- cypherpunks-legacy - lists.cpunks.org

1993 RSA Data Security Conference
by gnu 06 Nov '92

06 Nov '92

RSA is putting on a conference on public key cryptography on the SF peninsula on January 14th (conference) and 15th (tutorials and exposition). Last year's conference was smaller (one day) and quite interesting. Many of the best cryptographers in the world attended. It's worth prying yourself out of your usual haunts for. The conference is free. Registration is "extremely limited, and is first come, first served". Phone +1 415 595 8782 and ask for Conference Registration, to reserve your place (and a place for anyone else at your company who's interested). I'll see you there! John Gilmore

1 0

Re: a cryptographic deal with the devil
by George A. Gleason 06 Nov '92

06 Nov '92

Re. the digital wiretapping "compromise." As a telecom professional I absolutely resent and will resist any attempts to mandate backdoors into my PBXs. No compromise on that. Period. We've all heard the arguements many times: vast surveillance power, diminution of privacy, potential major security problems... I'd like to suggest something of a compromise which doesn't have these risks. Common carriers (local and long distance telcos) are currently required to provide access to line terminals when presented with a court order for a wiretap. This access could reasonably be extended to requiring the telco to connect a demultiplexer in the case of digital transmission, or some kind of appropriate signal splitter in the case of fiber optics. The agency requesting the tap would of course pay the bill for materials and labor. Now this gets law enforcement their demultiplexed signal path so they can tap only the intended target line, but it preserves the existing structure which prevents law enforcement "hacking," since no backdoors would be involved. For PBXs (this is my department), a slightly distasteful but acceptable compromise would be to have interconnect companies (the folks who install your PBX or key system) register with the local operating telcos, providing the interconnect company's name and contact information on the telco record for each subscriber. So for instance, General Widgets has XYZ Telecom install a new PBX; XYZ Telecom is required to inform the local operating company that they have just acquired General Widgets as a client. Now if a law enforcement agency gets a court order for a tap, they go to the local telco and ask who the interconnect company is for that subscriber. (We're talking here about a scenario in which one or a small number of extensions in a phone system are believed to be used for criminal purposes, so law enforcement has to tap those extensions only and not everyone who is on that phone system.) Now law enforcement visits the interconnect company and presents them with the court order, which requires the interconnect company to provide access to the line terminals of the suspect extension(s), and/or provide a demultiplexer etc. if needed (at law enforcement's expense of course)... and of course, under penalty of contempt of court, refrain from disclosing the situation to the client. Now this gets law enforcement their legitimately needed access to suspect extensions on PBXs, prevents interconnect companies from blowing the whistle to their clients, and still preserves privacy protections since there is no backdoor into the system. Now here is why I think the FBI wants backdoors: Recall that under the "war on drugs" etc., a ruling was handed down (I can't recall which branch of govt originated this) which says that a wiretap may be conducted for up to 72 hours for "investigational purposes" without a court order; and the material recorded may then be used to go and get a court order for a continuing wiretap. This places authority in the hands of law enforcement agencies to conduct taps any time they suspect someone of something, and then go see the judge after the fact. Now without backdoors, law enforcement has to depend on the goodwill of telcos to get access, and is kind of stuck when it comes to PBXs and key systems. I'm willing to bet there is a pretty substantial amount of "investigational" tapping going on, and that the FBI is interested in vastly expanding it. The compromises I'm proposing don't address this investigational tapping, and that's just fine, since that ought to be challenged in court or defeated one way or another. -gg@well

1 0

random numbers (resend)
by Peter Shipley 05 Nov '92

05 Nov '92

[my first send bounced so I am resending] Here is some code to run on a SparcStaion (I or II) to generate random numbers based on noise from the audio chip. I have not tested the true randomness of the out but cause I don't know how. Thus if anyone can please do and send me the results. I suppose some minor tweeking can be made for the mid range gain for a better spread. unfortunatly there is no way to turn off the u-law audio compression the chip uses to outputs data if you plot the out put you will see the problem the compression caused. #! /bin/sh # here be a shar file echo x - Audio/Makefile cat >Audio/Makefile <<'!E!O!F!' LD= $(CC) CC= cc LDFLAGS=-g CFLAGS=-g -I/usr/demo/SOUND -I/usr/local/include LIBS= -L/usr/demo/SOUND/ -laudio audio_rand: audio_rand.o $(LD) $(LDFLAGS) audio_rand.o $(LIBS) -o $@ test: audio_rand audio_rand | head -10000 > \#fff echo "plot '#fff'" | gnuplot !E!O!F! #! /bin/sh echo x - Audio/audio_rand.c cat >Audio/audio_rand.c <<'!E!O!F!' static char *_author = "Peter Shipley\n"; #define AUDIO_CHIP #include <stdio.h> #include <fcntl.h> #include <sys/ioctl.h> #include <sun/audioio.h> #include <sbusdev/audio_79C30.h> #include <multimedia/ulaw2linear.h> static char *_who_did_it = "Peter M Shipley (1992) [v0.1]\n"; /* X: 0-15 R: 16-31 GX: 32-33 GR: 33-34 */ #define MMR2_BITS 45 #define GX_GAIN 32 main() { struct audio_ioctl ai; extern void bzero(); int fd; int j, i; if( (fd = open("/dev/audio", O_RDONLY)) < 0) { perror("open:"); exit(1); } bzero(ai, sizeof(struct audio_ioctl) ); ai.control = AUDIO_MAP_ALL; if ( ioctl( fd, AUDIOGETREG, &ai ) < 0 ) { perror( "AUDIOGETREG ALL:" ); exit(1); } /* set audio input to a unconnected pin (audio input B) */ ai.data[MMR2_BITS] = ( ai.data[MMR2_BITS] & ~AUDIO_MMR2_BITS_AINB); /* set gain of the GX reg to the max */ ai.data[GX_GAIN] = 0x7F; ai.data[GX_GAIN+1] = 0x0; for(i=0;;) { char buf[BUFSIZ]; read(fd, buf, BUFSIZ); for(j=0; j<BUFSIZ ; j++) printf("%d %d\n", i++, audio_u2s(buf[j])); } } !E!O!F! #! /bin/sh echo x - Audio/reg.c cat >Audio/reg.c <<'!E!O!F!' static char *_author = "Peter Shipley\n"; #define AUDIO_CHIP #include <stdio.h> #include <fcntl.h> #include <sys/ioctl.h> #include <sun/audioio.h> #include <sbusdev/audio_79C30.h> static char * print_byte(); static struct _map { char *label; char size; } map[] = { "X filter", 16, "R filter", 16, "GX Gain", 2, "GR Gain", 2, "GER Gain", 2, "Sidetone", 2, "Frequency Tone gen", 2, "Amplitude Tone gen", 2, "MMR1", 1, "MMR2", 1 }; #define NMAPS (sizeof(map) / sizeof(struct _map) ) struct audio_ioctl ai; main() { extern void bzero(); int fd; int oldmmr2; int newmmr2; if( (fd = open("/dev/audio", O_RDWR)) < 0) { perror("open:"); exit(1); } dump(fd); /* bzero(ai, sizeof(struct audio_ioctl) ); ai.control = AUDIO_MAP_GX; if ( ioctl( fd, AUDIOGETREG, &ai ) < 0 ) { perror( "AUDIOGETREG MMR2:" ); exit(1); } ai.data[0] = ai.data[1] = 0; if ( ioctl( fd, AUDIOSETREG, &ai ) < 0 ) { perror( "AUDIOSETREG MMR2:" ); exit(1); } */ sleep(5); dump(fd); } dump(f) int f; { int i,k; bzero(ai, sizeof(struct audio_ioctl) ); ai.control = AUDIO_MAP_ALL; if ( ioctl( f, AUDIOGETREG, &ai ) < 0 ) { perror( "AUDIOGETREG MMR2:" ); exit(1); } k=0; for(i = 0 ; i < NMAPS ; i++) { int j; (void) printf("%s:\n", map[i].label); for(j=1; j <= map[i].size; j++,k++) { (void) printf("%10s", print_byte(ai.data[k])); if ( (j % 7) == 0) (void) fputc('\n', stdout); } (void) fputc('\n', stdout); } (void) fputc('\n', stdout); } static char * print_byte(c) char c; { static char bt[9]; bt[0] = ( (c & 0x01) ? '1' : '0'); bt[1] = ( (c & 0x02) ? '1' : '0'); bt[2] = ( (c & 0x03) ? '1' : '0'); bt[3] = ( (c & 0x04) ? '1' : '0'); bt[4] = ( (c & 0x05) ? '1' : '0'); bt[5] = ( (c & 0x06) ? '1' : '0'); bt[6] = ( (c & 0x07) ? '1' : '0'); bt[7] = ( (c & 0x08) ? '1' : '0'); bt[8] = NULL; return bt; } !E!O!F!

1 0

a cryptographic deal with the devil
by efrem＠informix.com 05 Nov '92

05 Nov '92

It is widely believed that the Police, the FBI, and other government agencies tap many more phones than they admit in public. This is not counting the NSA's monitoring of all the international traffic they can stuff into a computer. Now the FBI wants telephone switch manufactures to supply them with desktop phone tapping technology. Real-time delivery of all conversations straight to the agent's desk. This is scary, but it might an opportunity. Given that congress is likely to eventually allow the FBI this tech toy, would you go along with a deal that all taps would really require a court order and that the exact time and location of all taps would eventually be made public? No cheating possible. I believe such an compromise may be possible via cyptographic technology, I've not worked out the details, but here is a sketch of the idea. The legislation authorizing the tapping facilities would require that each tap be activated by a key supplied by a court. Each tap would require a new key. The switching gear would not only enforce the key mechanism, but transmit a record of the tap to some agency outside the court system. Both the courts and this agency would be required to periodically make public all old taps. Part of this information would include tamper-proof sequence codes and signatures to guarantee that all taps were in fact reported. The law would effectively be enforced by the switch hardware. We would not only know how many phones were tapped, but whose phones were tapped. This last would pose a privacy problem, The law could just require tappees being eventually informed of the invasion of their lives, rather than public disclosure. Problems: * We consent to the process, it legitimates phone taps. * The hardware would be in place for massive monitoring of communications if the government could get the public to accept dropping the limitations of the scheme. [It might be possible to limit the abuse by having the switches communicate and not accept anymore than a 1,000 taps a year.] * The lobbying for this would be difficult. Additional opportunity: * By proposing and lobbying for this type of scheme, it could be made obvious that the cops and mega cops want to maintain much closer surveillance than they are willing to admit. However, you'd have to be prepared for them to accept the compromise. They may figure on making all their illegal taps via other means. --efrem

1 0

random numbers
by unknown＠example.com 05 Nov '92

05 Nov '92

Here is some code to run on a SparcStaion (I or II) to generate random numbers based on noise from the audio chip. I have not tested the true randomness of the out but cause I don't know how. Thus if anyone can please do and send me the results. I suppose some minor tweeking can be made for the mid range gain for a better spread. unfortunatly there is no way to turn off the u-law audio compression the chip uses to outputs data if you plot the out put you will see the problem the compression caused. #! /bin/sh # here be a shar file echo x - Audio/Makefile cat >Audio/Makefile <<'!E!O!F!' LD= $(CC) CC= cc LDFLAGS=-g CFLAGS=-g -I/usr/demo/SOUND -I/usr/local/include LIBS= -L/usr/demo/SOUND/ -laudio audio_rand: audio_rand.o $(LD) $(LDFLAGS) audio_rand.o $(LIBS) -o $@ test: audio_rand audio_rand | head -10000 > \#fff echo "plot '#fff'" | gnuplot !E!O!F! #! /bin/sh echo x - Audio/audio_rand.c cat >Audio/audio_rand.c <<'!E!O!F!' static char *_author = "Peter Shipley\n"; #define AUDIO_CHIP #include <stdio.h> #include <fcntl.h> #include <sys/ioctl.h> #include <sun/audioio.h> #include <sbusdev/audio_79C30.h> #include <multimedia/ulaw2linear.h> static char *_who_did_it = "Peter M Shipley (1992) [v0.1]\n"; /* X: 0-15 R: 16-31 GX: 32-33 GR: 33-34 */ #define MMR2_BITS 45 #define GX_GAIN 32 main() { struct audio_ioctl ai; extern void bzero(); int fd; int j, i; if( (fd = open("/dev/audio", O_RDONLY)) < 0) { perror("open:"); exit(1); } bzero(ai, sizeof(struct audio_ioctl) ); ai.control = AUDIO_MAP_ALL; if ( ioctl( fd, AUDIOGETREG, &ai ) < 0 ) { perror( "AUDIOGETREG ALL:" ); exit(1); } /* set audio input to a unconnected pin (audio input B) */ ai.data[MMR2_BITS] = ( ai.data[MMR2_BITS] & ~AUDIO_MMR2_BITS_AINB); /* set gain of the GX reg to the max */ ai.data[GX_GAIN] = 0x7F; ai.data[GX_GAIN+1] = 0x0; for(i=0;;) { char buf[BUFSIZ]; read(fd, buf, BUFSIZ); for(j=0; j<BUFSIZ ; j++) printf("%d %d\n", i++, audio_u2s(buf[j])); } } !E!O!F! #! /bin/sh echo x - Audio/reg.c cat >Audio/reg.c <<'!E!O!F!' static char *_author = "Peter Shipley\n"; #define AUDIO_CHIP #include <stdio.h> #include <fcntl.h> #include <sys/ioctl.h> #include <sun/audioio.h> #include <sbusdev/audio_79C30.h> static char * print_byte(); static struct _map { char *label; char size; } map[] = { "X filter", 16, "R filter", 16, "GX Gain", 2, "GR Gain", 2, "GER Gain", 2, "Sidetone", 2, "Frequency Tone gen", 2, "Amplitude Tone gen", 2, "MMR1", 1, "MMR2", 1 }; #define NMAPS (sizeof(map) / sizeof(struct _map) ) struct audio_ioctl ai; main() { extern void bzero(); int fd; int oldmmr2; int newmmr2; if( (fd = open("/dev/audio", O_RDWR)) < 0) { perror("open:"); exit(1); } dump(fd); /* bzero(ai, sizeof(struct audio_ioctl) ); ai.control = AUDIO_MAP_GX; if ( ioctl( fd, AUDIOGETREG, &ai ) < 0 ) { perror( "AUDIOGETREG MMR2:" ); exit(1); } ai.data[0] = ai.data[1] = 0; if ( ioctl( fd, AUDIOSETREG, &ai ) < 0 ) { perror( "AUDIOSETREG MMR2:" ); exit(1); } */ sleep(5); dump(fd); } dump(f) int f; { int i,k; bzero(ai, sizeof(struct audio_ioctl) ); ai.control = AUDIO_MAP_ALL; if ( ioctl( f, AUDIOGETREG, &ai ) < 0 ) { perror( "AUDIOGETREG MMR2:" ); exit(1); } k=0; for(i = 0 ; i < NMAPS ; i++) { int j; (void) printf("%s:\n", map[i].label); for(j=1; j <= map[i].size; j++,k++) { (void) printf("%10s", print_byte(ai.data[k])); if ( (j % 7) == 0) (void) fputc('\n', stdout); } (void) fputc('\n', stdout); } (void) fputc('\n', stdout); } static char * print_byte(c) char c; { static char bt[9]; bt[0] = ( (c & 0x01) ? '1' : '0'); bt[1] = ( (c & 0x02) ? '1' : '0'); bt[2] = ( (c & 0x03) ? '1' : '0'); bt[3] = ( (c & 0x04) ? '1' : '0'); bt[4] = ( (c & 0x05) ? '1' : '0'); bt[5] = ( (c & 0x06) ? '1' : '0'); bt[6] = ( (c & 0x07) ? '1' : '0'); bt[7] = ( (c & 0x08) ? '1' : '0'); bt[8] = NULL; return bt; } !E!O!F!

1 0

Please remove me from the list
by Matt_Kelly 04 Nov '92

04 Nov '92

Please remove me from the cypherpunks distribution list. This is the second time I've asked. thanks, MattKelly(a)antioc.antioch.edu

1 0

Re: Update on Crypto Session at Hackers
by Paul E Baclace 03 Nov '92

03 Nov '92

Sounds good--too short, of course, so it is going to be challenging to prevent digressions. The "Crypto" stamp sounds like a good way for people to follow up on details/questions after the session. Paul

1 0

another service
by norm＠xanadu.com 03 Nov '92

03 Nov '92

I think that there is an even easier time stamping service than that described by Eric Hollander. It too requires a trusted service with a trusted clock. It has just two key pairs, A and B. The protocol is that you send it a message, perhaps with payment under public-A. It returns the message joined with the time provided by its clock under key private-B. The returned message provides evidence in the future that you held the original message at the time indicated in the returned message. This protocol requires no data base of keys. It was once the practice (perhaps still) to mail oneself (US mail) a certified envelope with information that one might want to prove that he had had at some earlier date. One would then keep but not open the delivered envelope.

1 0

Update on Crypto Session at Hackers
by tcmay＠netcom.com 03 Nov '92

03 Nov '92

Here's the semi-final information on the Crypto Session at Hackers this coming weekend: (Note: These inputs and volunteered talks came in over the past several days. There may be some changes of course. I'd like to see us get together---anybody who is reading this message---on FRIDAY NIGHT, in the refreshment room at MIDNIGHT, so we can plan, make any schedule changes, etc. This is of course up to you folks. I'll post a message somewhere prominent reminding you and listing time and location, if different from above.) * Time: Saturday afternoon, probably 3:00--4:30 (check the schedule!) (in an earlier update I mistakenly said we had only an hour) * Format: 7-10 minute talks by 4-5 people, followed by open discussion, questions, debate, etc. * Schedule: - Opening comments, groundrules, settling down, etc. (5-10 min), Tim May (I'll point people to a glossary posted on the wall and ask them not to interrupt the speakers with questions about the basics of crypto, such as whether DES has a trapdoor, etc.) - PGP, Fidonet, and Mail....Tom Jennings - Diffie-Hellman key exchange for rlogin, etc....John Gilmore - Anonymous remailers and DC-Nets....Eric Hughes - Digital time-stamping....either Stu Haber or W. Scott Stornetta - Open discussion, debate, etc. (45 minutes) I'll be the moderator, to keep folks to their time allotment. * If you have "poster" material, please bring it * We can also suggest that the discussion be continued later that evening, in one of the ad hoc sessions around midnight. Maybe some new topics can be be planned for a late night session. * Eric Hughes has suggested we stamp our badges with my red "CRYPTO" stamp, so that we can know ourselves (sort of a crypto oxymoron, eh?). See me and I'll stamp you, if you wish. Thanks for all of your excellent suggestions. I think this could be one of the best sessions at Hackers. --Tim May, 408-688-5409 -- .......................................................................... Timothy C. May | Crypto Anarchy: encryption, digital money, tcmay(a)netcom.com | anonymous networks, digital pseudonyms, zero 408-688-5409 | knowledge, reputations, information markets, W.A.S.T.E.: Aptos, CA | black markets, collapse of governments. Higher Power: 2^756839 | PGP Public Key: awaiting Macintosh version.

1 0

request
by SIVAI%FRSAC11.BITNET＠pucc.Princeton.EDU 02 Nov '92

02 Nov '92

Hello , I wish i knew the internet address , if any , to which one should ask so to subscribe to the 'sci.cryp' usenet's bb . I'd like to know either what cypherpunk stands for ? I'm interested in mathematical processing of datum ( well datas or datum....a key issue no doubt ) . Anyway i thank you in advance . Regards Louis Safer

1 0