DOCID I 4009810 J WUl'iJ00Uj DI1 Ul JWV l5lBl J i O'iJ UllDl5WlB 1Jl JW 1J lillJall 1J f WalW I1 lIll EO 1 4 c P L 86-36 SIGINT EXPLOlTATION 1990 o oo 1 LOOKING AT MR I l 2 A LITTLE PEP TALK ' o ' 4 ACCENTUATE THE NEGATIVE ooooooooooooooo REMEDIAL SOFTWARE ENGINEERING ooooo o ooooooooo 7 TELEPHONE PROBLEM HERE oooo oo oooooo __ NSA-CROSTIC NO 13 ooooooooooooooooooooo David H l CONOCES BIEN LA GEOGRAFIA oooooooooo I 16 C oA oA o NEWS o o o o o o o o o o o o o o o o o o o o o o o o o o o o W E o S oo o o o o o o o o o o o o o o o o o o o o o 1 7 A T A LETTER TO PRESIDENT ooooooooooooooooooooooooooooooooooooooo 18 LETTERS TO EDITOR ooooo -o oooooooo ooooo o o ooooooooooo ooooo oo 21 'fIllS BOSHMEN'f SON'fl tINS SOBEWORB Ml tTERIl J K e floo GQ I KQ illiZ CI'ell q' 2 eclassi fi ed and Approved for Release by NSA on 10-12-2012 pursuant to E O 13526 MDR Case # 4778 DOCID 4009810 Published Monthly by PI Techniques and Standards for the Personnel of Operations VOL V NO 4 APRIL 1978 PUBLISHER WILLIAM LUTWINIAK BOARD OF EDITORS Editor in Chief o Arthur J Salemme 52365 Collection 1 Cryptanalysis 1 Language Machine Support o P L llf89SSs L 1 49025 Irs 2 16 s IS303S Ma thema tics o Reed Dawson 39575 Special Research Vera Filby 71195 Traffic Ana1ysis o 1 Production Manager o Harry Goff 49985 For individual sUbscriptions send name and organizational designator to CRYPTOLOG PI TOP SECRE 447's 86-36 DOCID 4009810 P L e6NFmENTIAL IOINT EXPL OITATION 1990 April 78 CRYPTOLOG PageEll 1 4 c P L CONfIDENTIAL 86-36 86-36 DOOID 864 e09810 TOP SECRET l1MBRA EO 1 4 c P L 86-36 PEANUTS IN MII K CH April 78 CRYPTOLOG Page 2 TOP ECKET lJMHKJ P L 86-36 EO 1 4 c DOCID 4009810 EO 1 4 - -te7P L 86-36 'fOP SI3CRI3'f UMBRA I I I I Apnl 78 CRYPTOLOG Page 3 TOP SECRET UMBRA P L 86-36 EO 1 4 c DOCID A 4009810 UNCLASSIFIED LITTLE PEP TALK '----- 1 over the pa r c ple of yem RSl eADR' R5 'has been experimenting with computer soft'ware for cluster analysis Six programs implementing nine clustering algorithms have been acquired from outside the Agency and modified for use on CDC 6600 7600 computers These clustering methods are data-dependent i e the model most suitable for a specific data set will depend on the form and nature of the data The intent of the analyst must also be a consideration For example does the investigator wish to find natural groupings in the data or to partition the data base into a predetermined number of categories It may be necessary to try several clustering methods on the same data to determine the most appropriate algorithm PEP-I Probability Evaluated Partitions version 1 is one of the six clustering programs that R5l has adapted for Agency use PEP-l converts a set of input vectors into a matrix of pairwise similarities or dissimilarities see Fig 1 The choice of similarity or dissimilarity measure is an option of the program The user may select any of the following measures cross I C values dot products normalized dot products Euclidean distances and city block distances The choice of measure must be guided by the nature of the realworld problem and the intent of the investigator For xample does the analyst wish to a b a d e b e and valleys in the vector profiles or 1nstead match numerical values in corresponding components It should be mentioned that the analyst may construct his own similarity or dissimilarity matrix and then input the result into PEP-I PEP-I orders the matrix entries from highest to lowest and replaces the numerical value in each cell with its rank order Tied entries r ceive the same rank order The PEP-I algor1thm can best be understood by regarding the entities to be clustered as vertices of a complete graph i e an edge joins every pair of vertices see Fig 2 The edge connecting any two vertices is identified wi th the entry in the similarity matrix which measures the strength of the association between those entities b P L a a d Fig 2 A Complete Graph on Five Vertices PEP-l sequentially deletes edges from the complete graph according to the rank order of the entries in the similiarity matrix the edge corresponding to the weakest association is removed at each step of the algorithm multiple edges are ridded in the case of ties a d matc peaks g k The entry in this cell measures the strength of the relationship between objects d and a For similarity data the higher the value the stronger the association For dissimilarity data the lower the value the stronger the association 't---- o--- i Fig 3 Fig I Similarity Matrix on Five Objects -f'---- ---- m Removal of the edge joining hand j produces a disconnected graph consisting of two clusters April 78 CRYPTOLOG Page 4 UNCLASSIFIED - 86-36 DOCID 4009810 UNCLASSIFIED until a disconnected graph is produced At this point PEP-l pUblishes the two partitions the similarity dissimilarity value where the split occurs the rank order of this value and a probability that the disconnection could happen at random The PEP-I probability is merely an estimate the true combinatorial probability is very complicated to compute and is to be used only as an analytic aid to give a general impression of cluster validity PEP-I then reapplies the algorithm to each partition until the number of entities in a partition or subpartition is less than six Real-world problem i-----------------------j The anatyst must know what to measure to make the data refteat the _e Data i -------------------k---i The anatY8t mu8t ma e an appropnate _ Matrix of SImIaho'' ae larities dis di88imi- similari ties of simitaI'ity 1tL' ----------------I o the analyst wishes to cluster a similarity dissimilarity matrix of his own construction or feels that he can select an appropriate measure to derive similarity dissimilarity data from vector input o the analyst seeks to determine natural clusters in his similarity dissimilarity data matrix o the analyst requires some numerical indication of the quality of the clustering produced by an algorithm o the analyst wishes only the rank order information in his similarity dissimilarity matrix and not the numerical values themselves to determine cluster structure o the analyst is aware that PEP clusters are not necessarily compact and may in fact be serpentine or sausageshaped see Fig 5 o the analyst wants an algorithm that will not be harmed by the presence of outliers deviant observations in the data o the analyst wishes to cluster fewer than 100 entries fewer than 128 on the IBM 370 65 --6 ' 'The anatY8t may ohoose another atu8tering at- gorithm at this stage PEP-l ----------------------- Cluster output and splitting probabi li ties 1----------------------- Internat disposition Integration of of the invest' gator 'r--- information land his knowtedge of ' rea t-wortd prob-----L tem the l Cluster solution Fig 4 The Clustering Process Showing the Interaction in Italics of the Analyst and his Algorithm It should be emphasized that the analyst is unable to prescribe in advance the number of clusters to be found by PEP-lor the threshold levels at which the splittings occur PEP-I is a reasonable choice for a clustering algorithm if the following conditions are satisfied Fig 5 PEP-I clusters may assume a variety of shapes If you have any questions about cluster analysis or about the possibility of applying clustering techniques to your operational problem please contact either Chief R51 8518s orl 1 R5I 86265 U April 78 CRYPTOLOG Page 5 UNCLASSIFIED' I J P L 86-36 DOCID 4009810 Wegotta CONFIDBNTIA l ACCENTUATE THE P L IF 86-36 I l --- or some time now the subelement manager ligence is the intelligence gained from knowing that a target unit is performing normally or is has been engaged in a battle to defend his resources against reductions initi- perhaps inactive I ated for a number of reasons Managers at all levels have been barraged by a series 6f statistics designed to aid in making resourcesreduction decisions to show the efficiency of collection resources to determine the cost of product reporting and to allocate the resource expenditures for each function within the subelement of interest The usefulness of statistics as a tool for making these types of decisions has been widely accepted The accuracy of the statistics is rarely questioned and mapy decisions are based solely on this data The current period of dwindling resources and rising inflation with the concurrent review of intelligence expenditures has created an increased dependence on statistical methods as a means of determining efficiencies of operation Aided by a myriad of computers providing a capa- ---------------------------------- -------- bility to count and measure that was never Statistics as most statisticians will before possible statistics have emerged as the admit can be used to show whatever the statisguardian of our expenditures the protector of tician desires Remember the old Figures our resources and the practically omnipotent don't lie but liars figure adage E00 Ig f c champion of efficiency Savings have been made against t is in our very important p' ur 36 and some very necessary trimming of f t has rearea it 1S necessary to temper stat1st1ca1ry sulted These are significant accomplishments suggested resource reallocations with considerand speak well of the systems we have created ations of the negative intelligence not reflecStatistics by definition is a branch of ted in the statisticians' figures or to devise mathematics dealing with the collection analya method of measuring negative intelligence sis interpretation and presentation of masses contributions Until we can do that we must of numerical data The compilation of statisrely on the analysts to provide qualitative tics requires a collection of quantifiable data judgments through the subelement manager The Only those things which can be measured or importance of these qualitative judgments canotherwise quantified can be considered when con- not be overemphasized and they must be constructing statistics In measuring SIGINT prosidered along with the statistical inputs duction the statistician must therefore look Granted this makes assessments subjective to a at such things as the number of COPES objectives point and renders decision-making more diffisatisfied the number of minutes of copy by cult since decisions cannot be made solely on case the number of products produced he a quantitative basis these assessments number of intelligence requirements sat1sf1ed nonetheless can provide the information necesor some other similar quantifiable data sary to protect that important negative intelligence data which may be irretrievably lost This has caused great difficulties for some in a pri arily quantitative evaluation Rememsubelement managers when defending their reber that what we don't get is also important sources and has caused us to overlook one of In many cases we must accentuate the negative our greatest intelligence contributions which I call negative inteZZigence Negative intelApril 78 CRYPTOLOG Page 6 CONFIDENTllal IIMIBt VIA 68111111' 61IMBIEbB ePibY -DOCIO 4009810 L 86-36 UNCLASSIFIED Reprinted ftom Proceedings CISI Spring Confe1'ence 23-26 Hay 1971 Computer and Information Sciences Institute I I 1 i i I i rity Agency May 1977 S-217-66I 217 pp SC paper was awarded First Place among the Confer e n-c-e A-wa r d spresented by Mr Kermith H Speierman Assistant Ueputy Director Telecommunications and Computer Services in the Friedman Auditoriua on 23 June 1977 his paper will describe Remedial Software Engineering a systematic approach to identifying design deficiencies in software systems and implementing i proved systems economically I will combine ideas previously published by Yourdon Metzger and other authors to produce a methodology for dealing effectively wi th problem systems I will give no detailed explanation of software design methodologies but will assume that the reader either is familiar with them or will consult the references for clarification where necessary T Peter Prescription #36 asserts that rational action must be based on answering three questions 1 Where am I 2 Where do I want to be 3 How do I know I am getting there l This paper will explain how to use proven tools and techniques of software design and project management to answer these three questions It is based on my experience with some small- to medium-sized NSA systems where these principles have been applied successfully I leave it to the reader to decide whether or not Remedial Software Engineering might be applied beneficially to very large projects I chose the term Remedial Software Engineering' RSE for three reasons First to imply that its application would be to a system for which some pieces have already been created but which needs further work to be made useful Secondly to distinguish it from any particular software design or project management methodology And thirdly to imply some measure of orderliness in the craft of rescuing systems from disaster Equally sui table titles for the discipline described below might have been Creative Software Salvage or What to Do When Disaster Overtakes Your So tware Devel pment Project How Did I Get Here The premise of this paper is that the answer to Peter's first question Where am I is In charge of a software development project which is in deep trouble A quick review of some basic causes of project failure will lay the foundation for the RSE methodology One major source of trouble is the lack of any system design If nobody understands what is being built then it follows that nobody will be very successful in trying to build it Another source of trouble is a design that will not satisfy the real requirements In this case the implementers know precisely ILau en e J Peter The Peter Prescription New York Bantam Books 1972 p 159 April 78 what they are building What they don't know is that their product will never perform useful functions for the customer without major remedial action A second cause of disaster has to do with the project plan rather than the system design The system design is a detailed blueprint of all the elements of the system being built and their interrelationships The project plan on the other hand is a step-by-step description of how the pieces of the system will be assembled unit-tested integrated and tested as a system Where the system design specifies precisely what will be built the project plan details how it will be built If a project plan is missing or if it omits key considerations serious problems will result and will frequently manifest themselves as apparent hardware or software design problems If a house-building plan calls for installing the roof before laying the foundation it is almost inevitable that the house design itself will be attacked by the carpenters as being unworkable The plan and the design will then be modified on an ad hoc basis to make possible the immediate tasks at hand Usually there will be no distinction made between the plan and the design A third class of problems results from failure to follow the plan For example the system in trouble might have the wrong amount or the wrong kind of staffing even though the plan spells out exactly what staffing is required Assigning 25 troglodytes to build a condominium is unlikely to produce a livable structure even though the cavemen are handed a set of detailed blueprints and an hour-by-hour description of the thousands of sub tasks required Assigning 4000 architects to the project again with perfect design and planning documentation might yield similar results Although Remedial Software Engineering is called for when development efforts run into trouble there are other times when the application of the same principles might be beneficial For example if a functioning system is about to be upgraded to incorporate new hardware or software such as microprocessors data ase management systems etc the methodology detailed below might be an excellent preparation for the conversion Similarly a system may be producing acceptable results but only at an excessive software-maintenance cost RSE could help reduce the costs In short use Remedial Software Engineering whenever the pain associated with your current software efforts seems to be unbearable In these introductory paragraphs I have CRYPTOLOG Page 7 UNCLASSIFIED DOCID 4009810 UNCLASSIFIED defined Remedial Software Engineering stated its purpose and suggested when 1ts use would be appropriate In summary I quote the author of PatteI'nS of PPoblem Solving who said The most common difficulty in problem solving is not lack of information but rather the failure to use information that the problem solverhas 2 The remainder of this paper details three phases of making systematic use of the information on hand to answer the three questions of Peter Prescription #36 and convert incipient software disaster into a productive system Where Am I First gather the current requirements for the system you are seeking to build or modify and design an ideal system against those requirements Ignore for the moment the pieces of software you already have even if they are unit tested and function perfectly Use any structured design methodology of your choice and apply it rigorously Various groups some at NSA have applied HIPO Transform Analysis Transaction Center Analysis and Jackson Design methodol ogy 3 with success It appears as 2Moshe F Rubinstein PatteI'nS of PPoblem Solving Englewood Cliffs Prentice-Hall Inc 1975 p 8 3See Edward Yourdon and Larry L Constantine Structured Design New York Yourdon Inc 1975 for a discussion of the first three methodologies See also The Ja kson Design Methodology Infotech International Handbook Pasadena 1977 I ' I I I I though the particular tool used is not as important as the fact that some technique is applied uniformly and rigorously across the whole system Fig I is an abbreviated structure chart of a segment of a hypothetical system which must gather some data and display it on a terminal among other things Note that all functions are broken down into the smallest possible component parts and that all data flow and control are explicit You can trace the flow of solid arrows to see where the data goes and how it gets changed along the way Each module has a single function The modules may be combined together in-line to form a single program they might be a mix of separate programs and subroutines or they might even be a mix of hardware firmware and software The point is the structure chart ignores all those considerations and concentrates on the elements of the problem solution only Once they are understood then intelligent decisions can be made separately about how to package the modules You may have to resist a great deal of pressure to skip this step There are still some people lurking about NSA who will try to persuade you that time spent drawing pictures is wasted I suspect such characters pervade many institutions in the private sector too although I only deduce this from the work delivered by some contractors to NSA These people advise that we start attacking the problems directly even though some symptoms may be half-recognized Indeed it seems as though each of us tends to have a deep-seated urge to Fig 1 This paper will use several of the Yourdon definitions and conventions Among them are Solid-tailed arrows represent data Hollow-tailed arrows represent flags A box represents a module Arrows into a box interior indicate data flow into the middle of a module as opposed to through the module's entry point A module is any lexically contiguous set of computer instructions Atomic modules are the lowest-level modules A straight line represents one module invoking another module The higher invokes the lower unless otherwise noted Apri 1 78 CRYPTOLOG Page 8 UNCLASSIFIED DOCID 4009810 UNCLASSIFIED start modifying code too soon Just remember that design is cheaper than fabrication It is an abstraction which can be manipulated easily without intervening distractions such as the hardware peculiarities operating system vagaries or programming-language oddities If done properly the structure of the design will very closely depict the structure of the actual problem to be solved You will probably find that someone who has not had a hand in creating the mess you are correcting will be extremely helpful during this design phase Being uncorrupted by the faults of the existing software design he can give a more objective opinion of the fit between your ideal system design and the problem to be solved The trick is to keep yourself and your team from corrupting him You must now document the existing system design using the same notation you used for the design of the ideal system If you cannot document the design of the existing system then you should decide to replace it in its entirety In this case you might be able to salvage some code from the old system modules as described below This activity can be very dangerous however since you might preserve some of the major causes of your troubles By saving only atomic modules and resisting the temptation to add code to them you can protect yourself in most cases Now that you have two system designs the old and ideal lay them side by side and compare them You should look for correspondence between the two designs You might even have some modules from the old design that have already been coded and can be kept as is because they match the ideal design perfectly You are likely to have some modules that should be replaced because they are corrupt in design The lack of correspondence will be seen as confused flow of control and inconsistent interfaces with various superordinate modules The modules will also mix several functions without justifiable purpose sists of a single module with the functions hopelessly entwined This characteristic probably was due to a bad system design before the software implementation began or perhaps the design was not understood or followed When confronted with hopelessly corrupt modules such as the data gathering displaying one in Fig 2 you should salvage whatever atomic modules you can and then throw the remainder of the code away Fig 3 depicts how you might be able to extract the coding to construct your ideal Send to Terminal and Format a Line modules from the existing mess Very likely all the other functions are so dispersed and mixed up that you will be able to salvage nothing else 1ER WHILE FORMAT THE DATA 1 GATHERING IT AND DISCARDING GARBAGE' TH N 1 Fig 3 I DISPLAY IT You might also find some modules in the existing system that are moderately corrupt but have been performing satisfactorily for some time These represent a real problem particularly if you find yourself in the same tight resources situation as the rest of the world You will be tempted to keep them and thus corrupt your ideal system design Indeed you might have no choice but'to keep at least some of them because of the lack of time money or people Where Do I Want to Be You have already documented where you would like to be in a utopian world when you designed an ideal system But now you must reconcile your ambitions with the reality of limited resources You must answer Mr Peter's second question by defining revisions to the software system that make the best possible use of the Fig 2 shows a possible representation of the software already implemented as well as new existing design documented using the same consoftware to produce a reliable and useful ventions as tbe ideal design in Fig 1 Notice system that the data display portion of the system conIf you keep any poorly-designed modules you must recognize that you are doing so and realize MAIN the possible consequences The cost of mainCONTROL taining those modules will be excessive for TilE the life of the system They might contaminate SYSTEM the rest of your system if you do not take positive steps to prevent it They might not even be Dlnctioning as well as you think perhaps they just have not yet had sufficient PROCESS MERGE DATA AFTER opportunity to fail They might come 'back to EVERYTHING FORMATTING IT haunt you at a critical point in the project WHILE GATHERING IT AND DISCARDING A more subtle danger in using admittedly GARBAGE THEN corrupt modules is that members of your team DISPLAY IT might perceive a lower set of standards for the Fig 2 software project than you intend Actions April 78 CRYPTOLOG Page 9 UNCLASSIFIED DOClD 4009810 UNCLASSIFIED speak louder than words If you retain substandard modules from the old system make sure that every member of your team understands that you know they are substandard that you have weighed the risks and that you kept the modules for very specific reasons Emphasize that the modules are substandard and that you will accept no new modules of similar quality You should insulate all substandard modules from the rest of the system with interim interfaces that conform to all the standards of good system design Figs 4 and 5 show how this can be done Module 3 is typical of modules that can be found in many systems It has multiple entry points its logic depends on internal flags set by external modules it gets its data inserted into it by external modules and it returns its results by placing data directly into the middle of the invoking modules Let us suppose that module 3 has been stable for a considerable length of time under extremely severe testing conditions and further suppose that to replace module 3 would take 12 manmonths Let us also suppose that you want to replace module 2 a one man-month job and retain modules I and 3 MODULE 1 ------ MODULE 2 I MODULE MODULE 1 t I I I MODULE 2 MODULE I control three related functions - n I L __ and 3 are unchanged You are now in a position to redesign at your leisure module I and any other modules which 'invoke module 1 You can also deploy new modules which invoke module 3 through the interface without compromising their design because of deficiencies in module 3 After all the invoking modules have been converted then module 3 itself can be reconstructed Fig 6 depicts this last step The interface module has become permanent with the assumption of controlling functions An array of subordinate modules has been created with each submodule performing exactly one of the functions from the original module 3 -r--- J ' FUNCTI N 1 OF OLD MODULE 3 FUNCTI N 2 OF OLD MODULE 3 U 3 OF OLD MODULE 3 Fig 4 Fig 5 depicts the proper way to insulate modules I and 3 from the rest of your system The interim interface module has single entry and exit points and conducts its external communications data and control explicitly Note that the new version of module 2 uses the interface and follows all the rules of proper intermodule communication but that modules 1 MODULE 1 U MODULE 2 I I I I I I Fig 6 Another activity of designing is to incorporate the best commercially available products wherever possible rather than reinventing the wheel If a portion of your system design calls for modules to manage data consider using a commercially available data base management system There are now several on the market which have given many customers satisfactory results 4 By using one of them you can concentrate your own personnel on solving the customer problems rather than the computer problems Consider also some of the newer hardware technology that is available For example some of your modules might be best implemented on microprocessors When you complete step 2 of your remedial software engineering you will have a complete design plan for your system including how each module will be implemented ' Fig 5 4See ferbert L Gepner User Ratings of Software Packages Datamation December 1976 p 108 April 78 CRYPTOLOG Page 10 UNCLASSIFIED DOClD 4009810 UNCLASSIFIED Bow Do I K1'I01J I Am Getting There - 1 You have compared the existing design to an ideal system based on the current requirements and devised a design for the system you will implement Now you must plan the implementation effort and continually compare the actual progress to the plan to answer the last of Peter's questions 'don could have been studying NSA when he said that Some of the tools that some managers give too little emphasis to are training testing mechanisms and standards It is unfair to everyone involved to ask a newly hired college graduate or recently transferred computer operator to design code and test a piece of software for a computer he has never seen in Managing a Programming ProjectS is an excela language he knows nothing about to solve a lent reference for use in the project-planning phase and I will not repeat in this paper every- problem he only vaguely understands Instead he should be trained in the language he is to thing Metzger says in his book Several points use he should be handed a copy of the standeserve special emphasis however and Metzger dards his documentation software and test touches on them only lightly plans are expected to adhere to and he should It is essential to match your resources to understand the system design and project plan the task or vice versa Remember the cave documentation He should be working from a dwellers building the house Assign the right detailed functional description of the module number of people with the appropriate skills to or modules he is to produce The functional the remedial software effort Consider augdescription will include detailed descriptions menting your work force by contracting some of all interfaces between his software and the pieces of the system to the private sector if rest of the system The system should have you have a surplus of money and a shortage of built-in trace and audit trail features to fa people or if some particular skills are cilitate debugging and testing Given these unavailable to you directly I know of one tools you can safely leave the detailed deproject not at NSA where the configuration sign coding debug and unit testing to the management function was contracted successfully individual's programming team If you cannot staff your effort properly One last point I would like to add to Metzthen scale down your ambitions and undertake to ger's discussion of planning the project is deploy a more limited system If it is imposto consider carefully your system implementation sible to undertake less and still produce a strategy Harlan Mills of IBM has made some useful and reliable system then you should can- thought-provoking points about the working recel the project To do otherwise would be a lationship between computer people and their waste of taxpayer dollars One project manager customers 7 He suggests that rather than disat NSA recently mused as he stood in the shamappearing for several months of coding and testbles of an overly ambitious and understaffed ing as soon as you have gotten the customer to project If you are not big enough to take the sign a detailed requirements document you should bull by the horns then the best you can posdemonstrate another real piece of the system at sibly hope for is to stay a few steps ahead of least once a month This would greatly reduce the beast as you run in circles around the ring the understanding gap that frequeritly widens He was right only if you accept his assumpbetween the customer and the system implementtions that his size and boldness were the only ers on many software development projects It will help insure that misunderstandings which variables But the bull could have been either trimmed down in size or eliminated Shrinking occurred during the requirements defini tion phase the bull would increase the project manager's rel- are surfaced early before the whole system is coded and tested Mr Mills' arguments tend ative size and executing the bull would allow the manager to tidy up in a 'leisurely manner to be reinforced by my experience with several before regrouping for the next fray NSA projects and they suggest some guidelines for your implementation efforts Having achieved the right match-up between the project size and staffing make sure you Getting There provide your workers with all the tools they You have done your homework and rolled up need Dick Brandon president of ACT-Brandon your sleeves to begin the real work Since your recently said I am amazed today that we are system design has been thoroughly tested in the working with 1975 hardware using 1971 softabstract and you have a detailed project plan ware and managing as though it is 1960 -- and and you have even scattered all the right tools that we are trying to automate an organization about your workers' domain not much can possibwith a structure designed in 1944 6 Mr Branly go wrong Right SPhilip W Metzger Managing a Programming Wrong Project Englewood Cliffs Prentice-Hall Inc 1973 7Robert L Patrick Software Engineering and Life Cycle Planning iJatamation December 6 The Computer As Villain Datamation 1976 p 79 April 1976 p 14 April 78 CRYPTOLOG Page 11 Ji UNCLASSIFIED UNCLASSIFIED Remember you are beginning your remedial software engineering effort at a low point on the project Everyone has been working for 3 years and seems to be no closer to putting a system into production now than when they started Morale was low before you walked in but now the requests for transfer are outnumbered I only by the volume of help-wanted ads being clipped from the newspaper You will have to manage the remedial project better than any other project you have worked on Metzger covers most aspects of managing the remedial software effort in his chapter The Programming Phase s If you heed his advice on organizing your effort dealing with changes monitoring and controlling the work and carrying the water for your subordinates you will be in pretty good shape Since the premise of this paper is that the project is already in the middle of the programming phase or even the test phase a few additional points should be made First you are likely to encounter great resistance to change You will find that some optimistic members of the project team are caught up in the light at the end of the tunnel syndrome They will insist that the successful completion of the system is in sight just a few more corrections and a little extra effort will save the system Of course they have been saying this for months and have been reporting the system as 95% complete for about 50% of the elapsed project time They are expending all of their energies on trying to make whatever code they have work They never stop to seriously consider that their system design might be unworkable either from basic flaws present from the outset or because the design has been seriously corrupted during the implementation The trick is to convince them that things are really as bad as you know they are Sometimes a suspension of coding debugging testing and the assignment of a design analysis task culminating in a written report will do the trick Some people will even see the light if you brief them on your own analysis of the existing design Sometimes though you might have to simply pull rank on the individuals and direct them to abandon their design in favor of your revisions The result could be subordinates who are convinced that you sabotaged their good efforts just as they were about to bear fruit You will not have their commitment to the revised system and their support will be less than enthusiastic It would probably be better to transfer to another project anyone who has such an enormous emotional investment in the unworkable design that he refuses to listen to reason rou may also find that many of your program- mers have become obsessed with creating sophisticated code to the detriment of the system This is a common problem stemming in part from historical pressures to use every possible trick to cram many functions into the fewest memory locations possible It might also stem from individual recognition that the current system design will never work and the part of a successful team reward will never materialize The individual then internalizes his rewards by challenging himself to be even more clever in his production of code that will run faster in less core and do more work The trouble is that nobody including the original creator will be able to understand th code in 3 months when it has to be modified to allow for a requirements change The solution to this problem is twofold Orient every member of your team to the project objectives and build commitment from all to the development of a system that works Also establish and enforce programming standards for the project By ensuring that each individual understands the long-term benefits of never using part of an executable instruction as a constant never using self-relative arithmetic never hard-coding variables avoiding GOTO's etc it will become easier for you to enforce the standards Be sure to reward those who produce quality code as well as redirecting those who spend forever making a seldom-executed subroutine run faster violating standards in the process Remedial Software Engineering Swnmary Remedial software Engineering is the systematic application of the best available software design and project management techniques to bring about desirable changes in software systems RSE uses readily available information to analyze the current sta e of the system and implement an improved system First the remedial software engineer designs an ideal system to satisfy the requirements of the customers and documents the design of the existing system with the same design notation Next he compares the two designs and evaluates discrepancies between them Then he resolves the discrepancies by designing a revised system which combines elements of the existing system new elements and some temporary elements Lastly he manages the remedial project paying particular attention to the people problems By using a Remedial Software Engineering approach to problem systems the project manager will make rational choices among explicit alternatives lay the groundwork for productive team efforts and bring the expectations of management programmers and customers into harmony with reality He will have the management tools required to produce a quality product in a reasonable time SMetzger pp 69-112 April 78 CRYPTOLOG Page 12 UNCLASSIFIED DOCID P L 4009810 CONFIDENTM l 86-36 April 78 CRYPTOLOG Page 13 EO 1 4 c P I 86 36 EURONFIBHN'fIAI W U I I FA Q81HN'f EURlhI JnEbS aNbY DOClD 4009810 JfSA-erostie UNCLASSIFIED O IJ This month's NSA-crostic was submitted by guest NSA-crostician David H Williams P16 The quotation on the next page JaB taken fpom an NSA publication The fipst Letteps of the WORDS speU out the authop's name and the tit of the lJopk DEFINITIONS A Feels clumsily B Intellectual nourishment 3 wds c Russian man in the street D Inside scoop E Line from film by Word R cited in Bartlett's Quotations 14th ed 5 wds 105 -4- 51 188 F Industrial city in NE Ohio G Emphasizes H Pressure group I To be ------- is some danger HamLet Act III Sc iv 2 wds J of their appointed K Prolific motion picture director 2 wds L Reinforcing 213 '3'f 144 173 230 138 lZ9 98 Z08 171 123 -Z- % M Erases N Another line by Word R cited in Bartlett's Quotations O Gemstone P Eluder Q Word K's fiftieth film 1966 2 wds R Star of eleven films Every Day's a Holiday Night after Night etc author of eight of them 1893 S Vicinity 113 4S 147 92 82 6315 128 April 78 CRYPTOLOG Pll ge 14 UNCLASSIFIED DOCID 4009810 UNCLASSIFIED T Question asked of the cook at the monastery's fish-arid-chip stand 5 wds u Process of producing images by reaction of light V Intermittently 3 wds W Answer to query in Word T 5 wds 3 T 4 E 6 U 7 W 8 D 9 F 13 M 14 P 93 G 94 K 95 D 106 H 107 0 108 T 109 E 119 G 120 M 121 E 132 I 133 8 134 E 135 H 136 T 137 N 138 L 159 E 160 A 161 J 162 R 172 E 173 L 226 Q Solution next month I l April 78 CRYPTOLOG Page 15 UNCLASSIFIED _ DOClO 4009810 SIN CLASIFICACION lCONOCES BIEN L A GEOGRAF A Tony Melzer 05 Esta vez queremos que tu d s no solo el nombre de cada pats sino el de su capital tambien 1 3 Pais Capital 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 Las respuestas se ha11aran en e1 pagina 21 Tanteo 0-8 9-12 13-15 Cartografo y autor del texto en ing1es traducido en espaftol por Esteban Verm4s 16-17 abril de 1978 CRIPTOLOGO pagina 16 SIN CLASIFICACION I'Muy malo Ta1cualillo Bastante bueno IMuy bueno 4009810 L - DOCID 86-36 CONFIDENTIAL C A A NBWS B W E B M g 1978 P e fio PI ed As of this writing February 1978 it looks pretty firm that our featured speaker for May will be our #1 Member see reproduced membership card Watch for announcements about day and time He is certified as a Linguist and Special Research Analyst Our new president-elect is Francis X Parrino Chief of AI Office of Operations and Control He has served in the cryptologic community since 1951 first as an enlisted man in the Air Force Security Service then in several capacities as a civilian since 1955 r ANALYSIS I SSOCIATION THIS IS TO CERTIFY THAT VAVM B R Inman USN is a member in good standing of this association 1 '78 Dec J EUREURS nd out new t id nf-elecfl ' MMUNICATIONS for the year ending 31 C Pretitlent II U I L - - - - - - - - - - - - - - - J Wk 10 do lot He has performed at all levels of management through Office Chief I m mbet' on peS CAA members who transfer to overseas jobs are Jlhas attended the continued on our membership rolls free no National War College has served on the staff o of the NCRDEF and has served as Chief AI EO 1 4 c dues for the duration of their PCS tour overseas One of the things the CAA Board will be Also over the course of his NSA career he P L 86-36 wrestling with this year is how we can extend has completed studies at the University of Marythe aims and goals and benefits of the CAA to land earning a BA degree in Government and people who are off campus We want to talk Po H roo nA h o-ved overseas in the Office with those of you who have ideas about this IHe has also been awarded Have you been overseas and out of touch What the Civilian Meritorious Service Award do you think the CAA can do to help people He is certified in TA and SR and has been a who are now in the situation you were in U member of the SR Career Panel 6 EUR Qa Meet out new p e'idenfl I How fo be ome m mbetiF L Our new president is David Wo Gaddy a CAA dues are $1 00 peryear Youcan geti Cryptologist in D5 He earned his BA in U S membership applicalionsfrom any of the people History at the University of North Carolina and mentionecLinthese CAA news items Send the an MS in International Affairs at George Washing-c ompleted application and $1 00 to Tim Murphy ton University I B09 3W1l4 Then just sit back anclwatch your in-basket fill up with news bulletins and an nouncements about all the many and varied activities and programs of CAA U His cryptologic experience has included assignments as linguist bookbreaker reporter and Communications AnalYsis Association supervisor on a wide variety of problems and President David Gaddy 3247 targets in Southeast Asia He was selected for President-elect F r a n k P 0 r r in o 5B 7 9 1 the Armed Forces Staff College Norfolk VirgiSecretary _JI nia in 1967 and for the National War College in Treasurer _T i m Mu r i ph y ------ 3 7 9 1'_l1 1971 Board members Since his return from the War College he has held various senior planning positions including a tour at the Pentagon NCR Defense He is currently Deputy Chief Intelligence Community I'm running out of space I I wanted to give A_f_f_a_i_r_s _D_5 _ ---------------------------------i a biographical sketch o iour newest CAA Board member J and mention out Program Can't wedummntee chaired bj ge them in I '1 o 1 tr Y'lext mOnth April 78 CRYPTOLOG Page 17 I I LI _ _ r r CONFIBEN'FIAh ItJld4Btf 'IIA CQI4HR' EUR1h LfltIHS E JUt'f 86-36 UNCLASSIFIED american sl s1 Ors lt-t altte't 1 etllf e6 - The folWruing letter was sent on September 14 1977 to the President of the United States by RoyaL L Tinslell Jr President of the Americ Translators Association Dear Mr President Please accept my congratulations and sincere appreciation for your encouraltin attitude toward the study of foreign languages and cultures as expressed in your letter of 29 June 1977 to Congressman Dante Fascell concerning the Helsinki Final Act Let us hope that the foreign-language teaching profession has learned its lesson from the 1960s and will make better use of future support by the Federal Government I am a dedicated teacher of German and sometime Russian teacher but it is not as a teacher of foreign languages that I am writing this letter you will no doubt receive enough letters from my colleajtUes in the profession Rather as President of the American Translators Association I am deeply conc rned about the future of the translating profession in this country Although the study of foreign languages is critically important for the welfare of the U S no one can learn more than a few languages and most will do well to acquire minimal competence in even one foreign language Anything the individual learns about all the other cultures must ultimately be learned through the medium of translation You are more aware than most Americans I am sure that the United States is no longer the world leader in several scientific and technical fields This is no catastrophe in itself for surely every nation should have the right to excel at something I think you will agree however that unless our American scientists and technicians can keep abreast of the latest developments in their respective specialties there is at the very least undesirable duplication of effort and waste of time energy and money Translation provides the only feasible means by which we can keep up with developments in all areas of the world Reprinted by permission of The ATA ChronicLe newsLetter of the Ame1'1 aan Translators Association P O Box 129 Croton-onHudson N Y 10520 1 ra '01' e SOC I8t1 4 1977 september 1 In recent years there has been much duplication and waste even in the vital area of technical translation Cover-to-cover translations of foreign technical journals often provide translation of worthless material along with the crucially important There should be a screening process for each discipline whereby leading experts in the field would select important articles and books for translation by competently trained translators The latest developments in science and technology are now being published in more than 70 languages of some 100 countries Even though English still accounts for about 45% of this material and another 35 is published in Russian German and French some 20 appears in the less-well-known languages The information value of Japanese publications has increased remarkably over the past few years while Hungarian has become very important for electronincs Swedish for metallurgy power engineering and communications systems and Finnish for woodworking pulp and paper technology icebreaker construction etc Polish Czech Bulgarian and Rumanian technical literature is also becoming important as is Spanish and Portuguese When an agreement is finally reached that will permit us access to the technical literature of the People's Republic of China we shall have several decades of development to study and there is less than a handful of people in the United States today who are competent to translate technical Chinese The number of translators in the Federal Government has been steadily reduced over the past several years The shocking state o affairs in government translation was descrIbed by Richard S Relac ina report in The Federal Linguist Vol 6 No 1-2 Summer 1974 In what was presumably intended as an economy measure government translators were progressively phased ou by elimination of jobs and by refusal to replace personnel lost through death or retirement Many simply resigned because they were given GS ratings below those specified for the positions and were deni promotions Government translation needs were supplied to an increasing degree by contractors who recei1ed the work on a low-bid basis Despite claims to the contrary by government April 78 CRYPTOLOG Page 18 UNCLASSIFIED DOCID UNCLASSIFIED I Ij I procurement officers it can be documented that this procedure resulted in an incredible waste of money for a large number of marginally useful or completely worthless translations Most professional translators either refused to work for the ridiculously low rates paid by the contractors or in some few cases of which we are ashamed they simply did not spend the time and effort necessary to do a professional job The contractors were forced to recruit students immigrants war bri es etc in order to meet their commitments Some of these people did indeed become competent translators in time but at a staggering cost to the taxpayers Mr Lester C Benefeld Jr Director of the Foreign Language Research Branch of the U S Army Foreign Science and Technology Center was questioned during a meeting at Georgetown University in 1975 about the quality of this contract translation Mr Benefeld answered that most of these translations would never be used and that it was sufficient in any case if the Army's scientists and technicians could deduce from the equations tables and graphs what the article was about I One wonders why the government should have paid even low rates for translation that apparently was not needed Shooking state of affaiT's in government tI'ClnsZation In July of this year an official of the U S Bureau of Reclamation in Denver stated that there are government scientists and technicians who honestly believe that their Russian and German counterparts are incapable of writing understandable reports of their own experiments because every translation these American technicians have tried to use has been virtually incomprehensible Competent professional translators do not turn Russian and German technical writing into incomprehensible English and the Bureau of Reclamation asked the American Translators Association to hold a workshop for translators so that the Bureau could explain the tyPes of materials it needed translated Throu gh this' workshop the Bureau found several professional translators who could provide competent translation services for reasonable compensation This false economy move of the Federal Government has also had its negative effects on foreign-language education in the colleges and especially in the high schools of this nation Granted that the foreign-language teachers were slow to respond to the desire of students to learn foreign languages for practical use rather than as an entr6e into the literature of these languages government policies must also share a large part of the blame for the decline in foreign-language study in this country during the past few years Back in 1973 President Nixon's budget wiped out support by the National Defense Education Act of some 4500 courses in more than 80 languages and in all foreign areas -- at a savings of less than the cost of one F-lll fighter-bomber Marshall D Schulman The New YOT'k Times April 6 1973 p 39 With the Federal Government eliminating career opportunities and foreign-language programs in the schools indicating by its own example that foreignlanguage study was of little or no importance is it any wonder that young Americans have no desire to study foreign languages The United States is unique among the large industrial nations of the world in its attitude toward translation There is practically no training available for translators in this country Most of our professional translators are either self-trained through years of trial and error or they received training in European schools for translators and interpreters According to the Relac Report there were no more than a half dozen translator trainees in the entire U S Civil Service in 1974 Prior to 1970 the only training programs were at Georgetown University and at the Monterey Istitute of Foreign Studies Since then some 25 or 30 colleges and universities have initiated one or more courses in translation techniques but the only viable comprehensive training programs that have been added for technical translators are those at the University of California at Santa Barbara French German Spanish Carnegie-Mellon University French German Russian Spanish Stanford University German the University of Puerto Rico Spanish St -Mary-of-the-Woods College French Spanish and the Rose Hulman Institute of Technology German Russian No T'eUabZe statistios e r ist fOT' the amount of tI'ClnsZation in the United States Moreover most of the students in these programs all of them at the Rose Hulman Institutute are ot training to become translators -- they are future scientists and engineers who want to learn how to read foreign research i their own fields They will still be able to read only one or two foreign languages however Of a total 162 students enrolled in 5 of these programs as of October 1976 approximately 60 hoped to become translators The same 5 programs graduated 12 students in 1975 at least 10 of whom were employed in non-translation jobs or in positions where their language skills were of only minor importance April 78 CRYPTOLOG Page 19 UNCLASSIFIED DOClD 4009810 UNCLASSIFIED In contrast Canada has 8 training institutions for translators and interpreters and Western Europe has over 40 There is already a critical shortage of competent technical translators in this country and if the Bilingual Courts Act ever becomes law there will be utter chaos with every Spanish-speaking janitor pressed into service as a court interpreter In a courtroom situation where property freedom and even lives may depend upon A nationat disgraae accurate translation of not only words but of culturally significant gestures tones of voice facial expression etc we cannot afford to have untrained people acting as interpreters Translating and interpreting require training and experience NOT just bilinguality No reliable comprehensive statistics exist for the amount of translation performed annually in the United States the amount spent for translation the relative importance of various language combinations and of the different scientific and technical fields that use translations the number of full-time and part-time translators in the country etc etc etc On the other hand periodic surveys of almost every other aspect of business provide current statistics on how many hamburgers are consumed annually in Kalamazoo or how many pickles one can expect to sell in Atlanta Since 1970 I have contacted the Bureau of the Census partment of Commerce Congressmen Senators -anyone and everyone I thought might persuade the Federal Government to collect such data along with the myriad other statistics they are compiling continuously -- all to no avail How can we plan training programs intelligently how can we predict how many translators will be needed in 10 years for example and in which languages and technical fields without reliable data concerning the present state of the profession The American Translators Association simply does not have sufficient funds to gather such statistics The Soviet Union has these vital statistics The All-Union Center for Translation of Scientific and Technical Literature and Documentation in Moscow reports that its translation output doubled between 1970 and 1973 that it increased by 150% between 1973 and 1975 and that the 1976 output was 30% greater than in 1975 totally some 350 million words The All-Union Center is only one of several agencies in the Soviet Union The Chambers of Commerce and Industry translated 230 million words in 1975 Intourist translates about i7 million words De- annually and the Central Research Institute for Patent Information a little over 7 million words This was a total of more than 500 million words in 1975 and other smaller departments agencies and research institutions in the Soviet Union are reported to have translated several times that amount V N Gerasimov Transtation News Vol 7 No 1 April May 1977 pp 1-11 Moreover since 1975 all translations in the Soviet Union have been registered We have a National Translations Center at the John Crerar Library in Chicago that makes a valiant effort to eliminate duplication and reduce costs by making existing translations available for approximately the cost of duplication or by direct ing requestors to the proper source for the desired translation Unfortunately National Science Foundation funds for the NTC were cut off a few years ago and the NTC staff was reduced from 11 to 4 Over the past four-and-a-half years the NTC has' deposited or reported availability of 85 891 technical translations but even more were deposited with or reported to the NTC the staff of four simply could not process any more The NTC estimates that no more than half the technical translations made in the U S and probably considerably less than half are reported' to them due at least in part to lack of funds and staff to make the existence of the NTC known to everyone involved in the production and utilization of translations This Mr President is a national disgrace Although I have concentrated on technical translation because that is the critical field commercial translation banking reports correspondence contracts bills of lading etc is playing an increasingly important role as American business tries to compete in the world marketplace Nor should we forget the importance of translation in making literature and drama from other cultures available to the American public The only training program for literary translators' in the U S at this time is at the State University of New York at Binghamton In your efforts to encourage the study of foreign language and cultures I hope you will als remember the neglected but very important role of the translators of America -- in science and technology in business and commerce and in the humanities In closing I would like to offer my services as an individual and the cooperation of the American Translators Association to the Commission proposed by Congressman Simon Respectfully Royal L Tinsley Jr President of ATA Associate Professor of German University of Arizona Tucson Arizona 85721 Aprii 78 CRYPTOLOG Page 20 UNCLASSIFIED DOCID 4009810 UNCLASSIFIED P L 86-36 To the Editor CRYPTOLOG Letters to the Edito- Yes Art CRYPTOLOG ads do get results In a quarter-page f Her in the January 1978 issue I listed some new and improved aids for bookbreaking Within a week after the issue hit the streets the persons listed as contact points were bombarded with requests for information or copies of the programs It pays to advertise in CRYPTOLOGl r To the Editor CRYPTOLOG The note at the end ofl ------------ An unsoZiaited testimoniaZ fI'am a article A Linguist Looks at the 'Tube' satisfied subsaZ'ibeI' -- V V CRYPTOLOG March 1978 states that the article Bookbreaking and Crypto1inguistics was untouched by human hands Whenever i Coordinator P16 U read or hear such statements I start looking for human intervention I remember for exam- I pIe machine-translation output that used to be I' advertised as completely un-postedited that would prove to have hyphens appearing mysteriously but correctlyl or to have proper names output in capital letters such as EYZENKHAUER CRYFTOLOG March 1977 even though the names had been unrecognized by the computer Sam uel S Snyder Intzuence of u s Cryptologio Organizations on the DigitaZ From reading the article I get the impresComputeI' IndU8t National Security sion that the right-hand justification is done Agency Fort George G Meade Maryland by adjusting the spaces between the words in May 1977 36 pp S-2l7 130 U each line without bothering to use any compli ated hyphenation rules But I see last line One of the earliest cOllDnercial high-speed left column page 11 organi- and line 6 right lIne prInters came about because of NSA supcolumn page 11 justifi- Who put in those hyphens and when port a ed at least partly on experimentation a d InItIatIves by Agency engineers LikeHy Fennwatcher WIse one of the first practical charactersensing machines received early contract The authoI' of the artiaZe I'epZies encouragement from NSA U The word processing system used for my article accomplishes right-margin justification by inserting a maximum of two spaces between words and works best on normal lines of 60 to 80 characters Because of CRYPTOLOG's short lines th word processor occasionally was unable fUlly to perform its right justification on my article p 18 and in fact left a very ragged margin in the paragraph that Hy Fennwatcher mentioned 1 Honduras Tegucigalpa Therefore to improve the appearance of the 2 Nicaragua Managua paragraph my helper interactively i e on 3 Costa Rica San Jose the CRT screen hyphenated organization and 4 Panama Panama Panama City let the computer realign the rest of the para5 Colombia Bogota graph Seeing that the results were acceptable 6 Venezuela Caracas except for the last three lines she then 7 Guayana Guyana Georgetown hyphenated justification in the third-from8 Surinam Paramaribo last line and let the computer redo the last 9 Guayana francesa CaYena Cayenne two lines French GUiana 10 Ecuador This intervention would not have been necesQuito sary for a page of standard width but it is a 11 Peru Lima good example of the ease with which text can be 12 Brasil Brazil Brasilia 13 Bolivia revised on a CRT-based text handling system La Paz Sucre 14 Paraguay The alterations did not require retyping of the Asuncion article or even a page IS Chile Santiago 16 Argentina Buenos Aires IG92 '--17 Uruguay Montevideo SolUfion fo NSA - o'fi No 12 S'lu i 'Cono Bi n I G 08 i U D April 78 CRYPTOLOG Page 21 UNCLASSIFIED P L 86-36 Pl-1Wl 711-83-2629S D c L9Sjjl0 _ - jl fJ'--_---l--_ THIS 66CUMENT CON'fAINS CODeWORD MA'I'8RIt L 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