er IEIIEI UWW 1 IBl51illililli ca 11115 11Dl5 f 111 11B l 1UllD UlUl1lU W Vl JCE W ilfB J PL 86-36 50 USC 3605 • • • ' ' • • I •••••••••••••••••• I • •••••• 5 I • • 7 I • •• • • • 9 r •1 METHOD OF MEASURING NEGATIVE INTELLIGENCE •• •1 -CAN'T TELL WHEAT PROM CHAPP • ••••••• • J Gurin 11 --- -• S -CROSTIC NO 18 • David H TJH tDOCUMBNT CONTAINS CODEWORD MATERIAL · Non - Responsive IIQP IE 12 13 14 15 16 17 19 Williams 20 I JNon - Responsive Publi$ ed Monthly by Pl Tecbni ues and Standards # • for the Personnel of Operations OCTOBBR 1978 VOL V No O l 18P 8• 1•- I •ttP 9161111• B l• M ---------------- _________ I _ _ _ _ _ _ _ _ _ _1 · A123 i ----·-· _ ___ · · • · '1 PL 86-36 50 USC 3605 1 October 78 • CRYPTOLOG • Page 5 T8F_MIOBT 19 1BM EO 3 3b 3 PL 86-36 50 USC 3605 _ TOP OIIORIIT HJ IDR1fs EO 3 3b 3 PL 86-36 50 USC 3605 fli§S October 78 CRYPTOLOG Page 6 T9P 810RIIT IJ HDlll't I No n - Responsive I PL 86-36 50 USC 3605 81t8RII EO 3 3b 3 PL 86-36 50 USC 3605 W1um the author of this azoticle BUbmitted it to•CRYP'rOLOG in April he stated that he had delibenztety omitted much of the mo re technlcat deta a such as hoc the probabi U fi IUNB are derived but offered to anm ier any qMsti ons that the published caiticle might•engender ll1l o 'tunatety that offer no ui ei- applies since the author Nsigned in lafe TU118 1978 Questions q hoTJever be di noted to his f OlfflBl' asso tes in W322 •on 37648 1 R D Bu • Collection Editoi- i1978 CRYPTOLOG We Iarticle in the April Gotta Accentuate G the Negative pointed out the problems that exist with intelligence efforts that yield negative results The challenge is to make the 1110st of such results and produce negative but useful intelligence This article will att t to describe a •ethod develo ed by October 78 CRYPTOLOG Page 9 t • Piii III C - ' ' ' t - if · · smsm • mama arr EO 3 3b 3 PL 86-36 50 use 3605 I OCtober 78 • CRYPTOLOG l Non - Responsive 2 1119 Pag 10 llliiRI 5 Iii I 52ii71P • L 1 £25 UilCP r _ - - • - - -v EO 3 3b 3 PL 86-36 50 USC 3605 YOU CAN'T TELL THE WBJ AT FROM THE CBAFP WITBOITT A PROGRAM J Garln RS -· ----------------- r1- --- · ator or linguist fhalyst or whatever you may call the human processor who determines what shoald be retained and-what discarded accounts fot almost all the df ision-makinr at this stage What' s to Jje done if we do not have enoug h voice linguists to do the selection rejection job Only ene course of action seems to holi out any lijtelihood of success and there are• no money- ack guarantees to that offer get • the machine to help Easier said than donii for a lp•t of reasons Just what is 11 the machine to do7 One kind of selection of course is embodni chart which follows attempts to show • ied in the decision whether or not to copy a in grossly oversimplified terms what criteria particular signal But once it is decided that th transcriber uses to select or reject and• the signal is worth taking a nd that signal what kinds of automatic devices might be devis 9d carries a lot of traffic we face the problem te approximate those functions The purpose i of disposing of what is not useful to us resenting this chart is to indicate which _ • This problem of volwne reduction has proven paths could be followed by the DDR organizatioa to be especially troublesome for voice inter- • in providing assistance to the selection prob-• cept Except for devices which recoenize • lem which is growing in size and importance • dialed telephone numbers Although the mechanisms have been listed separately there is no reason to assume that • __ _ -- _ I there are no automatic selection they would not be used together perhaps to re3ec ion devices currently available The supplement the transcriber's efforts if that transcriber-scanner or voice intercept evaluproved to be the most reasonable thing to do F ·or years there has been general agree ment that the right way to deal with huge volumes of any kind of intercept is to exercise selection as early as possible Ideally the selection process should take place at the point of intercept so as to reduce the load on co11111unications channels and to avoid glutting the storage resources at the proce1sing site Whether the volume reduction takes place at the very first stage of the SIGINT process or later on it must be done and the sooner the better I I ----------------- 1 -October 78 • CRYPTOLOG • Paxt 11 llttWIE HTI SSPSTPIT S ttlllFI C S 1¥ UNCLASSIFIED ll8A•of'Ntlo No ·1 8 By uest NSA-crostician David H Williaas Pl6 I PL 86-36 50 quotati on on the net page was taken fro pubUshsd IIIOl'k of an NSAer The fizt it ttBztB of the 'IIORDS apsti out the authol''B and the ti tu of the IJO 'k DBUNI'rIONS 'flOHDS October 78 C YPTOLOG Page 20 UNCLASSlnED use 3605 1 UNCLASSIFIED I PL 86-36 50 use 3605 J 39 B 53 W 79 N 80 Q 81 H 82 B 92 W 93 Q 94 R 95 U 212 0 213 G 214 L 225 N 226 Q 227 P Solution next month I I October 78 CRYPTOLOG Page 21 I • I --- - - -- ---- -- No n D H N UNCLASSIFIED - Re s po nsive _ - _- -·· •