December 2003 - cypherpunks-legacy

real live Detweiler messages- anyone?
by jamiel＠sybase.com 17 Dec '03

17 Dec '03

I have a writing assignment coming up (nothing to do with much that is talked about on this list), and some example posts by Detweiler could be really handy as references. I have one from a long time ago, but could use some others- anybody save any for posterity? If so, could you forward me a couple of juicy ones? thanks- -j -- "Blah Blah Blah" ___________________________________________________________________ Jamie Lawrence <jamiel(a)sybase.com>

1 0

Re: (fwd) Re: BSD random() - any good (source included)?
by jamiel＠sybase.com 17 Dec '03

17 Dec '03

At 3:20 PM 07/07/94 -0500, Jim choate wrote 69K of forwards (so far) that I would much rather look at in sci.math or sci.stat.math than in my mailbox, unrequested. Didn't we have a little forwarding talk recently enough for your avarage televangelist viewer to remember it? Kindly cut it out. -j -- "Blah Blah Blah" ___________________________________________________________________ Jamie Lawrence <jamiel(a)sybase.com>

2 1

(fwd) Random Numbers - CHIKSN.FOR
by Jim choate 17 Dec '03

17 Dec '03

Newsgroups: sci.stat.math,sci.math,sci.math.num-analysis Path: bga.com!news.sprintlink.net!news.onramp.net!convex!cs.utexas.edu!swrinde!ihnp4.ucsd.edu!library.ucla.edu!csulb.edu!csus.edu!netcom.com!deleyd From: deleyd(a)netcom.com Subject: Random Numbers - CHIKSN.FOR Message-ID: <deleydCsIBAp.KuF(a)netcom.com> Organization: NETCOM On-line Communication Services (408 261-4700 guest) Date: Wed, 6 Jul 1994 06:56:49 GMT Lines: 526 Xref: bga.com sci.stat.math:1316 sci.math:15354 sci.math.num-analysis:3355 {Approx. 520 lines} C CHIKSN.FOR C C This is the program which impliments the chi-square test used to test C random number generators. Presented here if you would wish to play C with it yourself, maybe do some testing of your own. See the paper C "Computer Generated Random Numbers" sections 4 and 6 for an explanation C of what this program does. C C This is not polished code you put on a shelf and admire, this is code C you dig your hands into and work with to make it do what you want it to. C C The main routine is the one to tinker with. This code is meant to be C modified to suit your needs. The goal is to fill up the bins with C balls. Here's a brief outline: C C 1. ASK USER INPUT: C a. Number of dimensions? NDIM C b. Number of bins per dimension? NBINSPD C c. Total number of balls to throw at bins? NBALLS C d. Number of tests to run? NCHITESTS C e. Random number generator to use (if more than one defined) C f. SEED value to initialize generator with C C 2. CREATE ARRAY BINS and ZERO ARRAY C C 3. THROW THE BALLS AT THE BINS and CALCULATE PROBABILITY: C C LOOP (do CHITEST=1 to NCHITESTS) C zero BIN array C LOOP (do J=1 to NBALLS) C get random numbers r1,r2,...,rn) C increment BIN(r1,r2,...,rn) C ENDLOOP C CALL CHSONE to calculate chi-square probability C ENDLOOP C CALL KSONE to calculate Kolmogorov-Smirnov probability C C The main routine here is where you put a call to your random number C generator, or for speed you can attempt a direct implimentation of your C random number generator to save the overhead of a call. (It can make a C difference when you call the random number generator 100 million times C in one test). C C The main routine defines a large one-dimensional array called BINS, the C maximum size of which would depend on your account quotas and machine C specific limitations. The array BINS keeps track of how many balls have C fallen into each bin. The size of array BINS determines the maximum C number of bins a user may select for a test. (10,000,000 bins is a C typical number you may want to use if possible.) C C So steps are: C C 1. Check definition of one-dimensional array NBINS C that it's not too large for your account quotas C or system limitations. C C 2. Place your random number generator to be tested where it C bluntly says "PLACE YOUR RANDOM NUMBER GENERATOR HERE". C The output is an integer IRANDOM between 0 and NBINS-1 C (NBINS is the number of bins chosen by the user). C C Currently the program is set up to use subroutine RAN1, a portable C random number generator from the book "NUMERICAL RECIPES: The Art C of Scientific Computing". I've had trouble on our UNIX system C not making array R(97) static even though the code says to. C Compiling with the -static qualifier works. C C The random number generator being tested is used to "randomly" select a C bin for the ball to fall in, and the counter for that bin is C incremented. Note for a multi-dimensional test we calculate the C appropriate index into the linear array BINS by hand. After all the C balls are thrown we call the subroutines to do the heavy math. C C All the subroutines should be fairly standard FORTRAN-77 modified C versions of routines from the book "NUMERICAL RECIPES: The Art of C Scientific Computing" by William H. Press, Brian P. Flannery, Saul A. C Teukolsky, and William T. Vetterling, and you should look there for C further reference as to what the routines are doing. (Note: the book C comes in several programming language forms including C, PASCAL, BASIC, C as well as FORTRAN, so you can take your pick and rewrite this code in C any language you please.) C C Note: C CHIKSN.FOR is currently set up to be run by a process with a very C large page file quota (pgflquo). If you get a 'exceed quota' error C attempting to run this then all you need to do is change the line C which reads: C C INTEGER*2 BINS(20 000 000) !The bins. C C to something smaller like: C C INTEGER*2 BINS(1 000 000) !The bins. C C C To compile: C $ FORTRAN CHIKSN C $ LINK CHIKSN C or C % f77 chiksn.f !(Some UNIX F77 compilers require -save option) C C Sample run: C Test MTH$RANDOM in 3-D with 10 bins per dimension and 10 balls per bin: C C $ RUN CHIKSN C Input number of dimensions NDIM: 3 C Input number of bins per dimension NBINSPD: 10 C Total number of bins = NBINSPD**NDIM = 1000 C Minimum number of balls = 5*NBINS = 5000 C Input total number of balls NBALLS: 10000 C Input number of Chi-Square tests NCHITESTS (min=2) : 2 C Choose random number generator to test C (1) MTH$RANDOM, C (2) RANDU, C (3) ANSI C, C (4) Microsoft C C (5) Turbo Pascal C (6) DES C : 1 C Input starting SEED value: 1 C C BALLS= 10000 CHISQ= 993.0002441 PROB= 0.4524292 C BALLS= 10000 CHISQ= 974.0001831 PROB= 0.2915459 C KS D= 0.5475708 PROB= 0.5863269 C C----------------------------------------------------------------------- PROGRAM CHIKSN C Perform a CHI-SQUARE test on a sequence of sets of N random numbers C NDIM = number of dimensions C NBINSPD = number of bins per dimension C NBINS = total number of bins. NBINS = NBINSPD**NDIM C NBALLS = total number of balls. Should be at least 5*(NBINS**NDIM) C NCHITESTS = Number of chi-square tests to do. Must be 2 or more. C EBINS = Expected value for each bin. EBINS = NBALLS/NBINS C SEED = Initial seed value for random number generator C C Note 1: The maximum size of array NBINS may be determined by the users C page file quota (pgflquo in AUTHORIZE). Also, it is recommended C the user have a very large working set quota (wsquo,wsextent) C to reduce page faulting. This can greatly improve speed. C We use INTEGER*2 array here to save space. C C Note 2: The maximum number of Chi-Square tests that can be saved is C arbitrary (array SAVEPROB). The user may choose any value. C C Note 3: The user may choose any starting seed value. Some restrictions C may apply depending upon the particular random number generator C being used. For example, RANDU should always be started with C an odd value of SEED. MTH$RANDOM may be started with any value C of SEED. C C Note 4: The MTH$RANDOM generator is used by the VAX FORTRAN intrinsic C function RAN and the VAX BASIC function RND. It is defined as: C C SEED = 69069*SEED + 1 mod 2**32 C X = SEED/2**32 C C Note 5: The RANDU generator is obsolite due to very strong correlation C in 3d space. ( Prove to yourself using 65539 = 2**16 + 3 that C SEED[i+2] = 6*SEED[i+1] - 9*SEED[i] ). It is defined as: C C SEED = 65539*SEED mod 2**31 C X = SEED/2**31 C C The RANDU generator should be started with an odd value of SEED. C C Note 6: The C standard library function rand() is defined as: C C SEED = 1103515245*SEED + 12345 mod 2**32 C IX = SEED mod 2**31 C C This standard random number generator is defined in the book: C The C Programming Language C Brian W. Kernighan and Dennis M. Ritchie C Prentice Hall, 1978 C C The same generator is defined in the ANSI C version by the same C authors above, and the same generator is used in VAX C. C C Note 7: The Microsoft C version 4.0 library function rand() impliments C the following: C C SEED = 214013*SEED + 2531011 mod 2**32 C IX = bits 16-31 of SEED C C Note 8: The Turbo Pascal version 6.0 function impliments the following: C C SEED = 134775813*SEED + 1 mod 2**32 C IX = bits 16-32 of SEED C IMPLICIT NONE INTEGER NDIM !Number of dimensions INTEGER NBINS !Number of bins INTEGER NBINSPD !Number of bins per dimension INTEGER NBALLS !Number of random numbers per !chi-square test. INTEGER NCHITESTS !Number of chi-square tests to do C INTEGER*2 BINS(20 000 000) !The bins. (see note 1) INTEGER*2 BINS(200 000) !Less bins. (see note 1) REAL EBINS !Expected number of balls per bin REAL SAVEPROB(100) !Array to save results of !chi-square tests (see note 2) INTEGER*4 SEED(2) !Only SEED(1) result is ever used. INTEGER*2 W(4) !Seeds for RANDU EQUIVALENCE(SEED,W) !for RANDU COMMON / SEEDSTORE / SEED INTEGER I,J,K,MRANDO,NBYTES,CLEAR,CHITEST,INDEX,IRANDOM,NCLEAR REAL*4 FRANDOM, RRANDOM CHARACTER*8 TIMEBUF EQUIVALENCE (IRANDOM,FRANDOM) REAL FOR$IRAN !The RANDU random number generator REAL RAND !UNIX rand() INTEGER*4 xrand !BSD random() REAL RAN1 !test generator supplied REAL D INTEGER JISHFT,IRANDOM2,COUNT REAL RANDES !DES FUNCTION (not supplied) INTEGER KEY(2) REAL CHSQ,PROB !Chi-square value, !chi-square probability REAL*4 FNBINSPD !float(NBINSPD) REAL*4 TWO31F REAL*4 TWO16F REAL*4 TWO15F TWO31F = 2.0**31.0 TWO16F = 2.0**16.0 TWO15F = 2.0**15.0 C*DES KEY(1) = 12345 !Choose any number you want C*DES KEY(2) = 678901 !to initialize DES with C*DES CALL DES_INIT(KEY) !DES code not included. 104 FORMAT(' Input number of dimensions NDIM: ',$) 100 FORMAT(' Input number of bins per dimension NBINSPD: ',$) 105 FORMAT(' Total number of bins = NBINSPD**NDIM = ',I) 106 FORMAT(' Minimum number of balls = 5*NBINS = ',I) 101 FORMAT(' Input total number of balls NBALLS: ',$) 103 FORMAT(' Input number of Chi-Square tests NCHITESTS (min=2) : ',$) 102 FORMAT(' Choose random number generator to test'/, 1 ' /*(1)*/ xrand(),'/ 1 ' /*(2)*/ UNIX rand(),'/ 1 ' /*(3)*/ MTH$RANDOM,'/ 2 ' /*(4)*/ RANDU,'/ 3 ' /*(5)*/ ANSI C,'/ 4 ' /*(6)*/ Microsoft C'/ 5 ' /*(7)*/ Turbo Pascal'/ 7 ' /*(8)*/ DES'/ 8 ' (9) another random number generator (choose this one)'/ 6 ' : ',$) 107 FORMAT(' Input starting SEED value: ',$) 200 FORMAT(BN,I) C ***GET USER INPUT*** 10 WRITE(6,104) !Input number of dimensions READ(5,200) NDIM WRITE(6,100) !Input number of bins per dimension READ(5,200) NBINSPD FNBINSPD = FLOAT(NBINSPD) NBINS = NBINSPD**NDIM !Calculate total number of bins WRITE(6,105) NBINS !Total number of bins is... WRITE(6,106) 5*NBINS !Minimum number of balls is... WRITE(6,101) !Input total number of balls READ(5,200) NBALLS WRITE(6,103) !Input number of chi-square tests to do READ(5,200) NCHITESTS WRITE(6,102) !Choose random number generator to test READ(5,200) MRANDO WRITE(6,107) !Starting SEED value READ(5,200) SEED(1) SEED(2) = 1 !Used only if random number generator !uses bigger than 32 bits C INITIALIZE GENERATOR IF NEEDED C*XRAND CALL sxrand(SEED(1)) !Initialize xrand() CALL RAN1(-SEED(1)) !Initialize RAN1 generator C Calculate expected average number of balls for each bin EBINS = FLOAT(NBALLS)/FLOAT(NBINS) C CALL TIME(TIMEBUF) C WRITE(6,201) TIMEBUF C201 FORMAT(1X,A8) DO CHITEST=1,NCHITESTS C *** ZERO BIN ARRAY *** DO I=1,NBINS BINS(I) = 0 ENDDO C*VMS !Quickly set BINS(k) = 0, k=1,...NBINS C*VMS !Does the equivalent of above C*VMS !but a lot faster. C*VMS K = 1 C*VMS NBYTES = NBINS*2 !total number of bytes to zero C*VMS DO WHILE (NBYTES .GT. 0) C*VMS IF (NBYTES .LE. 65534) THEN !maximum number of bytes we can clear C*VMS NCLEAR = NBYTES !in one call to LIB$MOVC5 is 65535 C*VMS ELSE C*VMS NCLEAR = 65534 !max that LIB$MOVC3 can do in one call C*VMS ENDIF !(make nclear an even number so we can divide evenly by 2) C*VMS CALL LIB$MOVC5(0,0,0,NCLEAR,BINS(K)) !Clear a block of memory C*VMS NBYTES = NBYTES - NCLEAR !Number of bytes still left to clear C*VMS K = K + NCLEAR/2 !Number of bytes cleared so far + 1 C*VMS ENDDO C Main Loop DO J=1,NBALLS INDEX = 1 DO I=0,NDIM-1 C ***PLACE YOUR RANDOM NUMBER GENERATOR HERE*** C Set IRANDOM using whatever random number generator you choose C IRANDOM = integer between 0 and NBINS-1 c IF (MRANDO .EQ. 1) THEN c IRANDOM = INT( ( float( xrand() ) /TWO31F ) *FNBINSPD) c ELSEIF (MRANDO .EQ. 2) THEN c IRANDOM = INT( RAND(SEED(1)) *FNBINSPD) !UNIX rand() c ELSEIF (MRANDO .EQ. 3) THEN c IRANDOM = INT( RAN(SEED(1)) *FNBINSPD) !VMS mth$random c ELSEIF (MRANDO .EQ. 4) THEN c IRANDOM = INT( FOR$IRAN(W(2),W(1)) *FNBINSPD) !Infamous randu c ELSEIF (MRANDO .EQ. 5) THEN c CALL LIB$EMUL(1103515245,SEED,12345,SEED) !ANSI C c IRANDOM = SEED(1) .AND. '7FFFFFFF'X c IRANDOM = INT( FLOAT(IRANDOM)/(TWO31F) *FNBINSPD) c ELSEIF (MRANDO .EQ. 6) THEN c CALL LIB$EMUL(214013,SEED,2531011,SEED) !Microsoft C 4.0 c IRANDOM = W(2) .AND. '7FFF'X c IRANDOM = INT( FLOAT(IRANDOM)/(TWO15F) *FNBINSPD) c ELSEIF (MRANDO .EQ. 7) THEN c CALL LIB$EMUL(134775813,SEED,1,SEED) !Turbo Pascal 6.0 c IRANDOM = SEED(1) .AND. 'FFFF0000'X c IRANDOM = JISHFT(IRANDOM,-16) c IRANDOM = INT( FLOAT(IRANDOM)/(TWO16F) * FNBINSPD) c ELSEIF (MRANDO .EQ. 8) THEN c IRANDOM = INT( RANDES() * FNBINSPD ) !DES (not supplied) c ELSEIF (MRANDO .EQ. 9) THEN IRANDOM = INT( RAN1(SEED(1)) * FNBINSPD ) c ENDIF C Calculate index by hand. INDEX = INDEX + IRANDOM*(NBINSPD**I) ENDDO BINS(INDEX) = BINS(INDEX) + 1 !ball fell in this bin C IF ( MOD(J, 1 000 000) .EQ. 0 ) THEN C CALL TIME(TIMEBUF) C WRITE(6,302) J, TIMEBUF 302 FORMAT(1X,'AT BALL:',I,3X,'TIME=',A8) C WRITE(6,303) SEED(2), SEED(1) 303 FORMAT(1X,'HEX: SEED(2)= ',Z,' SEED(1)= ',Z) C WRITE(6,304) SEED(2), SEED(1) 304 FORMAT(1X,'DEC: SEED(2)= ',I,' SEED(1)= ',I) C ENDIF ENDDO 400 CALL CHSONE(BINS,EBINS,NBINS,CHSQ,PROB) SAVEPROB(CHITEST) = PROB WRITE(6,1) NBALLS,CHSQ,PROB 1 FORMAT(' BALLS=',I,' CHISQ=',F,' PROB=',F) ENDDO C Now see if all the chi-square values are chi-square distributed: IF (NCHITESTS .GT. 1) THEN CALL KSONE(SAVEPROB,NCHITESTS,D,PROB) WRITE(6,2) D,PROB 2 FORMAT(1X,'KS D=',F,' PROB=',F) ENDIF END C============================================================================ C From book NUMERICAL RECIPES: The Art of Scientific Computing C Here for demonstration purposes C Replace this with whatever random number generator you want to test C Initialize with negative number FUNCTION RAN1(IDUM) REAL R(97) SAVE R !(Some UNIX F77 compilers require -save option on compile) PARAMETER (M1=259200,IA1=7141,IC1=54773,RM1=3.8580247E-6) PARAMETER (M2=134456,IA2=8121,IC2=28411,RM2=7.4373773E-6) PARAMETER (M3=243000,IA3=4561,IC3=51349) DATA IFF /0/ IF (IDUM.LT.0.OR.IFF.EQ.0) THEN IFF=1 IX1=MOD(IC1-IDUM,M1) IX1=MOD(IA1*IX1+IC1,M1) IX2=MOD(IX1,M2) IX1=MOD(IA1*IX1+IC1,M1) IX3=MOD(IX1,M3) DO 11 J=1,97 IX1=MOD(IA1*IX1+IC1,M1) IX2=MOD(IA2*IX2+IC2,M2) R(J)=(FLOAT(IX1)+FLOAT(IX2)*RM2)*RM1 11 CONTINUE IDUM=1 ENDIF IX1=MOD(IA1*IX1+IC1,M1) IX2=MOD(IA2*IX2+IC2,M2) IX3=MOD(IA3*IX3+IC3,M3) J=1+(97*IX3)/M3 IF(J.GT.97.OR.J.LT.1)PAUSE write(1,100) R write(1,102) R(J) 100 format(f) 102 format(1x,'RAN1 = ', F) RAN1=R(J) R(J)=(FLOAT(IX1)+FLOAT(IX2)*RM2)*RM1 RETURN END C---------------------------------------------------------------------------- C CALCULATE THE CHI-SQUARE PROBABILITY. SINCE NBINS IS LARGE, IT IS JUST C THE CUMULATIVE GAUSSIAN DISTRIBUTION AFTER WE NORMALIZE THE VARIABLES. C OR ERROR FUNCTION. FUNCTION CHIPROB(NBINS,CHISQ) C Formula is the inverse of one given in Knuth for going the other way. INTEGER NBINS,DF REAL*4 CHISQ,Z DF = NBINS-1 Z = ( SQRT(24.0*CHISQ - 6.0*DF + 16.0) - 3*SQRT(2.0*DF) ) / 4.0 CHIPROB = ERF(Z) RETURN END FUNCTION ERF(X) C Return approximation to the complimentary error function erfc(X). C Return is not normalized err function. See book for details. C Adapted from book NUMERICAL RECIPES: The Art of Scientific Computing C Modified to return normalized error function erf(X) C (It's a polynomial approximation) REAL ERFCC,Z,T Z=ABS(X/1.414213) !Normalize T=1./(1.+0.5*Z) ERFCC=T*EXP(-Z*Z-1.26551223+T*(1.00002368+T*(.37409196+ * T*(.09678418+T*(-.18628806+T*(.27886807+T*(-1.13520398+ * T*(1.48851587+T*(-.82215223+T*.17087277))))))))) IF (X.LT.0.) ERFCC=2.-ERFCC ERF = 1.0 - ERFCC/2.0 !Normalize and compliment RETURN END C---------------------------------------------------------------------------- C THE FOLLOWING SUBROUTINES CALCULATE THE CHI-SQUARE VALUE: SUBROUTINE CHSONE(BINS,EBINS,NBINS,CHSQ,PROB) C Adapted from book NUMERICAL RECIPES: The Art of Scientific Computing INTEGER NBINS INTEGER*2 BINS(NBINS) REAL EBINS,CHSQ,PROB CHSQ=0. IF(EBINS.LE.0.) PAUSE 'CHSONE: EBINS must be > 0' DO 11 J=1,NBINS CHSQ=CHSQ+(BINS(J)-EBINS)**2/EBINS 11 CONTINUE PROB=CHIPROB(NBINS,CHSQ) RETURN END C============================================================================ C THE FOLLOWING SUBROUTINES CALCULATE THE KOLMOGOROV-SMIRNOV PROBABILITY SUBROUTINE KSONE(DATA,N,D,PROB) C Adapted from book NUMERICAL RECIPES: The Art of Scientific Computing C DF - degrees of freedom. Passsed to FUNC INTEGER N REAL DATA(N) REAL D,PROB CALL PIKSRT(N,DATA) EN=N D=0. FO=0. DO 11 J=1,N FN=J/EN FF=DATA(J) DT=AMAX1(ABS(FO-FF),ABS(FN-FF)) IF(DT.GT.D)D=DT FO=FN 11 CONTINUE PROB=PROBKS(SQRT(EN)*D) RETURN END C---------------------------------------------------------------------------- FUNCTION PROBKS(ALAM) C Adapted from book NUMERICAL RECIPES: The Art of Scientific Computing C Note the routine in the Numerical Recipes book erronously returns C 1 instead of 0 for large values of ALAM. PARAMETER (EPS1=0.001, EPS2=1.E-8) A2=-2.*ALAM**2 FAC=2. PROBKS=0. TERMBF=0. DO 11 J=1,100 TERM=FAC*EXP(A2*J**2) PROBKS=PROBKS+TERM C Error in Numerical Recipes book. Terminate if TERM underflows. C** IF(ABS(TERM).LT.EPS1*TERMBF.OR.ABS(TERM).LT.EPS2*PROBKS)RETURN IF(ABS(TERM).LE.EPS1*TERMBF.OR.ABS(TERM).LE.EPS2*PROBKS)RETURN FAC=-FAC TERMBF=ABS(TERM) 11 CONTINUE PROBKS=1.0 RETURN END C---------------------------------------------------------------------------- SUBROUTINE PIKSRT(N,ARR) C Adapted from book NUMERICAL RECIPES: The Art of Scientific Computing C See book for details. INTEGER N REAL ARR(N) DO 12 J=2,N A=ARR(J) DO 11 I=J-1,1,-1 IF(ARR(I).LE.A)GO TO 10 ARR(I+1)=ARR(I) 11 CONTINUE I=0 10 ARR(I+1)=A 12 CONTINUE RETURN END

2 1

(fwd) New ITAR Indictment
by Anonymous User 17 Dec '03

17 Dec '03

Are you serious? (fwd) From: chuckles(a)MCS.COM (Jason Skiles) Newsgroups: alt.security.pgp,comp.org.eff.talk,comp.org.cpsr.talk Subject: New ITAR Indictment Date: 8 Jul 1994 01:02:56 -0500 Reprinted with permission from the Chicago Sun-Tribune, 5 July 1994: >From Sun-Tribune Wires Washington, D.C.- In a Rose Garden press conference early this morning, Justice Department spokesmen announced the indictment of a Washington, DC. computer bulletin board system (BBS) operator, one Mr. Gil Bates, for violation of ITAR munitions export laws in connection with the operation of the infamous 'We got yer nudes here' BBS. According to federal authorities, many of the graphics files, or 'gif' files, named for their storage format, contained, when examined with a common hex editor, the value 0xAAAA, which can be used to encrypt data in such a way as to make it extremely difficult for intelligence and law enforcement agencies to recover the encrypted data. "0xAAAA first came to our attention a few weeks ago," explained Deputncrypted] file, by using a complicated algorithm known as 'exclusive or' and a 'mask' or encryption key, of, for example, "0xAAAA." Agents were stunned. "It's an entirely new and sinister turn of events in the field," said Miller. "We'd be helpless against such an attack." Miller went on to explain just how secure such a scheme would be. "When we need to crack some encrypted data, for example, email someone sends to his lawyer that we think may contain incriminating evidence," we usually just hand it to thenstitutional issues involved there, but we're working on it." He refused to elaborate. Once the technique was known, news spread quickly throughout the law-enforcement community. "This was a shot across the bow, a real wake-up call for us," said one Justice Department source. "We moved immediately, meeting with the vice-president and a professor from Georgetown. They were reluctant at first, but we mentioned organized crime and terrorists and they came around to our point of view." But why forbid the 'mask' or 'key' instead of the algorithm itself, the 'exclusive-or' technique? "That's sort of a funny story," explained Miller. "We were going to at first, but it turns out that the Clipper and Capstone chips [part of a government-designed key escrow system] make use of the algorithm in places. Of course, there were a few 0xAAAA's too, but the NSA assures us they've got a workaround." Bates loudly proclaims his innocence. "This is stupid, really stupid. It's just a 16-bit value, like any other. They can't restrict it. I'm... I'm at a loss. This is just too stupid to comment on." The case got weaker late this afternoon when it was revealed that the file in question didn't actually leave the country, but was retrieved by an FBI agent in Virginia. "Obviously this compromises our case somewhat," admitted Miller, "since Virginia isn't a foreign country. But someone in another country could have done the same thing, easily. That should count for something." Schneider noted that "Even though we'll probably have to drop the ITAR case, we did come up with something. He had a copy of PGP [a 'guerilla freeware' encryption package popular with subversives and criminals] and some files he'd protected with it." "We think they were maybe lists of children he an FBI agent in Virginia. "Obviously this compromises our case somewhat," admitted Miller, "since Virginia isn't a foreign country. But someone in another country could have done the same thing, easily. That should count for something." Schneider noted that "Even though we'll probably have to drop the ITAR case, we did come up with something. He had a copy of PGP [a 'guerilla freeware' encryption package popular with subversives and criminals] and some files he'd protected with it." "We think they were maybe lists of children he abused, or something," says Schneider. "Yeah, or something," added Miller. [Pre-flame apologies go out to those who object to the inclusion of 'serious' groups in the newsgroups list. If you know any silly ones where this would be more welcome, feel free to send it along.] ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ | Jason "Chuckles" Skiles | <Insert a clever quotation here.> | | chuckles(a)mcs.com | - <misattribute it here> | ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ | Any opinions I express are also those of every single school, company, | | and organization I've ever been in any way associated with. Honest. | ++++++++++++++++++++++++++finger for PGP public key+++++++++++++++++++++++++ ------------ To respond to the sender of this message, send mail to remailer(a)soda.berkeley.edu, starting your message with the following 8 lines: :: Response-Key: ideaclipper ====Encrypted-Sender-Begin==== MI@```%ES^P;+]AB?X9TW6\8WR:2P&2%`$A:^X<=%.A'J%;"Y7E2J[QT=&)]L M0`F:L=MI*O?R!?N6/E3TTZ6WF^B=ZP9][Y)B)J)4PF/%M3XOVYT^Y;!E*9Y9 $\U3XF@`` ====Encrypted-Sender-End====

1 0

Re: (fwd) Re: BSD random() - any good (source included)
by jamiel＠sybase.com 17 Dec '03

17 Dec '03

At 1:11 AM 07/08/94 +0200, Nobody wrote: >I guess cypherpunks would rather not write or even read code, but >simply gossip about usenet kooks and pedophiles. Nobody either - Is baiting for fun or - Actually thinks mailbombing 700+ people with easily accessible information on a subject tangental to this list is a Good Idea, and - doesn't have the slightest clue what proper conduct in a public forum is. For Nobody's sake, I hope it is the former. But if it happens to be the latter- Nobody, write me with an address I can reach you at and I'll send you lots and lots of code for you to read, every day, if you choose. That will help your productivity *immensely*. >c@lib@n -j -- "Blah Blah Blah" ___________________________________________________________________ Jamie Lawrence <jamiel(a)sybase.com>

1 0

cypherpunks mbone this sa
by Duncan Frissell 17 Dec '03

17 Dec '03

To: cypherpunks(a)toad.com K>The monthly meeting of cypherpunks will be broadcast on the mbone this K>saturday, july 9, from noon to 6. The ports and ID's are as follows: K> K>audio: 64264 51421 K>video: 43185 51863 K> K>This is being advertised using sd as well. Anyone in the New York area with a workstation? I'd love to listen (or watch). I'd bring beer! DCF --- WinQwk 2.0b#1165

2 1

Re: Mastercard, Visa, Access, Barclaycard, Amex, JCB ...
by jamiel＠sybase.com 17 Dec '03

17 Dec '03

At 11:20 AM 07/07/94 -0400, Paul J. Ste. Marie wrote: >The credit card swipers send the entire transaction in for both >approval and reporting. I'd imagine that the account number, merchant >number, and transaction amount are all there, since that's all needed. >Whether or not they encrypt the data I don't know. I believe modern card readers for at least MC/Visa use some form of encryption, but for backwards compatibility the central offices also work unencrypted. In the hotel I used to work in, the card reader certainly didn't encrypt. -j -- "Blah Blah Blah" ___________________________________________________________________ Jamie Lawrence <jamiel(a)sybase.com>

2 1

AA BBS case/update
by hkhenson＠cup.portal.com 17 Dec '03

17 Dec '03

[I wrote this for CuD, but decided the issue is urgent enough to post it other places before the next issue comes out. Apologies to Jim T.] ************* Re File 9 of the 5 July issue of CuD,--Some thoughts on the AA BBS, the Advocate writes: >Stuff about the AA BBS case. This case is essentially a war of ideas. >Can a backwards, pigheaded state like Tennessee set the moral and >cultural standard of a sophisticated state like California? So far they are making a serious attempt. >I say not, and like minded individuals agree with us. I would guess that 90+ percent of the people on the net agree, but they don't control the federal legal and police forces. >These "Reagan-Jungians" need to be beaten back. The best light is that >of the First Amendment. Bring the press in, point out the vital issues. While the press has been somewhat supportive, they don't get the connection between *their* presses and what an adult BBS provides. They don't understand the need to defend on the margins *before* you get your heart ripped out. ("First they came for the Pornographers . . .) >The judge will be embarrassed if the AP or Court TV is televising >what this action is about. Federal courts don't admit TV. The original bust with its obvious frame up of the sysop was reported on local (Bay area) TV. None of those involved seem to be embarrassed in the slightest. (Though Judge Brazil did remove himself from the case after being accused of serious breaches of conduct.) As a follow on what I posted mid January, it turns out that possessing kiddy porn is not a crime in the 9th district. The Excitement Video case in California ruled the law unconstitutional on appeal. Newsom (the TN prosecutor) specializes in porn and must have been up on this landmark case. The case was local to California, so the local prosecutors would have known about it as well. It has always been a mystery as to why postal inspector Dirmeyer did not have warrant for the kiddy porn he mailed to the sysop just before he came in. We now figure they left it off on purpose because even a corrupt judge who knew about the EV case would not issue a warrant for something he knew was legal! However, until the Supreme court rules on a case and unifies the law, possession of kiddy porn *is* a crime in the 6th district (where Tennessee is located). So, the sysop was indicted *there* after being framed for possession in California! Aside from the frame up, this raises the issue of: Can someone be charged with a crime in a different district of the country when what they did would not be a crime where it was done? The feds in Tennessee seem to think they can do it. (Ah, well. California has some odd notions of how far they can reach on things like sales and income taxes.) >Has anyone tried contacting the Playboy Foundation or the Guccione >Foundation. Contact people like Spider Robinson or William Gibson. >Publicity can only help. Yes Playboy Foundation, no Guccione. Playboy was marginally helpful. Does anyone have an address or number for any of those mentioned? >Especially given the candy ass tricks the prosecutors are trying out. Right you are! This is clearly political/religious persecution. (Does anyone know anything about the Conservative Caucus??) But what the hell can you do when the courts ignore their own rules and cater to the prosecutors? The court should dismiss this one on the speedy trial issue alone (40 days over the limit), but the judge has not ruled on several of the defendant's motions to dismiss, such as the NAFTA issue. The judge and prosecutor seem determined to break the defendant financially. For example, the last time Richard Williams (the AA BBS lawyer) went to Memphis for a hearing, neither the judge (Gibbons) nor the prosecutor (Newsom) assigned to the case showed up. All Richard could do was to turn around and come home with $2000 in plane fare and expenses down the drain. There is a hearing Friday, (July 8, 1994) at which the judge will ask Richard a single question--"are you ready for trial July 18" and to which he will answer "yes." This could be accomplished by telephone, but the judge said "show up or else." (Fortunately Richard was able to get a local lawyer to show up in his place and say "yes.") I very much doubt *they* will be ready for trial, since the judge has not ruled on the motions in anything approaching the time allowed by court rules. My bet based on watching this business since January is that the judge will stall till the trial starts, rule against all motions, and start a trial which will be overturned on appeal just to break the AA BBS sysop financially. There seems to be no rules against this vile misuse of judicial power--nor any forum in which you can complain. (Except the media--which is rather reluctant to support anyone whom the government has smeared with the "hot button" of child porn.) >Bring heat to Reno and Clinton. I haven't got a clue as to how to do this. I can't (and neither can anyone else who has tried) even reach Veronica Coleman, the local US Attorney, much less her boss Janet Reno. Actually, I feel for Clinton because there are likely people who *do* know how to hold his feet to the fire. My bet is that the NSA/CIA/FIB/XYZ knows (as someone put it on eff.talk) something Hillery does not. I am beginning to think that top politicians should fuck sheep and abuse children on live TV. Otherwise, those who know about their minor sins have an arm lock on them. J. Edgar Hoover abused the US Presidents this way for all of his long career. >If this case is to be tried, it should be in california. Judge Gibbons *did* rule on this one--denied. There wouldn't *be* a case in California. You can buy everything the AA BBS sysop was accused of selling within 10 blocks of the Federal Courthouse in San Francisco. >The Advocate. Keith Henson (who finds that the government disobeying the rules makes him itch!)

2 1

cypherpunks mbone this sat.
by Katy Kislitzin 17 Dec '03

17 Dec '03

The monthly meeting of cypherpunks will be broadcast on the mbone this saturday, july 9, from noon to 6. The ports and ID's are as follows: audio: 64264 51421 video: 43185 51863 This is being advertised using sd as well. The main topic will be discussion of Phil Karn's swIPe package, which was announced at Usenix last month. It allows one to set up encrypted IP tunnels over the exisiting IP network. For those of you interested in attending in person, the meeting will be held at Silicon Graphics in Cafe Iris, building 5. To get there, get off 101 at Shorline, and head towards the ampetheater. Turn right into the SGI parking lot after Charlston, and continue down to just before the circle. Building 5 will be on your right. NOTE: THE SGI PARKING LOTS ARE BEING RE-SURFACED THIS WEEKEND. SO YOU WILL HAVE TO PARK ON THE MAIN CAMPUS DRIVE OR ON THE CONCRETE PARKING AREAS. --kt Katy Kislitzin Silicon Graphics I/S Network Software ktk(a)corp.sgi.com

2 1

Question: Key Distr. in realtimeo applications?
by Jonathan Rochkind 17 Dec '03

17 Dec '03

Well, I've gotten a few hundred pages into Applied Crypoto, and am now educated enough to realize what the things are that I don't know. One question I have is regarding how protocols for realtime communications work, like say a encrypted voice conversation. I know there are such things,such as Clipper, but I assume Clipper/skipjack is atypical in many ways. But maybe I'm wrong. Anyhow, I'd guess that in such applications, a random one-time session key is generated for a symmetric encryption method. Is this correct? If so, how is this key distributed to both participants? You could use public-key cryptography in some way to distribute the session key, similar to what DES does, but then how do you distribute the public keys so as to avoid a man in the middle attack? Ideally, you want to pick up your crypto-phone, initiate a call to another crypto-phone which you've never called before, and which was possibly manufactured yesterday, and be able to exchange keys with it in a secure fashion. But I can't think of any way to do this, without opening yourself up to a man in the middle attack. How does clipper solve this problem? How do other potential realtime protocols? Or do they just ignore it, and assume that it's going to be too hard to do a succesful man-in-the-middle attack quickly enough for a realtime conversation? That would seem to me to be a dangerous assumption. I'd appreciate it if anyone could help me out.

4 3