ursuantto E O 13526 MDR Case # 5477 Ul£DVV0Ul£D11 0 lB JlBVV £DlB $UJ B L DlBV lB15 DlBlBl Ul l 1lJ ffi f OO W l1wWffi 3rd Issue 1990 L l'lllla1 BQCYlMISPliT CONTAI1 fS COBBWORB Mtft'fI3RIAlJ 86-36 CLASSIFIED BY NSA CSSM 123-2 DECLASSIFY ON OFi iAatiA A jeRe '§ D tellilii lation Requited NOT RELEASABLE TO CONTRACTORS DOCID 4012034 Published by Pl Techniques and Standards O L YIr No • 3 • • • • • • • • • • • • 3rd Issue 1990 i UIiI SHER BOARD or EDITORS ' EDITOR • • • • • • •• • •• Computer Ststemli Cryptanalysis • • Cryptol inquis t ic s Information Resoufc S • Information Scienc•• Information Security • Intelligence Reporting Language Linguistics Mathematics Puzzles • Science and Technology Special Research Classification Officer Bardolph Support Macintosh Support Illustrator 1 963-1103 1 - •·• C · · t i963-4382 __J 1 1 963-3258 963-3456 I I 972-2491 ' ' J H06 lfi--------1 1 1 963-5566 ·· I 963 - 64 30 1 1 963-4958 Vera R Filby 968-5043 963-5463 963-3369 • 8 968-1315 963-3360 To submit articles or letters by mail send to Editor CRYPTOLOG P1 NORTH If you used a word processor please include the floppy or cartridge with your hard copy with a notation as to what computer operating system and software you used via PLATFORM mail send to cryptlg@bar1c05 bar-one-c-zero-five note no 'o' via ALLIANCE send to PLBROWN note all caps attn CRYPTOlOG Always include your full name secure phone organization and building For Change of Address mail name and old and new organizations to Editor CRYPTOLOG P1 NORTH Please do not phone Contents of CRYPTOLOG may 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 All opinions expressed in CRYPTOLOG are those of the authors They do not represent the official views of the National Security Agency Central Security Service FOR OFFI6IAL USE OnLY DOCID 4012034 SECRE'FSPOKE P L 86 36 3rd Issue 1990 CRYPTOLOG page 1 Sl3GRE SPOR EO 1 4 c P L 86-36 1 4 c L 86-36 DOCID 4012034 S CK TSPO 3rd Issue 1990 CRYPfOLOG page 2 Sf lCRB'F SPOJBii DOCID 4012034 mm 0-m EEO 1 3rd Issue 1990 page 3 0 86 30 DOCID 4012034 3rd Issue 1990 page 4 DOCID 4012034 3rd Issue 1990 CRYPTOLOG page 5 S CRl l'l'SPOK li3 EO 1 4 c P L 86-36 DOCID 4012034 SECRE'f SP9IHl ••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••• SUBSCRIPTION TO CRYPTOLOG o Please enroll me as a new subscriber P L 86-36 EO 1 4 c o Please check my current mailing address o Please change my mailing address o Please change my name o Please remove my name from the mailing list because I am o retiring or resigning o going to a field site o taking long-term training • o too busy to read CRYPrOLOG though I know I'm missing a lot OLD NAME AND ADDRESS Name Last first miJ Org CURRENT OR NEW NAME OR ADDRESS Name Last first miJ Bldg Bldg Org ••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••••• Mail to Editor CRYPTOLOG PI NORTH 3rd Issue 1990 CRYPrOLOG page 6 SEC TSPOKE aCID 4012034 Asphalt and related bitumens native Bituminous limestone sandstone gilsonite short tons 1 980 562 Carbondio ide natural e I j 1 000 cu fl 1 109 530 Coal BItuminOus and lignite lhous short tons 602 932 Pennsylvania anthracite thous short tons 9 729 Helium Crude lhousand cubic feet 4 030 000 Grade A thousand cubic feet 647 000 Natural Gas million cubic feet 21 920 642 Natural gas liquids Natural gasoline and cycle products thousand 42-gal bbls 206 305 lP gases thousand 42-gal bbls 399 611 Peat short tons 526 000 pe i A S' 15 S'12 DI Abra iU V'IiiIM AsiJe' 'Q§ Barite 6oron minerals Bromine Calcium magnesium chloride Cement Portland Masonry Natural and slag Clays Emery Feldspar _ Fluorspar Garnet abrasive Gem stones estimate Gypsum 'm ' E ' m 34 ll 4 § § i nousand sFiOrf1 s 603 024 672 088 5 986 86 827 60 560 1 $600 10 648 15 753 81 261 87 990 1 020 391 883 4 4 4 4 381 001 21 27 1 268 718 67 53 400 883 23 25 1 284 600 69 106 53 932 258 332 58 691 264 415 • • 673 985 182 567 20 458 Nfl 9 905 20 209 8 869 8 411 1 874 2 390 38 354 280 736 • 269 221 18 837 NA 9 436 19 747 1 603 084 498 927 7 055 10 426 680 17 965 000 1 041 349 748 $8 561 144 2 795 509 100 769 46 843 21 599 3 455 615 201 764 378 457 565 760 sJl l r S TEll'7 §SlI thousand short tons thousand pounds short tons ' thousand 376·lb bbls thousand 2r-lb bbblS thousand 3 6-lb bls thousand short tons short tons long tons short tons ' short tons thousand short tons thousand short tons We make continual efforts to protect our classified facts to the degree that we sometimes forget to acknowledge the unclassified ones that should not be revealed outside of official channels Of course there is no way every unclassified fact with which NSA is associated could actually be compiled Below is a list of unclassified facts that are most frequently asked about in the Operations Organizations and that fall in that category $8 879 1 918 748 191 1 194 836 3 772 662 560 505 105 341 10 473 47 992 3 992 600 17 405 759 800 3 745 680 20 698 240 lp05ISAom • 13 923 1 936 2 396 35 132 286 155 4 4 P L 86-36 o The mere fact of SIGINT liaisonQr collaboration between NSA and GCHQ eSE DSD GCSB UKLO CANSLO AUSLO or NZLO o The terms NSAlCSS Representative Alaska NSAlCSS Representative EuroDe NSAlCSS Representative Pacific 1 L -J and their abbreviations The terms signals intelligence SIGINT communications intelligence COMINT elec- o Statements forwarded to contractor facilities or other government organizations which tronic intelligence ELINT and foreign indisclose the indoctrination status ofindividustrumentation signals intelligence FISINT also This statement is to read The individual when used out of context concerned is certified cleared and indoctrio Elementary principles of traffic analysis nated for access to TOP SECRET Special military cryptanalysis and cryptography Intelligence and any other accesses should be abbreviated such as TK B Me LM 00 etc o Individual job titles and descriptions that do not contain information otherwise listed above o The caveat HANDLE VIA COMINT as requiring classification NSA regulation lO- CHANNELS ONLY and combinations of or II provides unclassified job titles and job deother individual channel caveats including scriptions those identifying the BYEMAN TALENT KEYHOLE or LOMA control systems 0 o The fact of' overhead reconnaissance o 3rd Issue 1990 CRYPrOLOG page 7 FOR OFFIGMJ USB ONb¥ DOCID 4012034 SECRE'f r-------------------- r-------------------- Extract from National Cryptographic Habits by Preston Currier Collected Papers on Crypt4nalystic Diagnosis April 1969 8-194-074 3rd Issue 1990 CRYPTOLOG page 8 SECRE'f IIMIDL13 VIA COMI IT CIIMli'ffiLS may EO 1 4 c DOCID 4012034 P L 86-36 SECRE'f jE333 SOllTARY'IIREO L £01 4 c P L 86-36 PHILADELPHIA VIREO J L -4 YEllOW THROATED VIREO L 5 AURALI AnGUAGE IDEnTIFICATion O-CCO The problem ofidentifYj ng spoken languages at the Agency is an old one asl lvidenced by the existence of the second edition 1958 of A GUIDE TO SPOKEN LANGUAGE IDENTIFICATION Prod-03 Informal No 10 2 More recently it was the subject of an article by lin CRYPTOLOG December 1986 pp 11-17 The Language Identification Problem I C-ceo When I returned to NSA in 1987 after I an absence of 18 Yearsilwlissllrp sed to find it was more of a problem than ever whe 1 lasked me if I had any ideas on how to solve it Mter I had been thinking about it for a couple of months she sent me a copy of a suggesIthat tion byl had been adopted by DDO C-547-85 stating that J U I ad been exploring formal features in the same way that distinctive features are used in professional linguistic descriptions and had even been working by computer to systematize them But that approach would not work in the situation Barbara had described U My initial approa¢h was also based on an analysis of distinctive features but relying on features that are more subjective than scientific There was one overriding reason for this approach I had noticed years before that Polish and Spanish sound the same to me when I am far enough away not to distinguish words this is because both languages are accented predominantly on the penult the second syllable from the end of the word I'm told that English and Norwegian have a similar relationship and sound alike ifnot heard distinctly U Thus I started on the premise that languages might be classified by their aural impression on a listener so I compiled a list of seven 3rd Issue 1990 CRYPTOLOG page 9 EO 1 4 c P L 86-36 SECRET HANBLJil Vill COl UNT CHANNELS ONL¥ P L 86-36 DOCID 4012034 classes oflanguage based on predominant accentuation tone and vowel and consonant characteristics Though some of the principles were useful this scheme eventually proved to be too cumbersome to be practical My original analysis is shown in the chart on the next page Maryland Four have predominantly robin-like phrases comparable to a language family and can be identified by song as follows U At about this point it became obvious that this process might produce a scheme valid for some uses but not for this purpose Additional criteria could have been added to form more groups and more distinctive ones such as the exact type of stable accentuation nasality clicks retroflex consonants the presence of th sounds etc but they would have made the scheme even more complex and unwieldy to be handy Solitary vireo short phrases slower than red-eyed This can be compared to staccato Hungarian phrases U My next approach was based on field bird study I used to be able before losing a lot of my hearing to identify more than 100 Maryland birds by their songs and calls sometimes only by one feature which occurs in a stream of notes At times a family of birds rather than a single species can be identified before it is clear what the species is Vireos are a good example At other times different species have seemingly identical sounds so the observer has to listen until a distinctive sound is uttered In many ways this parallels human language we do not understand any particular words used in bird communication and perhaps we do not need to know any words in any particular language to identify the language perhaps we can identify languages by subjective means U This has a parallel in field ornithology as well American ornithologists insist that diagnostic markings be carefully observed for positive identification this would be equivalent to identifying a language by words which are understood British ornithologists accept the principle of GIS meaning General Impression and Size this corresponds to the subjective impression made by the sounds and mannerisms of the call CD Take the bird family Vireonidae for example Vireos are small birds olivaceous and grayish in color often hard to find and to see clearly and best identified by their songs Six vireos can be found in Red-eyed vireo short frequently repeated phrases This can be compared to Chinese Yellow-throated vireo diagnostic threeeights note This distinctive feature can be likened to clicks in Hottentot Philadelphia vireo very close to red-eyed vireo but more mellifluous and slower Comparable to Estonian overlong vowels D The ability of the listener to identify the language is what is important even if the guidelines are unscholarly Iflinguists find as some have that Japanese sounds like Turkish but is more singsongy the only question is Does it work Subjectively we find that for some people Greek is thin and Armenian is thick Romanian is like Italian with more sh sounds Portuguese is like Spanish spoken with a cold and even that Zulu reminds me of Italian Whatever works is correct FOUO Eventually I reorganized my findings and in the fall of 1987 presented them in a course at NCS listed as LG-220 Aural Identification of Languages to ten students The final course was organized with a combination of three potential approaches distinctive features subjective impressions parallels with bird calls w OOO The results of the session were mixed Everyone seemed able for a time to distinguish some oflanguages we started with Some who had to deal directly with collection in the field during the period spanned by the course found that they could isolate their priority languages and eliminate others quite well Other students could not identify certain languages at all 3rd Issue 1990 CRy ProLOG page 10 SECRET IIhP'l9LE YIA COl UNT CIIAP'iNELS ONLY DOCID 4012034 SECRE'f' Initial Distinctive Feature Analysis Class 1 2 vowels glottal 1 2 3 4 5 6 7 8 9 dominant accent long vowels tone example Tagalog Spanish Hungarian Lingala Ilocano Kurdish French English Serbian These criteria were based on my observation of how I hear the languages 1 2 vowels means broadly that there are many open syllables i e ending in a vowel rather than in a consonant which can make a language sound either very deliberate or fast Glottal refers to any sound deeper in the throat than normal American English sounds and therefore includes glottal stops and pharyngeal sounds Dominant accent means a predominant emphasis on the first last or penultimate syllable Long vowels refer to languages where long and short vowels are phonemically different sometimes difficult to determine without knowing the language because vowels are often lengthened for the sake of emphasis Tones refer to higher and lower pitches used in a consistent manner 5 CCO METHODOLOGY I P L 86-36 U Some members of the class never did accept the fact that the purpose of the course was immediate recognition of the language being heard They wanted to follow their own instincts to identify languages by words which they understood The great drawback here as Ihas already pointed out in the paper mentioned above is that there are recognizable cognates in many languages so language identification based on words alone is flawed from the start U Despite the early mixed results and mixed student evaluations the course has continued and several sections are taught every year mainly to students going on TDY and PCS Happily the two sessions given in FY90 received uniformly favorable student evaluations U As is common in teaching practice we try to go from the familiar to the unknown Everyone seems to recognize French-why or how that is I'm not even sure of the correct question to ask here-and almost everyone is quite sure of German Spanish and Chinese From there on we are on less secure ground U I begin by classifying the main accentuation pattern of non-tonal languages with practice for each type We use four patterns first syllable as in Hungarian Czech Estonian penult Polish Spanish Italian last syllable French Turkish Hebrew and mixed Russian Tagalog and Albanian 3rd Issue 1990 CRYPTOLOG page 11 UAP tQb VIA SECRET COl lIP'tT CIIANN bS ONLY DOCID 4012034 EO 1 4 c P L 86-36 SECRE'f CU From here we move on to other features which are not quite so distinct nasality French Portuguese long vowels Hungarian Finnish palatalization Russian Chinese retroflex consonants Hindi Tamil tone Asiatic Chinese Vietnamese and African Bantu languages Malagasy etc U The major Romance languages are presented as follows French nasal and final accentuation Spanish penult accentuation Portuguese nasal and penult accentuation Italian penult accentuation and long consonants Romanian like Italian but mixed accentuation and many sh and sht sounds _ _ J more often I am amazed at the ability of many students to identify a score or more oflanguages with facility after just a few hours of training U Arabic is identified mainly by the wa sound which precedes most clauses while this is a word meaning and it is not presented as a -EO 1 4 c word and knowing the meaning has 11 orr eall6 - 3 6 effect on the ability to perceive it and il1terpret it as an Arabic feature U Some languages are hard to characterize i e all Slavic languages have many palatal consonants but Russian adds a great deal of vowel palatalization Unfortunately the only way to classify Ukrainian is by the lack of this vowel palatalization and ultimately perhaps it can only be recognized by those who already have a knowledge of Russian CU The major Slavic languages are presented as follows Czech Slovak accented on first syllable Polish accented on penult Serbo-Croatian mixed accent but lilting quality suggesting a tone language Russian a great deal of palatalization ya ye yo and yu syllables Bulgarian t sounds in many final syllables t ta te to Ukrainian like broken Russian less palatalization 8-000 It may seem at first glance that this is not enough information on which to base the identification of a language particularly in view of the occasional problems which arise 1 W GGO Recognizing language family is particularly important with creoles and pidgins EO GGO At present we deal with about 50 languages per class in two different complementary ways language families and geographic areas as a major member of the Romance familY hut is also reviewed in terms of Central Mrica North Mrica and the eastern Mediterranean where Lebanon is located I I 8 GOO The problems involved in the range of languagesalso include several concomitant 3rd Issue 1990 CRYPTOLOG page 12 CR T EO 1 4 c IIANBLE VIA COMlNT CII Jl'l J ES O I V L 8 6 - 3 6 DOCID 4012034 1 4 c 86-36 st Cftt 'f situations One ofthese is the use of the same language in distinctive dialects in neighboring countries i e Farsi in Iran and Dari in Mghanistan at present only a person who knows the vocabulary and or syntactic differences between these two dialects can differentiate them This be sent to collecting stations at home and overseas as requests keep trickling in for more help in language identification D Retention is one problem which we face along with all other areas oflanguage training To enable students to practice and review at home we have found it possible with our more recent equipment to make special unclassified cassette tapes with many language specimens C-CCO I have speculated about other approaches that might be tried for certain languages One would be for students to learn the first one or two lessons of some language courses which are on tape or records in this way they could combine both the word recognition method and the subjective impression method Two obvious drawbacks are an increase in study time several hours per language and the lack of such material in a great many possible target languages FOUm We work principally with language tapes provided by the Voice of America because they are clear enough to teach the principles we operate with We add other languages as required Even so the range oflanguages available does not meet all ofNSA's possible needs and we are constantly looking for new language specimens D Another possible method is the time-based versus syllable-based feature the first of which accounts for the amount of slurring and the second for the degree of clarity and or staccato in a language This characteristic requires further investigation POUO Originally we began by using Jack Gurin's Language Library of recordings found in the NSA library in the past and now under P16 but many of them were poor because they were recordings of classes the teacher often used caretaker language because of the weaknesses of the student and the student often had a poor pronunciation Others were ad hoc conversations between two speakers of a language who had never met before and the sense of contrivance and artificiality comes through the recordings making them seem unreliable as real native specimens of the languages Ihas developed a 8 000 I remarkable computer program for recognizing graphic specimens oflanguages l say remarkable because I tested it very successfully with L -_ _ I He believes it can be applied _ _ to spoken language after some refinements have been made if so and if a small enough device can be developed the identification problem could be almost completely solved and LG-220 would have little reason fot continued existence FOYO Until then however LG-220 is one of the Agency's solutions to the formidable problem of aural language identification 0 3rd Issue I900 CRYPTOLOG page 13 SEeRE HltN9LE lit COMI-NT CIIItNNELS ONLY P L 86-36 aCID 4012034 Sift i7 g Wh a t fro ha ff _ -_ '----- ---- - - - -_ • d _ _ _--- -- P L 86-36 _____IR5 P L 86-36 1st Issue 1991 CRYPTOLOG ij ageh44 c S CRI'JT P L 86-36 IIMIDL-g VY cot' JnfJ' CIlA L8 O lL¥ 4012034 SIi CRE'i' Fig 1 DATA FLOW COLLECTED SIGNALS SIGNALS ENVIRONMENTS INTELLIGENCE PRODUCT EO 1 4 c P h 86-36 1 6R 61 1 ICIAL USE m Vy U Developing a good characterization of filtering tools will lead to identifying holes and gaps in technological capabilities and to identifying unfulfilled requirements that can be satisfied without new technology This should not be construed as an exercise in system engineering system or device design or building and assembling equipment to do SIGINT D The key to characterization is clear descriptions of data structures that are passed in the SIGINT system and how they are operated on to produce an intelligence product Once a function is well understood and described then science and engineering can proceed with specifics as to what can be done now and what needs more R D CD In Figure 1 there is description ofthe SIGrNT system at a top level with a data flow diagram The problem with this diode concept of unidirectional flow is that it displays only the reaction to a signal situation It does not display a system that must produce product first based upon who needs data then the source of the data and finally how the data are to be prepared and delivered U A more accurate snapshot of the SIGINT system is presented by the data flow diagram in Figure 2 The point here is not that Figure 2 is the only way to characterize the SIGINT process but that the SIGINT System should be shown as a process where the activities interact share data and filter make decisions about all kinds of data It is a data flow diagram that stresses the goes intos and goes out ofs without dwelling on the devices that transform the data The symbols as shown on the key are defined as follows An OUTSIDE INTERFACE is a process or organization lying outside the context of a system that is an originator or receiver of system data CRYPrOLOG page 15 SIWRE'f' IIA H LE VYt GOl Hl'ff' GH I tl'H3LS OULY 1st Issue 1991 DOCID 4012034 SECRE'f 'FOIt O'F'FICIAL t1S P L j' qL f 86 36 An ACTIVITY is a transformation of input data flow s into output data flow s STORAGE is a data store and DATA FLOW is a pipeline along which information of known composition is passed D Data flow diagrams can help characterize what filtering must be done because SIGINT product is produced from data that has been transformed We must consider data structures in motion data structures at rest and the necessary ways to act on them The diagram shown is only an approximation The most effective diagram shows how the activity is carried out and has a good data dictionary that correctly describes what the data should be and its form Filtering may have to include other data to be fused to the flow for good product output C CeO As there are standing requirements for documenting NSA-acquired system hl1r e4 chance that current filtering techniq les arij Faijijr 1st Issue 1991 CRYPrOLOG page 16 R T HA HJLE ¥L GOMHfP GIIttJH EiLS OnLY aCID T EO 1 4 c P L 86-36 1·20 4 oc 'JC • LJ UV - JV SECRE'f well documented understood and therefore characterized 1 U It should be no surpnse that a complete solu- P L 86-36 tion to the entire SIGINT problem may be too big to accomplish initially But we can at least try to see how much of the problem we can grasp what problems we can identify and how we can attack these problems CO-OOO Currently we are looking at the following LOOKING AHEAD CO-OOO Is there a stimulus to invent new filtering techniques Do we have to get serious about characterizing the filters we have or are they already characterized The author is hearing all kind of pleas for help 1 -- Yes there is a need but it is hard to believe enough people have a good grasp of what the 1st Issue 1991 CRYPrOWG page 17 SfilORBT EO 1 4 c IIMffiLE VtA eOMHfT' CIIAUSELS E9NIAb 4012034 SFiCR i T whole SIGINT System really does to glean out the needed intelligence product the author knows he doesn't have such a grasp FOUO NSA folks must help each other to better understand the problem of SIGINT filtering from an operational as well as a technological point of view Some of us have problems right here and now with which we are trying to cope sometimes all by ourselves Others of us want to talk about what we can do in the future At times neither group talks to the other because they are too busy and some are blinded by their own view of what time tag is most important Obviously both aspects have to be considered together to foster a better mutual understanding a foundation of experience and some assurance that the current problem of each new day is not a bigger and bigger compounded predicament of the past and present FOUO Is it clear that we'd better get a lot smarter about sifting wheat from chaff if SIGINT is to fulfill its potential That is an easy question to answer The harder one is will the right folks stand up and be counted when the call goes out for input to strategic planning for aspects of the filtering problem or will there continue to be fragmented efforts to solve such problems without the benefit of both an operational and technological point of view To the Editor CRYPTOLOG is still very much a winner providing a very enjoyable break in work when it arrives At this time I'm writing to ask for a copy ofthe do it-yourself packet on viewgraph preparation described in the 1990 2nd Issue With all the hightech stuffin the National Cryptologic School where I have been since December I still find occasions when viewgraphs are the best things to use in my endeavors Please send me a copy of that packet addressed E9 ITB P L 86-36 Thanks and warm regards unclassified CRYPTOLOG is a classified pUblication It may not be read in the cafeteria or in other insecure areas 1st Issue 1991 CRYPTOLOG page 18 S G T HJLE Y A GOMUff' GnA fELS O fLy DOCID 4012034 P L U On 12 May 1975 the SS Mayaguez a US merchant vessel was seized by Cambodian Communist forces in the Gulf of Thailand Within two days US Air Naval and Marine forces launched a rescue operation against Koh Tang Island where the ship and its crew were believed to be held During this operation the Mayaguez was recaptured and its crew returned This article described the actions taken by USM-7 Ramasun Station Thailand and its personnel in support of the operation In support ofthe operation on 13 May 1975 NSA declared SIGINT Alert One Vacater to be used on all SIGINT product relevant to Gulf of Thailand operations USM-7 was tasked with providing Tactical Reports Tacreps and Spot Reports on Cambodian communications relating to the disposition ofthe SS Mayaguez and its crew and to US rescue attempts In addition detachments ofUSM-7 personnel were deployed to several other locations to support these operations 86-36 The same day the alert was declared the station commander summoned the NSA Cambodian Technical Representative to USM-7 whom he requested to identify four Army Cambodian lin sts for immediate de 10 ent It is interesting to note that the summons came via a two-way radio link the same one employed at other critical times such as the evacuation of US personnel from Phnom Penh prior to the fall of Cambodia Their mission was to search for intercept record translate nd report all Cambodian VHF voice communications reflecting reaction to the recovery efforts L - Within one hour of that initial contact the five linguists four Army and the NSAtech rep and their working aids were flown directly from the nearby Thai air force base at Udom L -_ _ I ya US Air Force C-130 Shortly thereafter the linguists began search e fforts and succeeded in isolating several unidentified Cambodian voice links This search and development At USM-7 itself cast iron 24-hour continulnission a effort involved th ous cover was provided on eight Cambodian Strategic Air Command U-2Rreconnaissance Southwest Region and mainline Morse links As a aircraft result it issued a total of five Tacreps and four spot reports by the end of the operations The 8-660 During the three-day operation two reports discussed Cambodian defensive activities intercept positions were manned on a 24-hour as well as the results of US actions Meanwhile basis one for search and development of addiUSM-7 maintained contact with US Army Secutional Cambodian voice links and the other for rity Service providing informal progress reports to monitoring the most active frequencies Immediits commander ate one-liner Tacreps Were issued through a joint Army-Air Force effort 3rd Issue 1990 CRYPrOLOG page 19 St3€Rt3't' SPOlffi EO 1 4 c P L 86-36 DOCID 4012034 J CR T SPOKE were briefed e ill ----1 1 - as interpreters to assist in US Navy and Marine tactical operations at Koh Tang Island Their mission was to use bullhorns to advise the Camb HaJg 4 P L that the only purpose of the operation was to recover the Mayaguez and its crew -ffi7- Of the three linguist sent only one was successfully deployed to the island but he was unable to establish contact with the Cambodians This linguist was later rescued from the island along with the remaining Marine forces approximately eight hours after their deployment The two other linguists were aboard helicopters that had been damaged by small arms fire and were forced to return to the Thai mainland -fffl-On 15 May another team of three Cambodian linguists was also sen AREA OF OPERATIONS DEPICTING RELEVA 'JT 1 After the Mayaguez was rescued when the US prepared to launch a search-andrecovery mission for the US personnelmissing in action the linguistsl Iprepared a Cambodian text for broadcast and a possible leaflet drop to inform the Cambodians L lOCATIO S 800m The text on which each ' acrep was based was recorded by one linguist and immediately transcribed and translated by another linguist to provide the follow-up Spot Reports After the US recovery operation began -_- -_ issued ten Tacreps and nine Spot Reports on the basis of Cambodian voice intercepts The reports included reflections of Cambodian actions intentions and responses to US operation During the entire Mayaguez operation all reflections of Cambodian intent were relayed to USM-7 for possible follow-up traffic in HF Morse of the mission But the JoiIlt f p' concur so it was not dcme ThertinguiStlS--£e6 mained on alertab6ard ship until their return to USM-7 -terAll the linguists deployed returned safely to USM-7 soon after the successful recovery of the SS Mayaguez and its crew on 14 May U Though the recapture of the vessel and its crew was successful we were greatly saddened by the loss of our Army Navy Air Force and Marine fighting comrades who gave their lives Originally published in the Field Information Letter March 1978 ' ' fflt in the early morning hours of 14 May three additional Army Cambodian linguists at USM-7 3rd Issue 1989 CRYPrOLOG page 20 SEOBH'FSPOIGJ c 86-36 4012034 P L 86-36 The Three Faces of Collection The ongoing DDO restructuring is prompting discussion about what collection management is I am offering my view of it in the hope that it can serve others as well as it has served me I believe that Collection management has three components • the definition ofcollection requirements This is the process of deciding what we must have in order to produce the information required to satisfy our users' intelligence demands THIS CAN AND SHOULD BE DONEONLY BY THE TARGET ANALYST • brokering Collection This encompasses all actions between the statement of collection requirement and performing the collection task It includes all organizations that traditionally identify themselves as collection managers e g 04 staffs P5 A1 G8 B642 etc Personnel of these organizations have the know-how to convert the collection requirement stated by the analyst into the arcane lingo of our various specialized collection systems The broker model drawn from the commercial world accurately parallels the functions of our collection brokers rather than buying stock ourselves directly from a company we calion a broker a specialist who understands the mechanics of the stock market and makes our purchase for us As in the commercial world NSA brokers should also be advisors to their clients on the best way to satisfy a requirement The decision on the requirement however always rests with the target analyst For example I may call my stockbroker and say I want a long-term investment with a reasonable return and steady growth Please buy 100 shares of Amalgamated Circuit Inc The broker may respond that this has good short term prospects but for my long term requirements he recommends Integrated Widgets instead But thenthe choice is mine Collection brokers are more than a bureaucratic cog in the cryptologic wheel they are the human link between the target analysts and the collection system that assures we get the collection needed This demands an understanding of how the collection systems function and an insight into the true nature of the needs of the target analysts Regular dialog between analysts and brokers is essential for this process • Collection Evaluation This is an assessment as to whether the collection provided actually satisfied the needs of the analytic organization Perhaps the literal task was completely satisfied in which case the broker would count it as ajob completed but only the target analysts can say if the intercept provided by the task actually produced the results they were seeking In summary the three components of collection management are collection requirements definition performed by the target analyst collection brokering performed by traditional collection management organizations and third collection evaluation perfonned by both collection brokers quantitatively and target analysts qualita0 tively 3rd Issue 1990 CRYPTOLOG page 21 FOR OFFICfAL USE ONiN aCID 4012034 SECRE'f' SfOU - 8 ElcoL Last year marked the fortieth anniversary ofthe problem known for a long time as the VietMinh problem Fortyyears Thatmeans 1949 and the days ofthe Armed Forces Security Agency AFSA forerunner ofNSA US involvement in Viet Nam has left a scar on the public conscience that has not yet healed But it might be time for alittle-known account of the SIGINT side lOom IT TJ I TeD T1L1ILA D The Situation in 1953 All problems need some lore about their founding and the Viet Minh problem as it was called then has an amusing and instruc- French are fighting over in Indochina Oh was the reaction Well since you seem to know so much· about it do you think you could translate these messages His hand called what could the young officer do but agree run get the glossary or whatever it was he'd seen and retreat into the sanctuary ofa private spot He emerged a while later with passable renderings Yes thelanguage was Annamese as some called VietnamD ' Q ese at that early date • D d and the activity was D ' h' ' Viet Minh the altive story TheingrediSowh Chin a S a ents are an inquisitive '0 _ ready outdated name f---c -c ' young naval officer and ofRo Chi Minh's antiBY some puzzling new inFrench movement David Gaddy D9 ---1 And according to the tercept It seems that ' - USM-9 at Clark Air story the naval offiForce Base in the Philippines had intercepted cer became the first American COMINT unit to some strange-looking material The people at work the problem Arlington Hall who puzzled over such things of passing interest perhaps that the concluded that it was plain text but what lanoriginal notation for such intercept was VNG to guage denote guerrillas in opposition to the govern--cse7 Up pops the Lt jg as I recall the story He ment ofViet-Nam Later it was changed to VH for Viet Minh and as such it flourished The was the sort never to let time sli9 through his hands and he often spent his time in the modest war ended in 1954-'55 with the VH withdrawlanguage library thumbing through the texts ing north of the Seventeenth Parallel leaving a few well-chosen remnants behind In the second and becoming familiar with exotic languages Indochina war -the one we got involved in-we Let me take a look at that he evidently said You know combining his awareness of curagain dredged up VNG for the so-called Viet Cong Hanoi's campaign against Saigon then rent events with his recollection ofa text he had scanned I think this must be Annamese That's later rolled them together as VC for Vietnamese Communist 0 the language of the Viet Minh-the people the 0 1 10 LI QRO 3rd Issue 1990 CRYPTOLOG page 22 8EGRE'F 8POIB3 DOCID 4012034 WILLIAM F FRIEDlVlAN lVIEMORIAL AUDITORIUM PL 8636 How the Friedman Auditorium did NOT get its Name I suggest that the NSA Auditorium be named the William F Friedman Auditorium and that there be placed at the entrance of that auditorium a painting of Mr Friedman and a case displaying a sampling of his published works replicas of his awards and photos ofhim with other cryptologic leaders and national figures where each was apparently lost A note from me to Dr Tordella was met with more SILENCE Finally in early June 1972 a memo came from the Incentive Awards Branch saying the suggestion along with several others for honoring deceased NSA personnel would not be adopted It Those words began Suggestion No C-206-70 sent explained This decision apparently sic is based by me via NSA incentive awards channels on on the fact that the number of recent losses are November 3 1969 the day after William F Fried- regrettably high and that these together with man died The idea came so naturally because I those that can be expected within the next few had grown up revering Mr Friedman I teethed years greatly exceed the number of potential on his Military Cryptanalysis works in the late memorials 40's as a GI studying with Lambros Callimahos I thought What an incredible excuse Mr Friedand helped on Callimahos's revision of those man was in effect being viewed as some sort of a works in the early 50's starting as a civilian in AFSA's Technical Division under Mr Friedman willy-nilly choice for memorialization from among an immense set of equally deserving people-not The suggestion was acknowledged promptly by the as the pioneer he had actually been M Awards Committee in mid-December but then Down but not out I immediately sent the essence nearly two years of SILENCE followed to my of the original idea to the Commandant of the great surprise How naive I was then National Cryptologic School Frank Austin sugIn late 1971 to learn the suggestion's status I gesting that the National Cryptologic Course began an exchange of memos and phone calls with Center be named for Mr Friedman I justified it M awards people and eventually the Deputy as recognizing his early development of compreDirector Dr Tordella that continued until hensive cryptanalytic training and his numerous mid-1972 The awards people sought my pacontributions to the literature of cryptology That tience saying the original suggestion and a later was on the fifth of June 1972 and was followed by duplicate had gone to the Deputy Director's office SILENCE 3rd Issue 1990 CRYPrOLOG page 23 FOR OFFI61M USB ONL¥ 4012034 Some guys never learn In June 1974---as the fifth anniversary ofMr Friedman's death approached-I re-sent the original suggestion through Incentive Awards channels asking again· that the NSA Auditorium be made a memorial to Mr Friedman This version was duly assigned Suggestion No C-313-74 and very promptly returned without action This time the suggestion was characterized as failing to meet CSC and DoD criteria for Suggestion System processing Surprisingly those criteria were said to have become effective about the time of the original suggestion in 1969 although that suggestion continued to be processed for another 2 1 2 years I decided on one last shot Since there was a new Deputy Director namely Mr BufTham I decid d to get the 1969 suggestion directly to him I did that in a personal memo sent on August 15 1974 And guess what SILENCE againl I was crushed Five years of trying one method after another following Mr Friedman's death and still no memorial Imagine then my surprise and sense of achievement when one April morning in 1975 I rode the escalator to the third floor in the Operations Building and found WILLIAM F FRIEDMAN MEMORIAL AUDITORIUM over the auditorium doors Though stunned I composed myself enough to send another note to Mr BufTham this time thanking him for giving the project the support it needed and expressing hope I might have the privilege of attending the dedication ceremony I presumed would follow By that time I should have been prepared for what followed but I wasn't Again SILENCE I heard nothing a dedication was held and I was not invited And when I recounted my five years of effort and ensuing disappointment to Mr Callimahos who participated at the dedication my remarks got a blank stare Designation of the NSA Auditorium as a memorial to Mr Friedman had nothing to do with my suggestion BUT WAIT That was 15 years ago There's more As I write this it's late 1990 and I have just finished attending the agency's firstH C p-- 3 6 ogic History Symposium presented by NSA's Center for Cryptologic History CCH Bob Long time no see We were just talking about you a few days ago Fold friend and CCH staffer also at the Symposium said to me during a Symposium break He went on to say something like Tom Burns and I were looking at one of the Center's files containing material about you I What was he possibly talking about A trip to the Center produced the answer There in an aging file was a November 1974 Memorandum signed by the Director General Allen addressed to ADIL CNCS and ADPS reading 1 Attached is a suggestion for designating the NSA Auditorium as The William F Friedman Auditorium 2 For ADIL Please arrange a plaque and painting ofMr Friedman for the entrance ofthe Auditorium 3 For CNCS investigate that aspect of the suggestion concerning photos awards publications and other memorabilia ' 4 For CNCS ADPS Since November is the fifth anniversary ofMr Friedman's death I hope we may be able to accomplish this action sometime this month And what was the Attachment You're so right It was a copy of the August 15 personal memo I sent to Mr BufTham Now you know how the Friedman Auditorium DID get its name--and finally so do I It was then that I reached this shattering conclusion 3rd Issue 1990 CRYPTOLOG page 24 FOR OFFIOw USE ONL¥ D DOCID 4012034 eo FIDB I'fIAL - computer architectures to algorithms and computations Technical U The article serves as a summary of some factoring algorithms and their implementations especially parallel ones The WunderlichWilliams article is a description of an implement1ltion of a parallel version of CFRAC on the MPP together with a case study of the factorization of a 60-digit integer Literature Reports U James Reggia Granger Sutton 1988 Self-Processing Networks and their Biomedical Implications Proceedings of the IEEE 76 6 June 1988 pp 680-692 U This is a survey paper on neural networks in medical applications It may serve as a general introduction for the novice to the topics Reported by David Harris of neural nets perceptrons associative Editor's Note This material was submitted by memories Hebbian learning rules energy the late author to be published in CRYPTOLOG minimizing networks etc It has a useful as space permits See David Harris In bibliography Memorium 1st Issue 1990 U Duncan Buell 1987 Factoring Algorithms Computations and Computers Journal of Supercomputing 1 2 pp 191-216 D Douglas Jones 1988 Application of Splay Trees to Data Compression Communications of the ACM 31 8 August 1988 pp 996-1017 U The purpose of data compression is to U M C Wunderlich H C Williams 1987 A Parallel Version of the Continued Fraction Integer actoring Algorithm Journal of Supercomputing 1 2 pp 217-230 improve the efficiency with which data can be stored or transmitted by reducing the redundancy of its representation A compression algorithm takes the source data and produces a This issue of the Journal of Supercomputing compressed text It is undone by an expansion algorithm Static probability models for strings was distributed free by Kluwer Academic of text have their limitations when applied to Publishers at the AMS meeting in Providence many different sources of text Adaptive Huffman codes using tree balancing schemes is Duncan's paper discusses the computational structure of one way to handle this problem Adaptive the most effective methods for factoring integers arithmetic compression algorithms are another and the computer architectures--existing and U The paper describes the use of generalized used proposed and under construction--which splay trees to construct adaptive algorithms that efficiently perform the computations of these are extremely fast and compact When applied various methods The intent of this article is to Huffman codes splaying or a variant of it to use factoring and computers for factoring to yields a locally adaptive compression algorithm provoke general thought about this matching of I P L 86-36 EO 1 4 c I I CRYPTOLOG page 25 CONFIQKNTIAL H4 Nm E VIA COl I1 IT CgftWWEk Q TL¥ 3rd Issue 1990 P L 86-36 EO 1 4 c 4012034 eON FIBBH'I'IAL that is simple and fast although not optimal in in and modify code to try things out Since the parts of CAYLEY are interdependent the terms of compression When applied to designers of CAYLEY could not let this happen arithmetic codes splaying yields near optimal compression and asymptotically optimal time i€t Groups Graphs is a software package Finally the paper notes that as in other designed to overcome these difficulties It is a adaptive schemes the loss of one bit from the research tool for manipulating graphs on a compressed stream is catastrophic The author computer screen and for computing with them says there is a need for research on 'ways to protect against such problem He also suggests and their automorphism groups It does not have the ienormous scope of CAYLEY but is the possibility of using such compression portable and relatively friendly to the schemes in cryptography researcher This article describes the structure of thei package and some of its algorithms Many computational problems can be converted U G Brassard S Kannan 1988 The int problems about the automorphism group of a igraph and so this package might be helpful Generation of Random Permutations on the Fly Information Processing Letters 28 pp in clobbering them For example the package has an algorithm for often finding Hamiltonian 207-212 paths or cycles Such cycles have importance in generalizations of Gray code problems This is the basic paradigm described by both Aschbacher and Luks at the Providence The paper offers three meeting different solutions to the problem of generating a random permutation when only a small but unpredictable subset of its domain is ever to be queried Building the permutation on the fly allows one not to do the work of deciding on a value for the permutation until that value is requested The choice among the three depends U Richard Thomas 1988 Cayley graphs on the available time and space Two new data and group presentations Math Proceedings of the Cambridge Philosophical Soc 103 May structure techniques are introduced continuous 1988 pp 385-7 rehashing and a balanced tree scheme _----_ I II II U William Kocay 1988 Groups Graphs a MacIntosh Application for Graph Theory Journal Combinatorial Mathematics Computing 3 April 1988 pp 195-206 ------The goal is to come up with improved architectures for parallel processing The Cayley graph of a group G with a given set of generators has as points the elements of G and two such points are adjacent in the graph if one is gotten from the other by multiplication by one the generators The routing problem on the graphc m then be reduced to a type of wdrd problem in the group The possibility of finding efficient means ofrouting between the nodes of the graph is one recommendation for using Cayley graphs There is however a tradeoff between graphs with large numbers of verti¢es of 1One difficulty is that CAYLEY is an all-purpose package that is difficult to modify and is buggy The user just calls CAYLEY and depends on it to do the proper thing This is far from ideal for use as a research tool One would like to be able to go 3rd Issue 1990 Ll'mL CRYPTOLOG page 26 eONFIDRN'fIAL VIA COMHFf OIIAfHHlLS miLY P L 86-36 EO 1 4 c o tb 6 t012034 GONFIBBNtfIAL and the need either to have many interconnections or extremely long communications routes One is interested in finding Cayley graphs of a given number of vertices that are optilIlal for communication between nodes The purpoEje of the Thomas paper is to study groups with a certain kind of presentation and homomorphism anq obtain bounds on the number of generators irithe presentation and the order of the group D Richard P Lippmann 1987 An Introduction to Computing with Neural Nets IEEE ASSP Magazine April 1987 22 pp D At an R51 tea on neural nets technology this paper was recommended as an excellent source along with the book by Rumelhart and McClellan for beginners wishing to learn about neural nets The paper is elementary and expository D Ravinderpal Sandhu 1988 Cryptographic Implementation of a Tree Hierarchy for Access Control Information Processing Letters 27 pp 95-98 D David H Bailey 1988 Extra High Speed Matrix Multiplication on the Cray-2 SIAM J Sci Stat COMPUT May 1988 9 no riven a computer or D We are ' 3 pp 603 -607 --- communications network where users are classified into a rooted tree of security classes Each user is assigned a security class called his clearance Each file or message is assigned a security class called its sensitivity We want users to have access only to that information whose sensitivity is consistent with their clearances Sandhu intends to design a system for achieving this using a conventional cryptosystem such as DES Each security class has a distinct key Items are encrypted according to their security class and stored in encrypted form with the security class name appended Only users with access to the class key will be able to decrypt and read the item D The problem is to assure this desired access to key Sandhu describes a method due to AId and Taylor This method has many advantages but the terrible disadvantage is that when a new security class is added new keys need to be computed for many of the existing security EO 1 4 c P L 86-36 3rd Issue 1990 CRYPTOLOG page 27 C FIBI3 lqJIAL UAl'lDLB Yh OOMHlCF CIIA1iUELS OIiLY aCID 4012034 CONPIDBN'I'IAL classes and existing items must be either reenciphered or all previous keys stored The author presents a new solution to access control at least in the special case of a rooted tree hierarchy of security classes using one-way functions The root is assigned an arbitrary key One assumes that one has a family of independent one-way functions one for each value of a parameter p If b is an immediate child of c then the key of class b is the value of the one-way function with parameter p the name of class b applied to the key of class c Sandhu believes he can construct families of such one-way functions that are sufficiently independent to guard against users pooling their keys to recover keys to compartments they do not have but he cannot prove this U J Skorin-Kapov F Granot 1987 Nonlinear Integer Programming Sensitivity Analysis for Branch and Bound Operations Res Letters 6 no 6 December 1987 pp 269274 U The authors attempt to evaluate the sensitivity of solutions of non-linear integer programs to small changes in the right hand side or objective function coefficients Such sensitivity analysis has been done in the case of linear integer programming The authors assume the non-linear integer program has a relaxation to a continuous problem that is a convex program such that the Kuhn-Tucker constraint qualification applies as for example it will if the constraints are linear This lets them make use of the dual non-linear program Computational results are given for an example with a quadratic objective function linear constraints and 0-1 random variables The point is to get an idea of what additional information to keep along the way in the enumeration tree in order to be able to bound the optimal value of a perturbed problem effective also follows from a 1986 paper by Philippon Further when the ground field is a number field one gets an explicit formula for the constant appearing in the theorem U D G Rogers 1988 An Arithmetic of Complete Permutations with Constraints I An Exposition of the General Theory Discrete Mathematics 70 219-240 U D J Champin D G Rogers 1988 An Arithmetic of Complete Permutations with Constraints II Case Studies Discrete Mathematics 70 241-56 U Let N be an set of m integers ordered by increasing size The set N is symmetric if -n is in it whenever n is in it and -ni nm-i l for all i N is central if it consists of those integers less than some c in absolute value The purpose is to study the group S of all permutations of such sets N The permutation n is said to be complete if the difference I-e where e is the identity If S contains complete permutations then N is called an integral base U The purpose of these papers is to develop a theory of complete permutations especially on central N or symmetric N or similar types of N The arithmetic of complete permutations of symmetric integral bases turns out to be similar to the arithmetic of perfect systems of difference sets Perfect systems of difference sets are in fact an abundant source of complete permutations A constraint on a complete permutation is a specification of its behavior on a subset U Joseph Silverman 1988 Computing Heights on Elliptic Curves Math Computation 51 July 1988 pp 339-358 P L 86-36 I U Patrice Philippon 1988 A propos du text de W D Brownawell Bounds for the degrees in the Nullstellansatz Ann Math 127 pp 367-371 U Philippon shows that the major result of Brownawell's paper making the Nullstellansatz U oldme bout this paper on how to compute the canonical height of a point on an elliptic curve over a number field Tate gave a rapidly converging series for computing Archimedean local heights for real absolute values Silverman generalizes these methods to work for complex absolute values and gives an efficient procedure for calculating local heights at non-Archimedean places Thus we can 3rd Issue 1990 CRYPTOLOG page 28 GOHFIBI3H'I'IAL HAtiDL13 VIA GOMHl'J' OIl AtUlBLS OtlLY 4012034 £0 1 4 c P L 86-36 C6NFIBEN'f'IAL effectively compute heights over number fields having complex embeddings U This is potentially an important step in improving our ability to do computational work on elliptic curves For example heights are used in Zagier's e lgorithm for finding the integral points up to large bounds The canonical height of P can be defined in terms of a limit of heights of 2-power multiples of the point that is roughly the number of digits in the coordinates of the point and its multiples It is not practical to compute it as a limit but one can write the canonical height as a sum of local heights one for each absolute value on the number field These local heights are what Silverman teaches us to compute Computersuitable algorithms for these local heights are included Some examples of computing canonical heights of points on explicit elliptic curves are also included For example on the elliptic curve y2 21xy 494y x3 26x2 the point P 0 0 has canonical height 0 010492061 U William Gardner 1988 Simplification of MUSIC and ESPRIT by Exploitation of Cyclostationarity Proc IEEE 76 7 July 1988 pp 845-7 U This note says that the problem of dealing with direction-of-arrival DOA location with antenna arrays via eigenstructure methods as in the algorithms MUSIC and ESPRIT can be simplified by exploiting a property of modulated signals called cyc1ostationarity U The advantages gained include a reduction in the size of the array and thus a reduction in SVD computation work The disadvantages are that frequency parameters such as carrier frequency or baud rate must be known or measured the integration time for correlation measurement is longer and a different U Robert M Kuhn 1988 Curves of Genus correlation matrix must be estimated for each 2 with Split Jacobian Trans AMS May 1988 signal of interest Eigenstructure methods are 307 1 pp 41-49 intended to allow the individual DOAs of -- interfering signals to be determined provided --the number of sensors in the array exceeds the number of signal sources D A brief description of MUSIC and ESPRIT is included The modified methods should work on such bauded or carrier modulated signals as radar or radio communications The balance of the advantages and disadvantages for a particular problem will dictate whether the proposed new methods should be used in place of MUSIC and ESPRIT U J E Olson 1987 A Problem of Erdos on Abelian Groups Combinatorica 7 3 pp 285·289 EO 1 4 c P L 86-36 3rd Issue 1990 CRYPTOLOG page 29 OO NFI g8 TM IIiHfDLE VIA GOMRf'f k GIh'r HfELS mfLY aCID EO 1 4 c P L 86-36 4012034 eONFIB I3N IAL ' Editorial EDITORIAL POLICY A Letter to the Editor in this issue prompts this brief restatement of editorial policy CRYPTOLOG's criteria for publication are very like those of Cryptologic Quarterly articles may be on any theoretical doctrinal operational managerial or historical aspect of cryptology But CRYPTOLOG is a more informal publication and does not demand as CQ does that the articles make a genuine contribution to cryptologic literature It asks only that they be useful or interesting Nevertheless it does require that facts are accurate and that the writing is clear As for controversy CRYPTOLOG thrives on it But please not the cryptologic equivalent of the moon is made of green cheese factual error is not controversy About classification material up to and including TSC may be published but not if there are additional caveats -------------------- Preferences Classified over unclassified that's the justification for CRYPTOLOG's existence after all technical over non technical shorter over longer Notice to Subscribers Also most welcome are conference and seminar reports book reviews software and hardware reviews technical literature reports B L And letters We dote on Letters to the Editor Readers do too EJo 1 4 Distribution of this Issue reflects changes received by COB 26 October 1990 unclassified 3rd Issue 1990 CRYPTOLOG page • 30 e6Nfi'If N'fIAL IIAtiDLE VIA COMH'f'fl CHAUHELS O lLY 86-36 c 4012034 BOOK REVIEW Doublespeak From Revenue Enhancement to Terminal Living - How Government Business Advertisers and Others Use Language to Deceive You William Lutz Harper Row Publishers New York 1989 P L 86 '36 ReportedbY ---Jr16 As the featured guest speaker at the Cryptolinguistic Association's annual banquet in May William Lutz spoke to the audience about his research and experiences that led to the publication of his book Doublespeak I became intrigued by the examples he cited Immediately that afternoon I purchased a copy of his book to read for myself about the things he was describing to the audience Mr Lutz devotes a considerable portion of his book taking government and politicians at all levels and the world over to task for their abusive use of doublespeak What is most disturbing about Doublespeak is that as Doublespeak is language that pretends to government employees we are a part of the communicate but really doesn't It is language that makes the bad seem good the problem he describes When we or our political and executive leaders use doublespeak in negative appear positive the unpleasant dealings with the public or subordinates or with appear attractive or at least tolerable each other it reflects poorly on everyone in Doublespeak is language that avoids or shifts government service responsibility language that is at variance with its real or purported meaning It is As members of the intelligence community we language that conceals or prevents thought have an obligation even greater than that of the rather than extending thought doublespeak ordinary responsible citizenry to understand limits it correctly what is being communicated to us to Doublespeak is not a matter of subjects and act on the basis of an accurate comprehension of verbs agreeing it is a matter of words and the facts and to communicate to others fact agreeing Basic to doublespeak is accurately and clearly But as the author all incongruity the incongruity between what is too clearly demonstrates we are constantly said or left unsaid and what really is It is bombarded by messages that are sometimes the incongruity between the word and the intentionally designed to mislead distort referent between seem and be between the deceive inflate circumvent or obfuscate And essential function of language-communication--and what doublespeak does-- unfortunately we do not always have the time mislead distort deceive inflate circumvent and resources to strip away the doublespeak and obfuscate receive the true message Thus does the author define the subject of his book which is a bountiful collection of examples Readers should come away from reading Doublespeak with a new appreciation for and drawn from a variety of sources that touch dedication to communicating more clearly every facet of our daily lives media accurately and honestly I urge you all as advertising business professions government professionals in the intelligence business to take military politicians While Mr Lutz' message is this book to heart while you are churning out timely readable entertaining and amusing it is bureaucratic prose of any sort also sobering at times shocking and generally disturbing if not outright disgusting 3rd Issue 1990 CRYPTOLOG page FOR OFFICIAL USE ONLY 31 P L DOCID 4012034 How to Edit a Scientific Journal by Claude T Bishop lSI Press Philadelphia 1984 $14 95 Reviewed by I'-- lp16 Every NSA editor of any type-scientific endproduct newsletter-should have a copy of this excellent book Authors and Publishers should read it too so they might have some notion ofwhat is expected ofthem In just 132 pages Bishop has provided a do-it-yourselfmanual on every aspect of publishing a periodical He spells out the responsibilities and duties of editors and of authors as well He offers sample guidelines for referees for reviewers and authors and even sample letters His purpose in writing this book was to provide guidance to scientists who are not professional editors and who would likely end up learning the business by hard and sometimes bitter experience 86-36 Mter discussions about the referees' reports your Associate Editor agreed to seek the opinion of a third referee as arbitrator This referee found a major error in the theoretical part which escaped 4 authors 2 previous referees andtheAssociateEditor My gratitude is boundless I was saved from a very embarrassing situation But more than embarrassment can result In a seminar on ethics during the International Technical Communications Conference Chicago 1989 we discussed the responsibility of the editor who accepted a paper on a treatment for an eye problem and overlooked the fact that the researcher had not followed protocol for testing on humans Moreover the editor uncharacteristically endorsed the treatment Later an ophthamologist friend filled me in on the rest of the story The researcher went on to another university patented the medication and got A 'SIMPI£ cM of millions from investors But other researchers could not --J duplicate the results Evidently he had cooked the books besides violating protocol So now FDA SEC and IRS are after him Suits and countersuits have been filed The reputations ofall the individuals and institutions involved are affected All this because ofan editor's oversight compounded by an error in judgment A diligent editor would have stopped the researcher in his tracks The austere title gives no clue that this is a fun book to read Bishop illustrates his -r S'Tf I2ll'' Will P£ '2- 2£ v BN points with zany scenarios but he is dead serious about L quality in professional journals He points out that maintaining standards rests squarely in the hands of editors who decide what will be published He is concerned that the pressure to publish is resuItingin the proliferation of trivial periodicals-that serious scientists soon ignore By applying a significance factor he finds that ofthe 90 000 or so journals only 8 000 need be considered the remainder clog retrieval systems with GIGO garbage in garbage out The chapters on the literature of science editors editorial boards the review process referees ethEditors must also exercise great care in selecting ics and record keeping are of particular interest to editorial boards referees and reviewers and in preus at NSA Of lesser practical application are the paring guidelines for them because the review chapters on copy processing and printing and on process plays a key role in assuring that what is post-printing activities for NSA has its own rules published constitutes a reliable record It can save But they do offer considerable insight to the procthe reputation of authors reviewers editors and esses and a wealth ofterminology editors will find the publication itself Bishop cites a letter to the these chapters invaluable in dealing with both NSA editor from an author grateful for the rigors ofthe and commercial printers 0 review process CJI9 l3s idn Harris _ 5t mw M print 3rd Issue 1990 CRYPTOLOG page 32 FOR Oi VU YB ONl N - DOCWf 4 12034 P L 86-36 To the Editor As but one member of what must be a legion of CRYPTOLOG readers I always felt that this publication was intended as a vehicle through which authors could share with others selected technical expositions relating to our mission Expositions which not only explained tried and true methods of cryptology cryptography traffic and signals analysis and other technical fields but also documented and detailed new and innovative approaches to the problems we face in those fields Therefore I was puzzled to see so many articles dealing with managerial problems individual perceptions of Agency programs and explanations of corporate policy being published in the most recent magazine Among the ten major articles contained in the CRYPTOLOG 2nd Issue 1990 I counted only two- Planning for the Millenium Fhich were true sic an4 technical dissertations Six of the remaining eight articles- A Perception of the Tech Track NSA CSS and Counternarcotics Reassimilation Excellence through Evaluation Expert VLSI Designer Hired for Three Days and Rebuttal are not technical treatises and would seem more appropriate as submission for some other publication such as the Agency's Quarterly Management Review QMR Each of these six easily fit into the QMR's Quarterly Management Topic Human Resources and or Telecommunications and Computer Services categories The remaining two articles- The Need for Intelligence and BalloonInt Civil War Style -are perfect candidates for the historical publications of the NSA archives researchers leading edge of technical development CRYPTOLOG deserves better support in terms of input from our technical professionals and should be reporting more of what they are achieving Nontechnical articles of the type noted above should be addressed to and published in other more appropriate Agency vehicles P L 86-36 Don't you agree 1 lchief £1 The Editor replies If there aren't more technical articles in CRYPTOLOG it's not for want of trying Actually we do get a lot of technical articles over the transom and through solicitation that we do not publish Sometimes it's because they are prizewinners in essay contests and automatically go to Cryptologic Quarterly for consideration and increasingly it's precisely because they are about new and innovative approaches -advanced techniques that are closely held Sorry Fred you won't be reading about them in CRYPTOLOG As an internal publication aimed at serving our But you are right We are getting more nonemployees in the technical fields CRYPTOLOG technical than technical articles and especially a provides a valuable service Surely our technical flood of management articles of a general nature experts can provide the technical articles CRYPto the point that I've had to resort to a printed TOLOG should be carrying and in quantities rejection form Why is this I believe it's because sufficient to warrant publication of more than two the fashion in the agency is but a half-beat behind issues per year We are a unique organization that in the outside world Just as in universities replete with talent and full of important achieve- there are fewer PhDs in physics and math and ments Indeed some of what we do is on the 3rd Issue 1990 CRYPTOLOG page 33 FQK QFFIGIhb UBB o ax P L DOCID 4012034 more MBAs so at NSA anyone who can is disdaining the tech track for management It's simply a matter of more reward for less effort So ambitious tech types write on management or non-technical subjects as a way of planting the flag I haven't given up though and I continue to buttonhole everyone who might possibly contribute a technical article I hereby make a plea to whom it may concern Thanks Fred for giving me this opportunity As for publishing the cited management articles in QMR and the historical articles in historical publications the persons in charge of those publications think differently as you will read in their responses In any case the line between managers and technicians is blurred Technicians do not work vacuumpacked in monastic cells solely on tasks handed down from above but participate in decision-making and in the running of the agency They sit on promotion boards reorganization task forces career panels They order dictionaries and software They provide grass-roots input to multi-million-dollar computer systems They hire a specialist for a oneshot job In short they do management things 86-36 cians about technical things for technicians But there is room for much else tUlder that umbrella The technicians' viewiof such management programs as the technical track is very different from that of generalists it needs to be articulated Pate's piece doe s it well There is plenty of room for history in the CRYPTOLOG the archivists will certainly welcome the material for their histories Similarly with other articles mentioned From what I recall of the QMR it is simply not a vehicle for any ofthe articles in that issue William Lutwiniak PI Retired Amicus Curiaereponse tol'-- ---' Editors' lots ate not always happy ones Some readers adore their publications some abhor them otherreaders take what they can get Mr tobservations about the last issue of CRYPTOLOG were certainly written out of concern for the quality of technical publication in the Agency And it is true that the articles he points out are not strictly technical in the sense that we usually define the term i e signals CRYPTOLOG is somewhat more than a medium for analysis cryptanalysis computer science etc technical exchanges among technical types As Gen We can look at this situation from two direcWolff then DDO set forth its charter in A Letter of tions however First and not meaning to be Introduction in the first issue its level of infor- flip an editor doesn't always receive a good supply of the kinds of articles he or she would mal exchange invites short articles and letters on in this case items of a more technical like any subject The editors have interpreted any nature The editor however still wants to subject to mean anything pertaining to NSA And publish matter of value to his or her readership so it provides individuals be they managers or This brings us to the second point Articles technicians a forum for indulging in speculation for which address not-so-technical subjects but advancing futuristic notions for query and chalwhich are perhaps peripherally related and lenge It has become neutral ground where managers and technical types can meet discuss and argue therefore still have value to the work force should not be shunned because they do not meet with one another having first checked their shoulour tight definition of what is a technical subderboards at the transom ject From the Founding Publisher There is some but not much merit tolL- argument I agree that most of what is published in CRYFTOLOG ought to be by techni- _ I Somewhat in this light I couldn't agree more wit tatement that our technical experts can provide the technical artiCles CRYPTOLOGshould be carrying The same bservation could be made about 3rd Issue 1990 CRYPTOLOG page 34 FOR OFFICIAIa 1J8 omx P L 86-36 P L 86-36 DOCID 4012034 Cryptologic Quarterly the journal I am privileged The fact is however because of busy schedules lack of enthusiasm for writing minimal awareness of theeJl istence of publication vehicles or a hundred other reason$ Agency employees don't submit articles in the volumes we editors would like to become accustomed to M my of those who do send us articles though theydon't necessarily contribute technical treatises still write about topics that interest a large segment of the NSA population who don't have access to the Quarterly Management Review or other publications which while more appropriate do not receive the wide distribution that would make the information they contain even more useful toedi A wordaboutl eference to what the NSA archives researchers should be writing First of a11 1 Imay or may not be aware that the archivists and the researchers read historians are now organizationally two different groups The historians my organization while personally interested in topics like The Need for Intelligence and Balloon-Int would not be inclined to write about them The first would need much more elaboration and in-depth research to make it worthy of a cryptologic history monograph or even an article for Cryptologic Quarterly The second while fascinating and even amusing does not carry a cryptologic theme and therefore would not be fuel for any cryptologic historian's production Our criteria for publication of material as a monograph or as a Quarterly article include to quote from the CQ's title page whether or not the article or monograph is of sufficient substance and interest to make a genuine contribution to cryptologic literature By the way Cryptologic Quarterly articles may be written again to quote on any theoretical doctrinal operational managerial or historical aspect of cryptology Although I have not seen an expressed editorial policy for CRYPTOLOG I believe that it is implicitly very similar narrow i e technical aspects of cryptology but we should also have an abiding interest in the broader issues that affect the cryptologic effort To ignore those broader issues is to make us less than we can be Barry Carleen D91 Executive Editor Cryptologic Quarterly FromQMB I am respondin 1 1 uggestion that six articles on management published in 2nd Issue 1990 are more suitable for the Quarterly Management Review QMR than for CRYPTOLOG A few words are in order about QMR It serves as a management tool which tracks performance trends within the Agency Information that is placed in the QMR is factual and presented in graphic and or narrative form limited in both length and amount of detail It is not intended to provide an analysis of the data or to present personal opinions Rather the data illustrate the status trends comparisons and accomplishments of ongoing and new activities in a graphic or statistical manner for individual management analysis Also included are topics that are of general interest to or impact on a wide segment of Agency management but again factual in nature Historically CRYPTOLOG has offered an opportunity for individuals to contribute opnions to relate experiences and to share knowledge on a variety of subjects The articles mentioned fall in this category Therefore the several individuals involved in preparing the QMR read and appreciated the articles cited and were not perturbed that they were published in CRYPTOLOG The article Excellence through Evaluation would be a suitable for QMR if the contents were factually oriented and summarized in one or two pages -- CRYPTOLOG and Cryptologic Quarterly are at their hearts technical journals Both publications recognize however that not only should we as cryptologic professionals be concerned with the ----'1 Chief N313 QMR Coordinator o 3rd Issue 1990 CRYPTOLOG page 35 FOR OPPleIXL tJSE ONti P L 86-36 DOCID 4012034 1 Hiawatha mighty hunter He could shoot ten arrows upwards Shoot them with such strength and swiftness That the last had left the bowstring Ere the first to earth descended This was commonly regarded As a feat of skill and cunning 2 One or two sarcastic spirits Pointed out to him however That it might be much more useful If he sometimes hit the t rget Why not shoot a little st 'aighter And employ a smaller sample 5 Hiawatha in a temper Quoted parts of R A Fisher Quoted Yates and quoted Finney Quoted yards of Oscar Kempthorne Quoted reams of Cox and Cochran Quoted Anderson and Bancroft Practically in extenso Trying to impress upon them That what actually mattered Was to estimate the error 6 One or two of them admitted Such a thing might have its uses Still they said he might do better If he shot a little straighter 3 Hiawatha who at college Majored in applied statistics Consequently felt entitled To instruct his fellow men on Any subject whatsoever Waxed exceedingly indignant Talked about the law of error Talked about truncated normals Talked of loss of information Talked about his lack of bias Pointed out that in the long run Independent observations Even though they missed the target Had an average point of impact Very near the spot he aimed at With the possible exception Of a set of measure zero 4 This they said was rather doubtful Anyway it didn't matter What resulted in the long run Either he must hit the target Much more often than at present Or himself would have to pay for All the arrows that he wasted 7 Hiawatha to convince them Organized a shooting contest Laid out in the proper manner Of designs experimental Mainly used for tasting tea but Sometimes used in other cases Randomized his shooting order In factorial arrangement Used in the theory of Galois Fields of ideal polynomials Got a nicely balanced layout And successfully confounded Second-order interactions 8 All the other tribal marksmen Ignorant benighted creatures Of experimental set-ups Spent their time of preparation Merely shooting at a target 13 July 1967 P L 3rd Issue 1990 CRYPTOLOO page 36 FOR OFFICtAL USE m LY 86-36 DOCIO 4012034 9 Thus it happened in the contest That their scores were most impressive With one solitary exception This I hate to have to say it Was the score of Hiawatha Who as usual shot his arrows Shot them with great strength and swiftness Managing to be unbiased Not however with his salvo Managing to hit the target 10 There they said to Hiawatha That is what we all expected 11 Hiawatha nothing daunted Called for pen and called for paper Did analyses of variance Finally produced the figures Showing beyond peradventure Everybody else was biased And the variance components Did not differ from each other Or from Hiawatha's This last point one should acknowledge Might have been much more convincing If he hadn't been compelled to Estimate his own component From experimental plots in Which the values all were missing Still they didn't understand it So they couldn't raise objections This is what so often happens With analyses of variance 12 All the same his fellow tribesmen Ignorant benighted heathens Took away his bow and arrows Said that though my Hiawatha Was a brilliant statistician He was useless as a bowman As for variance components Several of the more outspoken Made primeval observations Hurtful to the finer feelings Even of a statistician 13 In a corner of the forest Dwells alone my Hiawatha Permanently cogitating On the normal law of error Wondering in idle moments Whether an increased precision Might at the risk of bias If thereby one now and then could Register upon the target - __ ' ' - · iI-4 ·· 'i· ·o i 3rd Issue 1990 CRYPTOLOG page 37 FOR OFFICMI J USE O L¥ _ w DOCID 4012034 TillS QOCYk U NT CONTAIN COQI W JIU iI TIiiIU L NOT R b ASABLE TO CONTRACTORS