W UV JW iJl1 $ WlEVU $W lD V l3lD l3 OOl3UlI1Jl3 f OOUl lbUi 11J OCT-NOV 1986 ______ ' ' 'I A NEW KIND OF JAPANESE U o o o o o o SIREN No 5 U o o o o o o A SUMMER LANGUAGE COURSE U o o o o o ON THE LIGHTER SIDE U o o o o o THE REPRESENTATION OF PREFIXES o o CONFERENCE REPORT U o o o oo TECHNICAL LITERATURE REVIEW U o SOFTWARE REVIEW U o o o o o o o o BULLETIN BOARD D o o oo oo LETTERS U o o o o o o o o FROM THE PAST o o o o o o RESULTS OF READERS' SURVEY U o PUZZLE U o o o o o o o o o o o o oo 9 o 12 o o 13 o o o o o o o o 15 o o oo o 16 o NoCo Ge r son o o o o 17 o J 19 i I _ 23 o o o o o o o o 8 14 22 o ' ' o o o o o o o o 27 o o o o o o o o 31 o o o o o o o o o 32 o o o o o o o o o o o o o 33 'fHIS BOEURUMEN'f 30N'flaIN8 CODEWORD Mla'f aIA NOT RELEASABLE TO CONTRACTORS -f9P--SEGREJ L 86-36 CLASSIFIED 8Y NSA CSSM 123-2 DECLASSI Y O A eAey's OrigiRiltir lg DeterRCIiflatiefl ReEJtJireei DOCID 4010004- L 86-36 Pljblishedby Pl Techniques and Standards ' ' -'-- VOL X Ih N s 10-11 ' PUBLlS 7 ' October-November 1986 - - 1'-- _ BOARD'OF EDITORS Editor ' o ' -_ ---Jt963-11 03 Computer C o lIection Systems o o o o o o o o o o o o o o o o o ooo o o o ' Cryptanalysis o Cryptolinguistics o o o o 963-1103 963-5877 963-5238 963-1596 Index ' ' ' k963-5292 Information Science oooo 1 1 972-2268 Information Security George F Jelen 859-1211 b 1i963-384S Intelligence Research 1 Language 1 963-3057 r63-5566 Mathematics Puzzles ' ' 9 6 3 - 6 4 3 0 Science and Technology ' 968-8075 Special Research Vera R Filby 968-8014 Traffic Analysis ' ' Robert J Hanyok 963-5734 1 To submit articles or letters by mail send to Editor CRYPTOLOG Pl HQ SAlS7 If you used a word processor please include the mag card floppy or diskette along with your hard copy with a notation as to what equipment operating system and software you used via PLATFORM mail send to cryptolg at barl c05 bar-one-c-zero-five note no '0' in 'log' SHARE THE WEALTH U Even in the best of times not everyone who could profit from it has been able to go to conferences Partly it's because there's never that much money But it's also because we can't shut down a whole operation for a conference -- someone has to mind the store Nowadays money is very tight So people who are lucky enough to get to go to a conference ought to share the wealth and write it up for CRYPTOLOG in addition to the customary trip report Why write it up again when you can specify the distribution of each trip report and reach the right people Because the probability i very great that there are people who should be informed that you don't know about The Agency has gotten very large and a good number of employees going to conferences haven't been around long enough to have a feel for who might be doing what Also our endeavors are becoming more complex and interdisciplinary and the logical distribution may miss some key projects that are related Only the fictional Mycroft Holmes could keep on top of things all by himselfl So write it up for CRYPTOLOG and increase the probability that your trip reports reach the people who would benefit And the chances are that yo'u'll get something out of it too -- making connections with people on similar projects or with like concerns Always include your full name organization and secure phone number For Change of Address mail name and old and new organizations to Editor CRYPTOLOG Pl Please do not phone Contents of CRYPTOLOG should not be reproduced or disseminated outside the National Security Agency without the permission of the Publisher Inquiries regarding reproduction and dissemination should be directed to the Editor I CORRECTIONS I I The caueat NO CONTRACT should be placed on the couers ofthe Aug Sep 1986 issue I The article titled Collection Management in the same issue should be classified CONFIDENTIAL rather than CONFIDENTIAL CCO FOIt OP1'IEURIAL USB OtHN DOCID 4010004 EO 1 4 d SSORET SPOI S I i Oct-Nov 1986 CI YPTOLOG page 1 SISOIlET SPOKlG DOCID 4010004 EO 1 4 d SElGHffiT SPOKS OeL-Nov 198 CRYI'TOLOG page SJ4 OltB'l' SPOlCB 2 DOCID EO 1 4 d 4010004 SECftE't' SFOKE CRYPTOLOG Sf130UBT SPOK Oct-Nov 1986 page 3 DOCID EO 1 4 d 4010004 SHOUS'P SPOICH Oct-Nov 1986 CRYPTOLOG page SEOUS' ' SPOKS 4 DOCID EO 1 4 d 4010004 ECllET SPOIH3 Oct-Nov 1986 CRYP'l'OLOG page 5 BCUliJ'l' SPOI lt EO 1 4 d DOCID 4010004 Si-3CnE't' SPORB CRYPTOLOG page 6 SIWIUil't' SPOnEl Oct-Nov 1986 E10 ELLE DOCID 4010004 Oct Nov 1986 page 7 811gl ll1rll E01 4 d DOCID 4010004 S CftEr t' SFOKE BULLETIN BOARD II I DATA CONVERSION U PODO The Data Conversion Center formerly known as JOBBER can convert your hard copy to machine-readable form using an optical scanner or a keyboard At the same location the Magnetic Media Conversion Center can also in most cases convert data from one machine-readable form to another For information about specific conversion pairs call I IT1431 963-4777 FORMATIONQF HP3000 USERS' GROUP U U In response to many requests an NSA Users' Group for the HP-3000 is being formed Programmers as well as end users are invited to join For information please calli I I tr2136 96 3 i 748 P L 86-36 CRYPTOLOG SIWltB'l' SPOIH OcL-Nov 198B page 8 OCID 4010004 BROREIf' A NEW KIND OF JAPANESE U I -if ' I - t -t' I o HlT'ADH' H pn P L December 1986 o CRYPTOLOG o page fo0 1 4 c St- OlUYF P L 86-36 Hl'''WD VIA GO H P'tT OILYHfELS OULY 86-36 r ' ' ' ' DOCID 4010004 ' EO 1 4 c P L 86-36 SHGIlB'I' December 1986 CHYPTOLOG page 10 SI'WRK'I' IIAHBLE VIA eO'PtHPf ' OIIAPfP'fE f S OP'fLY DOCID 4010004 P L 86-36 EO 1 4 c December 1986 CltYP'l'OLOG page 11 HA mLH Sl ell 'p VIA GOMUt' ' GIIAPUfSLS OPtLY 4010004 g Clt hl'i P L 86-36 EO 1 4 c L ---_------- P L 86-36 EO 1 4 c Dcccmber 1986 CHYPTOLOG pagc 12 SHCUET IIAtmLEl YlA GOllH tT GnA t rKbg gP'i b DOCID 4010004 gCglS'F SPOKS A SUMMER LANGUAGE COURSE U P L 86-36 U I have just returned from a great experience dinner that night I met the guys that would be in I was at the Indiana University campus in my class There where 12 of us all together in the Bloomington in an intensive Chinese language beginning class All the Chinese students and program It was a fantastic two months teachers ate at special tables The Japanese and Korean students also had there own tables This FOUD The opportunity came about rather was a great opportunity to talk to the teachers and suddenly in May while I was working as a clerical practice Chinese as time went on assistant in B61 I came to the agency as a Senior Clerical Assistant in 1985 but with a desire to get U Classes started on Friday morning We had into a language program It did not take long In four classes a day each an hour long with an hour the second week of May a memo came in to the between classes That hour between classes was office about an intensive East Asian Summer usually spent studying for the next hour's class Language program being offered in Indiana I The two hour lunch also was usually spent knew this was exactly what I was looking for I studying The dorm cafeteria offered sack lunches thought my chances were slim but after talking to so we did not have to waste time walking 20 the chief and the deputy chief of my division I minutes each way for lunch decided to submit my name After about two weeks I was told that there was a very good possibility U At the end of the day we would all walk back to that I would be able to go During the last week of the dorm It was a 20 minute walk along tree lined May I was informed that I was going to Indiana to paths over wooden bridges and along a stream begin an intensive Chinese language course on This was an excellent way to relax after a day of classes After dinner and about half an hour of June 13 relaxation and reading the newspaper we resumed U When I arrived at the university I was studying about four hours a night One of our overwhelmed by the size of the campus It was native Chinese teachers had the room two doors beautiful I went straight to the dorm to check in down from mine If anyone had a question All of the language students stayed in the same anytime she was always eager to help The dorm the women on one floor and the men on teachers seemed to be truly interested in each another It was one of only two dorms on campus student' sprogress They were always around that was air-conditioned This was nice in the when someone had a question or a problem They middle of the summer I was told there were had office hours in the evening when the students television movie and exercise facilities available could go and ask questions or just sit and talk It in the dorm I do not know if this is true because was nice to be able to sit and use Chinese one-onafter the first day there was no time for any of one with the teacher these things U All of our time was not spent studying though U I met the other girls who would be in my class Every morning at 0630 there was Tai Qi Chinese that afternoon and we all became fast friends At shadow boxing for anyone who wanted to learn Oct-Nov 1986 CI YPTOLOG page SHCnK' ' IlQK 13 4010004 EeIUft' There was also a Chinese drama class calligraphy and drawing and singing There were Chinese movies with English subtitles for beginners on Friday nights Occasionally there was a party after the movie where we could relax and dance after a hard week of study Saturdays were usually spent doing all the things a person did not have time to do during the week There were trips to the store to get away from campus for a while laundry cleaning and studying There were volleyball games between the different classes Sundays were usually spent studying all day Every second or third Sunday the teachers tried and do something special Since there was no dinner served at the dorm on Sunday they cooked their own Chinese f od and we would have a picnic or just get together and talk Then it was back to the dorm to study I' SPOlfE L 86-36 BULLETIN BOARD' SLAVIC TERMINOLOGY U rouo Now in distribution is the third printing of the ever-popular Comparative Bulgarian Russian Polish and Czech Terminology 8-203 185 PI Language Publications No 1 January 1972 compiled by The multilingual format provides a base for a comparative method of U I kept a diary while I was there When I look solving translation problems at it now I see that just about every other word is study But for every study there is a fun and that is what I remember most Although there was --_---I For a copy write to a lot of hard work the teachers and students made IP16 lIQ A187 Phone orders it all fun are not accepted P L 86- 3 6 I f 1987 LINGUISTIC INSTITUTE U U The Linguistic Society of America th e Association for Computational Linguistics and the American Association for Artificial Intelligence are co-sponsoring the 1987 Linguistic Institute at Stanford University from June 29 to August 7 I987 The theme is Contextual and Computational Dimensions of Language The Institute includes conferences and seminars of various lengths from a few days to 6 weeks with intensive courses in __- - -- - I But at this time I am unable to do much I still have a few years of study before I become effective in my job But I look forward to it because itis definitely much more interesting and more challenging than being a secretary I think I am going to like this L - FOUO I found out when I returned to work that sending a beginner to Indianawas an experiment on the Agency's part I was the first person to go there to take a beginning course I think the experiment was a success because there is talk of sending more students next summer I hope this is true because it would be great to see more people have the experience I did And I must admit that I hope it is true because I would like to go back next year and take a more advanced course I think even if the Agency does not send me next summer I will go on my own for a long vacation spent studying and having fun 0 EO 1 4 c P L 86-36 P L Oct-Nov 1986 African languages Arabic morphology and linguistic and language processing courses on syntax semantics and phonetics The aim of the conference is to integrate linguistic theory I with artificial intelligence psycholinguistics discourse analysis and computational linguistics For further information colltl l ctl r16 963 103 0 86-36 CItYP'l'OLOG o page 14 SB6U1S'1' SPGKK I EO 1 4 c P L 86-36 DOCID 4010004 ON THE LIGHTER SIDE U A test paper making the rounds in the National Cryptologic School is rolling them in the aisles Unlike most funny exams the answers were right You too might enjoy reading the unorthodox responses orl Ian R U15A student to questions on Russian History Question 1 Describe the reign of the following Russian Tsars What did they do which was significant for Russia and what events took place P L 86-36 A Ivan IV B Peter 1 C Alexander I D Alexander II E Nicholas II Answer A Ivan IV The Terrible One hard-core tsar He got the landowners boyars and the church to fork over some cash to help run the country Created a 6 OOO-man political police force to help persuade his populace to follow the straight and narrow Freaked out and killed his son during a tantrum Alexander's troops beat him back to France and made Russia a European power to be reckoned with Only head of state that showed up at the Congress of Vienna But still your basic tsar on the homefront D Alexander II The Tsar Liberator Freed the serfs technically in 1861 Real liberal guy he tried to push through a bunch of reforms but it wasn't fast enough and some hot-headed terrorists blew him to smithereens E Nicholas II The last tsar This poor slob was hen pecked by a warhorse of a wife Wanted to be big and tough like his dad but it was no go Blundered into WW 1 After a few uprisings he gave up some power forming the Duma But it was too little too late Overthrown in 1917 he and his family met an untimely end See Anastasia starring Ingrid Bergman Question 2 Why was Lenin successful in the October Revolution when all other revolutions failed What was the shortcoming of the' Provisional Government Answer The Germans secretly shipped Lenin B Peter I The Great Tall dude who liked the into Russia to get the Russians off their backs It worked The provisional Government and West Modernized the army and created a Kerensky didn't pull Russia out of WW I navy He was a tsar who knew how to profit which was probably why the October Revoution from Western technology But no liberal was took place in the first place AWOL soldiers he He did nothing that would weaken his workers and your basic Joe Blowsky didn't absolute rule Serfs remained chattel take kindly to going back to the front and so C Alexander 1 Napoleon came saw and Lenin moved in and with the treaty of Brestbarely escaped with his brandy intact Litovsk got Russia out of the war CI YPTOLOG page i5 1'0 it Oi'l'lGlftL USJoii aptLY Oct-Nov 1986 DOCID 4010004 Question 3 What was the New Economic Policy and what was its effect on the Soviet Union Answer In 1922 after the Civil War the Soviet Union's economy was a shambles So Lenin devised this new economic plan which was nothing less than a good strong dose of capitalism It worked and in 1927 everythingh was back to its pre-Civil War level Question 4 What steps were taken by Stalin to set the country back on the road to socialism What problems were encountered What effect did the purges of the 1930's have Answer Stalin came in and ended the NEP and put socialism back on its dreary path Not too many people liked that So he got rid of them Exiled and or killed millions Decimated the officer corps so that when WW II came arond his leadership cadre was wanting Serves him right Question 5 What were the major events during Khrushchev's regime Answer Let's see There's the denunciation of Stalin that put the Soviets in a tizzy and really upset the Chinese His Virgin Lands project to expand agriculturally into the vast regions of eastern Russia was a big flop Gary Francis Powers took a nose dive in his U-2 and with that and so did relations with the U S And in 1962 he tried to sneak missiles to Fidel and blinked when Kennedy stared him down Banged his shoe at the U N Visited Disneyland and was shocked by Shirley Maclaine on the set of Can-Can in Tinsel Town 0 REPRESENTATION OF PREFIXES u fHD Provided b P L 86-36 REPRESENTATION PREFIX Factol byWhich Unitis Mutliplied International Symbol Common Use Symbol Wouble Case FORM I FORM Single Case Lower Jl Single Case Upper exa 1018 E E ex EX peta 015 P P pe PE tera 10 12 T T t T giga 109 G G g G mega 106 M M rna MA kilo 103 k k k K hecto 102 h h h H deka deca 101 da da da DA deci 10-1 d d d D centi 10-2 c c c C milli 10-3 m m m M micro 10-6 11 11 u U nano 10-9 n n n N pico 10-12 p 10-15 f P f P femto P f atto 10-18 a a a A Oct-Nov 198G 'on CIWPTOLOC page 16 OPJ'IOIAL USI OP'JLY F DOCID 4010004 SBOUH'I' I o CONFERENCE REPORT TALC 1986 oio o o o o o o o o EO 1 4 c EO 1 4 d P L 86-36 Target o o o Aquisition and o o o Location o o Conference o o o 86 s oeo o o o N C Gerson W3 o 'f'ltis cr tiele js clussified SECRET 'lYSeD tP'l fes tir8ty Oct-Nov 1986 Cl YPTOLOG page 17 SECUj ij'I' HANDLE VIA OOMHi'f' OHAUUELS OUbY o - o Sl 10UH'P 4010004 EO 11 EO 1 4 d P L 86-36 Oel-Nov 19SG CRYI'TOLOG page IS SECltt03'f IIA mLE VIA GOMIWI' GIIA T TEL8 O TLY CONFIBHNTU L Technical Literature Review A Trigraphic Cipher with a Short Key for Hand Use by Joseph R KruskaI Cryptologia July 1985 P L 86 '36 ReviewedbYIL - r2 r3 -rl 1 P12 In this article the author presents a simple Playfair-like cipher which is as he states easily used for manual enciphering and is genuinely trigraphic 1 THE SYSTEM The author's scheme requires that the number of rows m and the number of columns n are both odd As an example let us take m n 5 combining J with I and construct a Playfair square with an appropriate keyword T 00 01 02 03 04 o L GAB 10 11 12 13 14 D E H I 20 21 22 23 24 Q S 30 31 32 33 34 W X Z 40 41 42 43 44 F K M N u V P C2 C 3 c t r l r3 -r2 c1 c3 -c 2 r t r 2 - r 3 c1 c 2 - c3 where the calculations of the components are carried out mod m n To illustrate if the plaintext trigraph REF is to be enciphered we 0 1 E 2 1 F 2 2 and write R calculate the ciphertext coordinates as 2 2-0 1 2-1 0 2-2 1 2-1 0 2-2 1 1-2 This segment ofthe review is unclassified CRY corresponding ciphertext trigraph has coordinates which mod 5 5 is 4 2 0 2 0 0 resulting in the ciphertext trigraph WYC The deciphering procedure is only a little more complicated If the ciphertext trigraph is RI'C t R2 C 2 R3 C 3 then the transformation which is inverse to the enciphering transformation produces the plaintext t R 2 R3 t C 2 C3 t R I R3 teC I C3 n t R I R2 t C I C 2 as can readily be checked For example The second square shows the row and column coordinates of the 25 alphabetic characters Both the plain and cipher components are alphabetic the coordinates will however be useful in calculations If the coordinates of an on-cut plaintext trigraph are r l c 1 r 2 c2 r3 c3 then the Oct-Nov 1986 t R2 R3 l r l r3 -r2 r t r 2 -r3 r t The form of the decryption operation shows why it is that we must insist that both m and n are odd If m were even then t would have no meaning But for m odd t makes perfectly good sense t is that number which when CRYPTOLOG page 19 eON It'll ISN't'IAL IIANDLEl YIA GOMI ttp OIIA U ElLS O LY Eo 1 4 aCID 4010004 L c multiplied by 2 yields 1 mod if m 5 5 then t cor07 lOEN'i'lAL 86-36 ffi So for instance 3 since 3 2 6 1 mod Kruskal remarks that if plain ABC corresponds to cipher XYZ then we also have correspondences between the pairs BCA and YZX CAB and ZXY CBA and ZYX BAC and YXZ and ACB and XZY ad-Nov 1986 CHYPTOLOG page 20 60NFII ENTIAL IIA mLB VIA 60MHF 6IIld'Hf8LS ONLY EO 1 4 c DOCID 4010004 P L 86-36 Oel-Nov 1986 CHYPTOLOG page 21 eOPH'lIlBP 'I'IAL IIAUDLE VIA COMHi' ' GIIA HiELS OPlLY DOCID 4010004 1 4 c P L 86-36 60NFIDEN't'IAL BULLETIN BOARD FORMATION OF XEROX USERS' GROUP U U A Users' Group is being formed at NSA for users of the Star Viewpoint and XDE Interested persons should calli Y44 972-2345 ------ - -P L 86-36 FOREIGN LANGUAGE VIDEOS U FGYO Videotapes in Russian and German are shown every Thursday at lunchtime in a c onference room They feature news and documentaries Russian tapes are shown 11 0012 00 and German tapes from Austria and East and West Germany are shown 12 00-1300 U Programs and meeting rooms are posted on the CLO Bulletin Board on the north wall by the escalators near the entrance to the cafeteria POUO For other information about the German videos c a l l P 1 6 963-1103 and about the Russian videos A2COG 963-1180 '---- - ' '- P L 86-36 TERMINOLOGIES U FOUO Several terminologies and glossaries in SIGINT and related specialties are being compiled by P13D For a list of publications in the planning stage or already completed call or writel IpI3 968-816l P L Oct-Nov 1986 CRYPTOLOG page 22 COfWIII lilN'fIAI IIAUDLE VIA COMIU'f CiIAtoflq'EL8 OIq'LY 86-36 DOCID 4010004 TOP SKGRK'F P L 86-36 SOFTWARE REVIEW P13 FOUR PC-BASED EXPERT SYSTEM SHELLS U o ExSys o Personal Consultant o Guru o Expert-Ease U Are any of the many expert system packages advertised really useful Is even the best PC package suitable for building expert systems for NSA applications FOUO Overall I found that all the PC-based expert system shells have some potential for NSA applications With the abundance of ASTWs in NSA and at its field sites expert system shells have considerable potential for U In general one might expect assisting both junior and senior analysts with complex analytical decisions Greater potential o cheap easy to use products providing roughly the same set of features as in Mycin exists in areas where personnel rotation creates an expert system research project of the a training problem and round-the-clock 19705 with the rule capacity limited by the operations make it difficult for even an expert PC's memory and disk storage to keep track of the myriad of constantly o limited processing speed and either no changing detail Field sites NSOC and extensibility or very limited extensibility DEFSMAC are excellent candidates for expert o a variety of user interfaces systems which assist with important timeo generally poor documentation sensitive decisions about analysis and reporting of observed phenomena particularly when there U In the four packages tested I found are conflicting data from multiple sources o more features and a greater variety than I U Following are specific comments on the expected though some of the basic Mycin individual packages features were not available in some packages ExSys $300 o better user interfaces but disappointingly some aspects of the user interfaces seemed D ExSys is the easiest to use of the four needlessly complex or difficult expert system shells I tried and it has some powerful features as well though it does has o generally more capacity than I had some limitations believed possible o documentation in some cases even worse than I had expected Oct-Nov 1986 D When you start to build an expert system with ExSys the first activity is to define a rule The rule constructor requires qualifiers CRYP1 'OLOG page 'FOP s sonwl' 23 flAN'flfjE ViA COMfU'P CIIAfHf8LS OULY EO 1 4 c 4010004 86-36 TOP SI 30RF3'P and choices for the IF and THEN portions of the rule While you are in the middle of making the rule you go off to the side and build a qualifier which is an enumeration data type whose name is the subject and verb of a sentence and whose values are possible objects for the sentence When you have built a qualifier you select one or more values for the rule you are building ExSys automatically builds a rule that reads like English Once a qualifier is built it can be used in other rules everything is integrated nicely Before you know it you have built an expert system U One particularly nice feature of ExSys is the optional ELSE part to a rule This does not appear in any of the other three expert system shells reviewed U Some of ExSys's shortcomings Qualifiers cannot have certainty factors if you must deal with uncertainty this may not be an acceptable system Choices the final output for ExSys must have probabilities associated with them and these probabilities will be averaged if multiple rules produce the same answer wi th uncertainty That is as far as ExSys goes with uncertainty no uncertainty is allowed in the IF portion of a rule The manual is skimpy in some of its explanations For example the order of rule selection is according to the numerical order of choices I found that out not by reading the manual but by rearranging the choices in alphabetical order and thereby messing up my system o The on-line explanation merely displays the current rule This is not as useful as Personal Consultant's ability to display the entire chain of reasoning on-line Texas Instruments Personal Consultant $950 U Personal Consultant a commercial implementation of the EMycin system has several of its nice features o enumeration data types very handy for user interaction The user can select from a list of values instead of typing them in thereby avoiding mistyping o runs under Lisp it is extensible in Lisp and offers Lisp's advantages for processing symbols e g text This has a dark side also every 15-20 minutes the Lisp system stops to do its garbage collection a oneminute interruption on my PC o multiple rule contexts to help organize the rules o comprehensive explanation capability both during a session and after the session is concluded o an intuitive certainty factor system U Personal Consultant has some surprising problems the documentation is badly organized and hard to use a few features that I had not liked in the mini-Myein package a reunchanged in the TI package and worst of all it requires a key disk in drive A one of my pet peeves U Some other complaints about Personal Consultant Oct-NoY 198G EO 1 4 c P L 86-36 CHYPTOLOG page 24 TOP SISOUB'l' IIAPWhS YIA GOMIt ' GHANPfELS DULY 4010004 TOP SKCUK'l' o The rules are given names then when the rules are listed on the printer the names are omitted and numbers are used o A user with no Lisp training will find some of Lisp's ideas confusing for example parentheses all over the place and the ubiquitous garbage collection Non-Lisporiented users would have trouble even with the installation U Mycin does offer a lot of power for building expert systems and the Personal Consultant offers that power in a nice package except for the copy protection CPS eeO Personal Consultant would be better than ExSys in an application with a complex knowledge structure It also has graphics to display either input data or to explain its recommendations U After running through the demonstrations I tried to build a small five-rule expert system of my own This took about ten hours I had previously built the same five-rule system with ExSys in two hours and with Personal Consultant in 2 hours and I later built an equivalent system with Expert-Ease in 2 hours U Some specific problems o Inadequate error handling In ignorance I used display a reserved word in Guru both as a variable and as a rule name The compiler did not tell me that I had used a reserved word When I tried to run the erroneous expert system Guru didn't stop me the message misplaced LOGIC appeared on the screen I knew something was wrong but the error message gave me no clue whatever o Incomprehensible description of certainty factors in the manual Although the claim is made that there are 16 ways to combine certainty factors there are actually four very similar probabilistic ways that can be paired to make 16 combinations U A new version called Personal Consultant Plus $3 000 has graphics metarules frames DOS program execution the capability to insert a procedure to determine the value of a variable and the capability to insert a procedure to be executed when a variable changes GURU $3 000 U The documentation of Guru is bulky The four manuals -- a two-volume reference manual and two user's guides -- are intimidating I found almost no use for the two user's guides o Lack of enumeration type that would allow you to specify a limited set of string values for a variable e g Sunday Monday Tuesday Wednesday Thursday Friday Saturday You should be able to access this through a menu function missing in the current version of Guru rather than having to write a subroutine U Guru is difficult to use but broa d in its applicability It would probably be good for a rule-based application for which access to a data base or spreadsheet or remote computer file is important fi'S-OOO When the bugs are worked out Guru may be useful for complex data base retrievals or for help with decisions based on information in a databas 1 Oct-Nov 1986 CRYPTOLOG page 25 TOP Sf JCIW'l' IL ' DLE VIA GOMI 'lCP CHAPUfSLS OptLY P L 86-36 4010004 DOCID I I 86-36 TOP clu 'r I some applications in which it would fail miserably U My second choice would be Personal Consultant For general power without respect to cost for a Lisp connection for good certainty EXPERT EASE $495 U Expert-Ease is a different kind of expert system shell It builds rules using examples given by the user a technique developed by Donald Michie in Scotland This appears to be a very easy way to do it It was surprisingly easy to set up the toy system I used to test all the packages U It was easy learn and to use and it does allow linkage of multiple rule sets and does deal with uncertainty in a limited way three categories but it has several major defects it has no high-level language connection and it is not extensible it provides no explanations of its reasoning chain factors and for explanations it is the best of the four Personal Consultant Plus offers even more features metarules graphics frames external DOS programs although at a higher price U Guru has a number of problems but its concept is a good one It is the only shell I know of that integrates a data base and spreadsheet with the expert system it also has a natural language interface not evaluated graphics and a communications interface A new release will be available by the end of 1986 U Expert-Ease is the only product to build an expert system from user-provided examples Other than that there is nothing to recommend it 0 it has no reference manual The manual is a tutorial without an index U Expert-Ease runs under UCSD Pascal and requires a Pascal partition to run it on a harddisk system causing problems for someone who wants to run it alongside DOS or Unix programs The lack of extensibility and the lack of explanations make it inadequate for the kinds of things expert systems are noted for It might be useful for someone who wants to experiment with building an expert system using examples instead of constructing rules SUMMARY U Of the four systems reviewed ExSys offers the most power for the money It is superior to the others for ease of use power cost and external program connections But there are Oct-Nov 1986 - _ _ - - _ _ _ - _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ REMINDER I I I I I I I you want to be sure of receiving i 'CRYPTOLOG after you move or are reorganized send a change of iaddress to Editor CRYPTOLOG i iPl and include your name and i both old and new organizations i If I I I I I I I I I L _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ CRYPTOLOG page 26 '1'01' SIWH8'1' IhUWLB YIlt OOMHf'P OIIA HH3LS ONLY DOCID 4010004 SHeREif' - _ _-_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _-_ _ _ _ _ B L 86-36 1'0 the Editor the way NSA was funded The good old days ended about that time and the program petio 1 article on collection elements and subelements began Each part of management Aug-Sep 1986 is thought- the Agency was funded to perform specified provoking l' i like to know if the situation as functions each had to justify its existence and Ihasportrayed it is so Has the NSA its budget every year Consequently each collection manager become a wart on the insisted on exercising complete control of assets smooth skin of SIGINT operations Do other that were associated with the targets for which collection managers feel asl Idoes the subelement was funded There suddenly was no longer a pool of intercept positions that S-CCO For the younl er readers a little belonged to the Agency and that could be lhistory may be in assigned as required to perform functions that expansion otl order Before he arrived elements of the needed to be done I Agency within Production PROD included General Studies GENS Asian Communist Studies ACOM ' J That was when GENS became A I 1 I and All Other ALLO There was another element that was specifically concerned with the management of collection resources Collection COLL although it was charged with other responsibilities as well In fact shortly after the element had been organized it changed its name to Collection and Signals Analysis OOSA to reflect more accurately the element's responsibilities Even that name fell short It omitted consideration of Non-Morse General Search COSA-41 Morse General Search COSA-42 and others ACOM became B and so forth And that was when the collection managers who had managed the Agency's resources found themselves often in conflict with subelement managers who insisted on managing the resources they had to be responsible for At this time the role of the collection manager began to be eroded to the state described by I I rFQTJO Please note that I am not disputing the events and facts described by I I Those dates are true I am just mentioning additional factors that were involved in the evolution of the role of the collection manager There are probably other factors as well The 8 GOe That's enough of the gratuitous point that I am trying to make is that it is not history lesson Other readers with better a simple problem Solutions therefore will not memories than mine will almost certainly find be simple if they exist at alL some inaccuracies in the foregoing The fact is P L 86-36 that at that time things started happening to EO 1 4 c P L 86-36 Oct-Nov 1986 CH YPTOLOG page 27 SIWIU ' ' IIA fDLB VIA COMHfT CHA f fEf S O p ff Y DOCID 4010004 P L EO 1 4 c P L 86-36 86-36 Sii3C IU 'l' FOUQ It woUld be useful to have responses to queries from other collection managers I I IPD41 1 - To a Mixed-Up Fireman on a Diesel Locomotive U Just thought I'd let you know that I spent part of my time here on this hoJ iday 11 November reading your article in CRYPTOLOG U I enjoyed it I hope it brings the results you want P L 86 36 manager is in some way always associated directly or indirectly with many of those functions This was not very well reflected in the article if at all It L -_ _ aD5 To the Editor t8-000 I read with interest the article on collection management Although the historical portion was somewhat informative the remainder was rather insensitive misleading and a bit upsetting at least to me I would 8-000 In fact if it wasn't for the highly dedicated efforts of our relatively small work force of collection managers the current ability of this agency to satisf its SIGIN'l' re uirements would be greatly diminished and or in a messy state of disarray And as NSA's mission continues to grow the functions and roles of our collection managers will most likely continue to expand not diminish ' - -_ _ I GGot I believe that the role of the collection manager and collection management as a profession is absolutely essential if NSA is to continue to satisfy its SIGINT requirements I further believe that the collection manager in the performance of the multiplicity of his or her duties and functions in the process of getting the product to the consumers is absolutely vital and should be so acknowledged Today's collection manager can be and should be very proud of his or her chosen profession to 000 This is not to say that collection managers cannot or need not improve upon their past performance and accomplishments - they can and they are very willing to In fact today's collection manager is in a better position than ever before to do a more effectiv L L I ' 'k-e-w ' j ' s-e- -I ---c-a-n lt-i ' m-a-g-l - n-e--a-g-r-e-a-t-d e-a-l ---of job more efficiently due to the modern tools of support for your recommendations coming from the computer age coupled with recent the Signals Collection Career Panel collection telecommunications upgrades The most professionals and aspirants or the' ColleCtion significant shortcoming of today's collection Association Some criticism perhaps manager is due to the lack of a meaningful training program All of the efforts and all of 8-COO As collection managers we are the tools and communications enhancements continually involved in a multitude of sins to will not optimize the collection manager's skills or maximize his or her productivity unless they include are fully and formally trained to handle the many and various collection-related matters in which they become involved A good complete well-structured formal collection manager's training program is sorely needed to remedy the initial lack of technical subject-matter related and literary expertise of our new collection A good collection Oct-Nov 1986 EO 1 4 c P L 86-36 lIA HJ l l CI YPTOLOG page 28 Sl e U I 'I' VIA OOAH Wf' GIIA H'H3LS OnLY DOCID 4010004 t' L 86-36 manager trainees This approach I believe would be a positive step in the right direction I L - is the capability to feed back decrypts to field sites seven days a week without having an analyst on hand to monitor the process G 8 4 1 r - - - - - - - - - - - - - - - - - - To the Editor FOUO We read with great interest two letters to the editor concerning the ODYSSEY CAMS article in the CRYPfOLOG We in the organization where the process was begun would like to address the various points that were made in these two unsigned letters So to 'A Retiring Cryptanalyst An Old Timer ' and anyone else who may be interested please read on G CCO Both letters showed concern over lost messages and perhaps never seeing key messages These fears are unfounded since all messages that pass the various thresholds and process through the system are accotinted for 1 I G GGO The final point in common i8 a concern over the quality of AG22 data We see 8 CeQ Let us address the points common to no point in debating this issue Given our both letters Each advocates interactive editing successes as demonstrated by the above either very early or before statistics the quality must not be too bad Interactive editing would defeat the purpose of There is an additional capability which was not this process fully eXDlained in the oril inal article I I I I the shortest possible time with an absolute minimum of human intervention In order to understand why for us the issue of 'minimum human intervention' wa$ and is critical let us give a few statistics about the volumes we are faced with I ' L - I b I An add Itiona enefit we derive ---I Oct Nov 1986 iii FOUO After we first started using the CAMS process B53 gave demonstrations and talks about ODYSSEY CAMS to interested people and organizations the most notable being for the 1982 CA-305 course and for the 1982 CISI Spring Conf le ren c e We will be happy to discuss CRYPTOLOG page 29 SISG I t W l ' E O 1 4 c IIAtfDLE VIA OOMfh rf OfIAN'H'ELBft ML'ti6-36 DOCID EO 1 4 c P L 86-36 4 O lOQO 4 t' L 86-36 SEOllB'J' any of these points further or demonstrations to anyone interested to give This organization does not view ODYSSEY CAMS as a universal processing method It is not the be all and end all for Agency cryptanalysts For the original purpose for which it was created ODYSSEY CaMS is a very successful process and we highly recommend it to any organization that is ta sked B53 Management and Cryptanalysts P L The following letlel is els8si fiecJ 0000 lila OONTRAOT ii f i 8 e i i e I To the Edi tor CRYP'l'OLOG is a classified publication It may not be read in the cafeteria or in other insecure areas EO 1 4 c P L 86-36 Oct-Nov 1986 CltYPTOLOG page 30 SI30IlI1T IIAPWLB VIA OOMH' OIIAUUBLS OULY 86-36 DOCID 4010004 'POP BECnElif' Ur tBR O 1 4 L FROM THE PAST U Oct-Nov 1986 q'Op CRYPTa LOG page SHCIU 'f' tTMHftA 31 c 86-36 DOCID 4010004 RESULTS OF THE READERS' SURVEY CU In ' the May-Jul 1986 issue we used a variety of type faces word spacings and leadings Also for some articles we right-justified the text for other articles we didn't An additional variation a more subtle one consisted of using for some articles l2-point type at 80% and for others 10-point type at 100% these two are theoretically equivalent The purpose of this experiment was to determine what combination of variables would provide the most readable text given the limitations and bugs of the current word processing software And so we asked readers for their opinions Meanwhile we were moved to another press and a new page size Of the 38 responses received one-third idiosyncratically as might be expected of the NSA population gave only a few top and bottom choices so the totals are not the same for all entries Overall people were more in agreement about what they didn like than about what they did like The scores show an aversion to fonts designed for optical character readers and to closelyspaced words and lines The few comments indicated preference for serif type faces And some respondents hadn't noticed any difference at all This text is set in 12-pt Classic double space between words with 15-pt leading reproduced at 90% now that we are using the new standard page size of 8ixl The result approximates ll-pt type considered ideal for narrative text We believe that this combination will be very readable -About right justification we try it out for each article When it causes grotesque spacing-usually because of a sequence of long words -- we revert to ragged rights We expect to rightjustify routinely beginning in mid-1987 when upgraded software with automatic hyphenation is to be installed In the summary below the NET column shows TOP 3 minus BOTTOM 3 for a net score ARTICLE Odyssey C3 Linguist AI Documents Team Passwords I Editor TOP 3 BOTTOM 3 NET TYPE FACE SIZE LEADING WORD SPACE JUSTIFlED 14 15 2 12 Classic 12 80 13 pts Double No 2 18 9 -7 Classic 12 80 12 pts Single No 18 12 1 17 Classic 12 80 15 pts Single No 3 16 14 -11 Classic 12 80 12 pts Single Yes 11 17 1 10 Modern 12 80 13 pts Single Yes 8 21 8 Classic 101100 11 pts Single No 12 15 8 Modern 10 100 11 pts Single Yes 5 21 5 Modern 10 100 12 pts Single No 6 14 6 Modftaf 121100 12 pts Single No 6 19 2 4 Modern 12 80 14 pts Single Yes 8 8 8 Classic' - 12 80 14 pts Single Yes 1 15 9 -8 7 21 10 21 3 P L MIDDLE 9 4 Class Ita 12 80 14 pts -21 OcrA 12 80 13 pts Single No 17 -17 OcrB 12 80 13 pts Single No 2 1 -Titan 12 80 14 pts Single No 86-36 Oct-Nov 1986 CRYP'1'OLOG page 32 POU OFl'I6IAL USB ONLY '-Sing Yes DOCID 4010004 P L 86-36 PUZZLE b L -- -----J Wheel of Fortune has become one of the most popular TV game shows in history The object is to solve a word or phrase by guessing the letters that appear in it For a bonus prize each contestant is invited to supply five consonants and one vowel then given 15 seconds to solve a phrase with only those six letters revealed Most contestants choose the letters l N R S T E which are the ones revealed in the phrases below A brief clue is provided for each An additional clue is that all the solutions relate to cryptology Note that' Person' need not indicate a proper name EXAMPLE Place N Thing N T N l SE L R EN T ET S ST T T N SOLUTION NATIONAL SECURITY AGENCY MONOALPHABETIC SUBSTITUTION 00000000 I Person L L 2 Place R L 3 Person 4 Thing - - -- L L 5 R 9 Thing T R E N T R E 12 Place LET L E R E s s 14 Place 15 Thing 18 Thing L N T E R S R T N E R R T - - - - R L T R R E 5 R E 5 T R - - T N - - - E L E R - - - - -- R 5 - R 19 Historic thing 20 Thing N E L TEN N E - 17 Phrase R T S N II Thing 16 Thing N S R E s T 10 Place 13 Thing L L - - - - - 7 Thing N - -- L E R L E R N E N E R - - - - 6 Thing 8 Place T N 5 Organization R - E R T - L S E N S N E Oct-l ov 1986 CH YPTOLOG page N L 5 E T E L E N N N N 33 FOil OFI'IOIAL USFi ONLY R T R This document is from the holdings of The National Security Archive Suite 701 Gelman Library The George Washington University 2130 H Street NW Washington D C 20037 Phone 202 994-7000 Fax 202 994-7005 nsarchiv@gwu edu