete specification will be made public). This looks like a very nice system, and unlike Mondex doesn't rely entirely on the hope that criminals can't get at the data on the card. Peter.

5 4

Re: Announcement: Very Good Privacy
by Mark Rosen 17 Dec '03

17 Dec '03

> What puzzles me is that he included two cyphers that are _extremely_ easy to > break, the vignere cypher and the ascii cypher. Why include these? And what > is his new permutation of RC4 and DES? The reason I included ASCII and Vigenere was because they are very fast; on a friend's machine, I can get 1.7mb/s with ASCII but only around 600k/s with NewDES, the fastest "secure" cipher. I do agree that it is kind of redundant to include both ASCII and Vigenere, as from a cracking standpoint, they are both identical. As for Psuedo-RC4, that is just to try to avoid a lawsuit from RSADSI; RC4 is a trademark. NewDES is an actual algorithm; it's even mentioned in Applied Cryptography. It does have some holes and is in fact less secure than DES (refer to Applied Cryptography for details on this); I included that because it is fast and provides much, much more security than ASCII or Vigenere, even if it is less secure than DES. Mark Rosen FireSoft - http://www.geocities.com/SiliconValley/Pines/2690 Mark Eats AOL - http://www.geocities.com/TimesSquare/6660

1 0

Re: Announcement: Very Good Privacy
by Mark Rosen 17 Dec '03

17 Dec '03

> > I have written an encryption program called Very Good Privacy > > Trademark violation here. Probably not a good thing. Is PGP trademarked? Are you sure? And if it is, is naming something Very Good Privacy a violation? Why? Legal people, please help! > I'm not sure how an encryption product that uses encryption > algorithms weaker than Pretty Good Privacy can be described > as being better than PGP. > > Especially when all the algorithms listed have known problems > of one kind, or another. << And yes, I know that the known > problems -- in some instances --- are entirely theoretical in > nature. >> In your quote, you deleted the smiley about VGP. It's a joke. Your claim that VGP is weaker than PGP is unfounded, as VGP uses IDEA, which is the symmetric algorithm used in PGP. ASCII and Vigenere are very weak, but they are fast and geared toward people who simply want to stop their brother or sister from getting into something. As for the security of other algorithms, BlowFish is used in PGPfone and RC4 is used in SSL -- since you control the key size in VGP, you can make RC4 virtually unbreakable (not to mention the problem of the attacker figuring out the key size in the first place). Mark Rosen FireSoft - http://www.geocities.com/SiliconValley/Pines/2690 Mark Eats AOL - http://www.geocities.com/TimesSquare/6660

1 0

WinSock Remailer Back Up
by Joey Grasty 17 Dec '03

17 Dec '03

-----BEGIN PGP SIGNED MESSAGE----- Y'all: Due to a misconfiguration on one of the two machines that operate the WinSock Remailer, messages were lost from about Wednesday, 27 November, to Sunday, 1 December. We apologize for the lost messages. Regards, Joey Grasty Jim Ray WinSock Remailer Operators -----BEGIN PGP SIGNATURE----- Version: 2.6.2 iQB1AwUBMqJvkg6sYKeTQAOtAQGrNwMA2MYpRHLtkeksVcoyx6T9+V+36fey8FXM sdbxI3DjN8Xq8Se81R7+qXNWxgVx2/HxQtu4w9Ea1GQz4B8cjYdlMVLXNIgdGrHy 3We8saHtvVIzCTFnvFIEn6MStNAZb+gw =9Vso -----END PGP SIGNATURE-----

1 0

NEWS.COM DAILY DISPATCH
by The CNET newsletter 17 Dec '03

17 Dec '03

*************************************** NEWS.COM DAILY DISPATCH Wednesday, November 27, 1996 San Francisco, California, USA *************************************** WELCOME! *************************************** The NEWS.COM DAILY DISPATCH highlights the up-to-the minute technology news presented by NEWS.COM. It tempts readers with coverage of today's hottest stories, news in the making, (in)credible rumors, and other indispensable information. For complete versions of the stories updated all day long, head over to: http://www.news.com/?nd *************************************** CONTENTS SCOOPS AND TOP STORIES Online rivals Gates and Case circle each other Attention Comdex survivors! Need someone to feel yur paynnne? AOL's prix fixe might cause another service delay Intel's multimedia chip: More suitable for fun & games than business? ANNOUNCEMENTS An easy way for you to customize NEWS.COM Late-breaking stories just a click away with Desk Top News Send us your questions, comments, flotsam, and jetsam *************************************** SCOOPS AND TOP STORIES ONLINE RIVALS GATES AND CASE CIRCLE EACH OTHER The online world has become a killing field. CompuServe pulls the plug on its consumer-based Wow service; Prodigy, at the bottom of the heap, transmogrifies Kafka-like into an ISP. Now, as the mighty Web threatens to consume all in its path, these two cyberbehemoths engage in a struggle for the whole enchilada. http://www.news.com/SpecialFeatures/0%2C5%2C5770%2C00.html?nd ATTENTION COMDEX SURVIVORS! NEED SOMEONE TO FEEL YUR PAYNNNE? Okay, you've been through a lot, granted...but speaking out about your Comdex trials and tribulations can help the all-important healing process begin. So please, send us your war stories. We will publish the five best, and the winners--picked on a totally subjective basis by our editorial staff-- will get supercool CNET T-shirts. Send your tales to suggestions(a)news.com, and include the words 'Comdex stories' in the subject of your message. P.S. Don't forget to check out our comprehensive coverage of the sublime and the ridiculous at Comdex '96. http://www.news.com/Comdex/?nd AOL'S PRIX FIXE OFFER MIGHT CAUSE ANOTHER SERVICE DELAY Just when you AOLers thought it was safe to go back online, the possibility of another service glitch rises up on the cyberhorizon. Starting on December 1, AOL's 7 million users have the option of switching to an all-you-can-eat pricing offer; the resultant "usage surge" may cause service delays, according to a recent AOL announcement. Wise users might consider getting their heavy surfing done this long weekend, once that turkey-induced coma sets in. http://www.news.com/News/Item/0%2C4%2C5783%2C00.html?nd INTEL'S MULTIMEDIA CHIP: MORE SUITABLE FOR FUN AND GAMES THAN BUSINESS Intel's latest technological achievement, the MMX Pentium processor, would certainly be ideal for running sophisticated business applications--that is, if there WERE sophisticated business applications for it to run. At this point, MMX seems more useful in entertainment and gaming settings than in the executive suite. The $64,000 question: will developers build for this flashy new kid on the block? http://www.news.com/News/Item/0%2C4%2C5782%2C00.html?nd *************************************** ANNOUNCEMENTS AN EASY WAY FOR YOU TO CUSTOMIZE NEWS.COM So many bits, so little time? Sounds like you need Custom News. Identify the topics, keywords, or sections you're most interested in, and Custom News will a create a page of headlines and summaries for all stories that match your criteria. Check it out at: http://www.news.com/Personalization/Entry/1%2C21%2C%2C00.html?nocache=1 *************************************** LATE-BREAKING STORIES A CLICK AWAY WITH DESK TOP NEWS How would you like having split-second access to the very latest news on the Net? Our Desk Top News feature puts our 20 most recent stories right there on your desktop for you to review at any time. Here's how it works: 1. From any story, click Desk Top News in the top right. 2. A window will open showing our last 20 stories. 3. Click on a headline to display the story. 4. Desk Top News updates itself every 30 minutes. 5. You become known as Ms./Mr. Cyber-Info. It feels good. http://www.news.com/Help/Item/0%2C24%2C12%2C00.html?nd *************************************** SEND US QUESTIONS, COMMENTS, FLOTSAM, AND JETSAM If you have any questions, suggestions or remarks, send us a message: newsdispatch(a)cnet.com *************************************** CNET: The Computer Network http://www.cnet.com/ http://www.news.com/ http://www.gamecenter.com/ http://www.download.com/ http://www.search.com/ http://www.shareware.com/

1 0

Ignoramus Chewed-Off on IPG algorithm
by ichudov＠algebra.com 17 Dec '03

17 Dec '03

Hi, I was sort of tired of endless talk that "IPG algorithm was not peer-reviewed, blah blah blah, so we won't even look at it, blah blah blah", and decided to look at what Don Wood writes and try to see how his program actually works. Of course, I am not an expert in cryptography, and will appreciate all corrections. The web page to look at is http://www.netprivacy.com/algo.html, and it describes IPG algorithm in some detail. First of all, the description of the algorithm is extremely unclear. I understand that this may be Don Wood's writing style, but it is certainly not the most efficient style for precise communications. I suggest that Don tries to rewrite his description to be more structured. Second, I seriously suspect that his algorithm of "trimming" is NOT going to work right. Just to remind everyone, he generates pseudo-random A(JV), B(JV), C(JV) such that 16384 < C < 20361 B < 12227 A arbitrary (at least the web page contains no restrictions on the value of A). and then goes on to "trimming" -- a process that obtains a new value of A that is LESS than 16384 through this algorithm: DO JV=JV+1 IF JV=53 THEN JV=0 A(JV)=(A(JV)+B(JV)) MOD C(JV) UNTIL A(JV)<16384 We shall first note that THERE ARE CASES WHEN THIS ALGORITHM WILL NEVER STOP! For example, if all A values are _initially_ 16385 and all C values are 16386 and all B's are 0, it is obvious that the pseudocode above will be stuck in endless loop. No good for IPG algorithm. in fact, if only some triplets of A, B, and C have B == 0 and 16384 < A < C, these triplets will always be ignored (skipped) by his trimming process. Second, IPG's claim that ``Of even greater importance is the second role of producing a distribution of the numbers 0 to 16383, so that the number of each of those numbers produce, asymptotically approach an even distribution, in the same way that theoretically a true random number generator produces a set of numbers whose frequency asymptotically approach an even distribution'' is FALSE: I do not see why the distribution of A's should necessarily contain all numbers from 0 to 16383, less to "approach" an even distribution. For example, if all initial A, B, and Cs are even numbers, we'll obviously never get an odd number for A(JV), no matter how many iterations. If, say, 40 out of 53 of triplets A, B, and C contain only even numbers (not a statistical impossibility!), then about 75% of resulting values of A(JV) will be even. No good for the keystream. There are many more imperfections than just related to all-even triplets. - - - - Let's go on, to the description of the "scrambling tables" and actual encryption. He uses three tables, DIFF, DISP, DETR, each containing 4096 elements. DISP is randomly generated (or so I understand his term "prescrambled"), DIFF is a random transposition of DISP (same values as in DISP, but in another order), and DETR, again, is filled with some random data. It does not really matter to my analysis how DISP and DETR are generated. Here's how he encrypts the data (in a loop for I from 0 to 4095): D1=(D1+A(JV)) AND 4095 D2=DISP((D2+D1) AND 4095) F=DETR(D2)<<8 D3=(D3+D2) AND 4095 BUF(DIFF(I))=BUF(DIFF(I)) XOR (F+(DETR(DISP(D3)))) I.e., D1, D2, and D3 are regenerated each pass of the loop and then are used to encrypt item DIFF[I] (i.e., 2-byte word with index DIFF[i], NOT I). That is very bad: what it means is that we are not guaranteed that all BUF[i] will be scrambled! It is fairly clear that about 1/(e*e) (e is the base of natural logs) will be left unscrambled!!! This follows from the fact that even with perfectly random numbers used as DIFF[i], the probability that a cell J will be scrambled is 2/4096 (it is 2 and not 1 since each hit covers two bytes and not one). Therefore, the probability that cell J will be left unscrambled is (1-2/4096) ** 4096 which is approximately equal exp(-2) == 1/(e*e). It means that 1 out of seven characters will be left unencrypted at all. No good for a stream cipher. I think that his cipher could have been made stronger if he actually did not use DIFF[i] as an index and used I instead. Still, of course, his algorithm has nothing to do with one time pads. Since this is clearly not a matter of high theories but a simple question of accepted definitions, there is little point to argue about it and I preferred to address details of his algorithm instead. There may be more problems with IPG algorithm that I simply missed. I would appreciate your corrections and notes. Thank you. - Igor.

1 0

Re: The Difference Between The Right And Left
by Murray Hayes 17 Dec '03

17 Dec '03

On Wed, 27 Nov 1996 22:47:06 -0800, Troy Varange wrote: > The French Revolution perhaps provided the best > definition as the origional coiners of the terms. > > The Left is for radical change and the Right is for > evolutionary change. The Jacobin (sp) sat on the left side while the other guys (I forget their names) sat on the right. Of course both used to be on the left before they klled everyone on the right after which an expansion of sorts happened to fill the seats. I think the event was called the night of the long knives? > > That would put mainstream politics decidedly on the right > wing side of the political spectrum. I would have thought main stream would have been in the middle. > The left wing is a rather motly collection including many > anarchists, communists and nazis, yet many of the same > types are decidedly right wing. mhayes(a)infomatch.com It's better for us if you don't understand It's better for me if you don't understand -Tragically Hip

1 0

Re: Ignoramus Chewed-Off on IPG algorithm
by ichudov＠algebra.com 17 Dec '03

17 Dec '03

Igor Chudov @ home wrote: > > Hi, > > I was sort of tired of endless talk that "IPG algorithm was not > peer-reviewed, blah blah blah, so we won't even look at it, > blah blah blah", and decided to look at what Don Wood writes and > try to see how his program actually works. > > Of course, I am not an expert in cryptography, and will appreciate all > corrections. The web page to look at is http://www.netprivacy.com/algo.html, > and it describes IPG algorithm in some detail. > > First of all, the description of the algorithm is extremely unclear. I > understand that this may be Don Wood's writing style, but it is certainly > not the most efficient style for precise communications. I suggest that > Don tries to rewrite his description to be more structured. > > Second, I seriously suspect that his algorithm of "trimming" is NOT > going to work right. Just to remind everyone, he generates pseudo-random > A(JV), B(JV), C(JV) such that > > 16384 < C < 20361 > B < 12227 > A arbitrary (at least the web page contains no restrictions > on the value of A). > > and then goes on to "trimming" -- a process that obtains a new value > of A that is LESS than 16384 through this algorithm: > > DO > JV=JV+1 > IF JV=53 THEN JV=0 > A(JV)=(A(JV)+B(JV)) MOD C(JV) > UNTIL A(JV)<16384 > > We shall first note that THERE ARE CASES WHEN THIS ALGORITHM WILL NEVER > STOP! For example, if all A values are _initially_ 16385 and all C > values are 16386 and all B's are 0, it is obvious that the pseudocode > above will be stuck in endless loop. > > No good for IPG algorithm. > > in fact, if only some triplets of A, B, and C have B == 0 and 16384 < A < C, > these triplets will always be ignored (skipped) by his trimming process. Note also that if B(K) == 1, his algorithm will need to make C passes through the loop for JV == k, in order to generate a new value of A(JV). This is very inefficient and results in a bias for triplets with high Bs -- because they will generate good A(JV) more frequently. - Igor.

2 2