Applied Understanding Technology (AUT)
A New Capability for Enabling Concept
Engineering
Technology: n [Gk technologia systematic treatment]
1: the terminology of a particular subject technical language;
2a: the science of the application of knowledge to practical purposes;
3: the totality of the means employed by a people to provide itself with the objects of material culture.
Merriam Webster’s International Unabridged, 1961
The Need
Logic, mathematics, and science are essential to engineering, but what distinguishes engineering from them is that, by the application of handbook knowledge, safety factors, etc., the practicing engineer reliably goes beyond the ever-present predictive limitations and solves problems to the good of society. Modem technology extends engineering practice to the systematic, controlled replication from which whole industries grow. Industry, in turn, is the generator of value — which powers the economic world. It now is recognized that the actual reality addressed by technology is that of a single, global system of systems of intricately interconnected complexity. The subsystems that comprise this unity are of all varieties — not just "technical", but social, political, and ecological, as well — both man-made and natural. Included even are the complexities of governance, for at the same time that we strive to cope with these problems through policy, programs, law, and regulation, we cannot abide erosion of the rights or aspirations of our citizens. Actions taken to address the new reality, whether by government, business, interest groups, or individuals, inherently have a profound effect on that reality, thereby affecting us all. For most of them, we have no science. For all of them, the many constituencies need clear, reliable explanations that they can understand, for the modern world marches on ever faster and with greater consequences.
The world has passed by the stage where togetherness, talent, good intentions, and generous sponsorship of money, resources, and time can be counted on to get the job done properly (World War II style). Too much is new; there’s too much to do. There is no assurance that we can, once again, engineer our way through. There is very high risk that repeatedly, and in many contexts, there simply will not be enough time to consider all the factors. We simply can’t tell without those science-based predictions what is the best way to proceed — unless there is a way to by-pass that necessity for directly applying science and reliably obtain the benefits of a systematic technological approach without it.
There is such a way — a technology for understanding — the focus of this proposal for action.
It meets, head on, the urgent need for a new extension of "technology" so that
that term can apply to any subject at all, whether or not there is an
undergirding scientific or engineering base.
The nature of the new approach is easily appreciated: Most creative, preparatory steps in performing any design do not make use of detailed calculations, but come from the designer’s experience-based understanding of what those calculations will show. Understanding is exercised time after time, much more often than detailed calculation, as the "design envelope" is explored. For each actual calculation, program, or implemented breadboard design there are many cycles of predesign sketchings, schemings, and imagined possibilities that come before it. That is where the creative problem-solving takes place. All that activity is done merely on the basis of an understanding of the calculation that might be performed were this trial sketch to be a final candidate for detailed evaluation. With the AUT technology, actual concept engineering — engineering the concepts, themselves — becomes possible (not conceptual engineering, which happens all the time, and is quite a different matter).
That is the opening. Understanding can rigorously and concisely be expressed directly in a kind of graphical algebra — opening the field of "technology" to any subject whatsoever — even nontechnical areas thought to be beyond the reach of true engineering.
Given the proper medium to work with, the crucial creative steps of engineering and design that go beyond rote science and are what distinguish the practice of a technology (even of science) from science, itself, can be carried out with full rigor, completely in the realm of pure understanding. When there is a proper science, it is used, of course. But when it is lacking, the direct knowledge and experience of the participants in the project takes its place. It is the steps of the practice that are of equal rigor. The engineering still is sound, even if the scientific underpinnings are weak or missing.
This is made possible by a new discipline of Applied Understanding Technology (AUT) which may be interpreted to be either
<<applied understanding>technology> or
<applied<understanding technology>>.
The distinction is akin to applied mathematics versus pure mathematics. The pure side is a technical language (the fundamental meaning of "technology") that may be thought of as a graphical algebra for expressing and manipulating generic thoughts. The mathematical strength of this algebra is what makes AUT work. Its graphical simplicity is what makes it practical. Like any algebra, it provides no solution to any problem (the solution comes from the user), but it provides free and reliable expression for any subject to which it is applied.
Not only does AUT open new domains to rigorous treatment, but also it enhances every old area as well, because to each it brings a new concept — digital understanding — the prospect of perfect agreement, with no room for misunderstanding. The principle is that of digital information or FM transmission. In each case, it is the discrete possible states of a complex circuit that form the information medium, immune to drifting analog component values. In AUT, the more graphical structuring (in the linking of many small expressions into ever larger expressions, like multidimensional chain-mail body armor) the more the possible realm of meaning is pruned and squeezed — until only the single intended meaning remains — perfectly understood because it is unique. If that is not the desired meaning, the only recourse is to make changes that make it right. (Also, there still may be unresolved differences, if the prospect is not yet achieved. But each component can be understood perfectly by all so the problem is understood — allowing the engineering of a solution to proceed.)
The Product
— that is the ubiquitous deliverable product of the Applied Understanding Technology Project. To establish this new mindset, the deliverable items at the end of a two-year project launch will be:
· Video Master Lectures for Exposition
· Teacher Workbooks for Education
· Workshop Exercises for Training
· Anthology of Best Practices
and already a growing body of successful results from early pioneers.
The adaptable and reusable templates of the Anthology (a repository), being the distilled and annealed essence of actual application experiences of AUT Project participant organizations, drawn from the broadest range of subjects, from interoperability standards for the shop floor to considerations of international trade negotiations — with every variety of ecological and societal enterprise (the activity, not the organization-implying noun), isolated, integrated, or otherwise in between.
Significantly, both technical and nontechnical participants (management, end-customers, etc.) can both use and collaborate in the practice of AUT. In fact, in its fullest use for important work (AUT is simple enough to be used casually, as well), their involvement in the process is required, in order to ensure that all pertinent realms of expertise and the needs of all who may be affected by the project outcome are acceptably accounted for throughout the project.
The Rationale
For the ordinary algebra of arithmetic, there are rules of formation, transformation, and interpretation for the variables and operators of the algebra that result in a unique perfect understanding of the meaning (value) of a formula, and that value that can be confirmed by calculation. Similarly, the rules of AUT’ s graphic algebra do the same — where the "calculation" step corresponds to some form of experimentation or debugging of phenomena by any combination of reality, computer simulation, or readers’ imaginations, well-discussed and well-documented.
As to the form of the graphic algebra, it consists merely of boxes and directed arrows that can be named and labeled, respectively — along with a few other notations. As its variables, the graphic-algebra language of AUT directly absorbs (as names and labels) the nouns, verbs, adjectives, and adverbs of whatever natural language introduces a topic, unchanged — freely accepting all their imprecision and ambiguity). As its operators, in conjunction with its own graphic notations (through which all terms are absorbed), it treats the generic natural-language prepositions, conjunctions, pronouns, definite articles, etc. that enrich the descriptions of the topic (along with limited punctuation for naming, etc.) all as elaborate forms of punctuation — having meaning only in the context of the other terms.
The net effect is that the combined word-and-graphic language has only two parts:
and as with ordinary punctuation, this word-and-graphic punctuation completely dominates the meaning of the technical terms, even though it has no meaning on its own. The resulting language is called an "algebra" because, as its calculation, that "punctuation" (like a virus infecting a living cell) dominates the interpretation of all the words (whatever their source), and squeezes out ambiguities and restricts generalities in a completely unobtrusive manner — purely by structure.
Unlike a virus infection, however, the control over meaning exerted by the graphics is entirely beneficial — for only that fact makes the rigorous use of natural-language terminology (essential to the addressing of any subject) possible at all!
The idea is easily grasped: All the separate formulas written in the graphic algebra are about the same subject, so their meanings interlock as one big structured formula. The more graphical structuring there is, the more strict is the interpretation allowed each word individually, as their meanings all link together into one collective meaning. Although initially there may be many possible ambiguous meanings and much lack of precision even for single word meanings — as structure is added, more and more is squeezed out until (as in arithmetic) only one meaning remains — unique, and perfectly understood by all. The sentences, paragraphs, and chapters of a novel are a good analogy — better than a technical report (which follows the same structure) because the individual appreciation of the plot and characters rightly is left up to the individual reader. Generic designs all are of that sort — as are good laws, regulations, and plans.
The Pragmatics
So unique meaning perfectly expressed and understood is possible, in theory. But what about the practice? Is it practical? Actually the language of AUT Applied Understanding Technology is twenty years old, so it is accepted and well tested already — it’s only the understanding of its better application that is new here. The new twist on an old language involves unlearning some of the old ways the language and methodology of its use have been taught. That is why for the AUT Project, it is referred to as an algebra, rather than just a language — to emphasize and make possible the change of mindset (which seems quite radical, at first, to current practitioners).
The language upon which AUT is based is the original "SA" Structured Analysis language of SADT (Structured Analysis and Design Technique) — best known in its decade-old IDEF0 (Integrated Definition) functional-modeling form. It requires courage and a leap of faith in going from the familiar comfort of a top down 2Dflat modeling world to AUT’s 3D outside-in treatment — and to do so well enough to save time and money while doing a better job! With a single view, Everything is decomposed from a single top in terms of non-overlapping parts. But with many separate partial views, each with its own "top", and all collectively covering the subject (like a globe), each view must be detailed just enough to properly link by overlapping! Not all practitioners will make the change, and IDEF0 will continue successfully. But for new people who have nothing to unlearn, or for IDEF users who already feel the need to link separate viewpoints, or who recognize that they never have actually used major portions of their models (even though those parts seemed essential for completeness, at the time, and elicited much heated debate to get agreement) — AUT is the way they ought to go! Those users can do more with less — in the new way.
For the full twenty years, the old modeling language has had the goal of treating any interesting subject in "a language for communicating ideas." To corral such generality and ensure the quality of both analysis and presentation, there always has been a Reader/Author Cycle, in which there is shared team responsibility for both the substance and the modeling of each subject. Multiple "readers" with technical, management, or customer expertise learn the language and make terse written comments on every aspect of draft models. There is only one requirement for writing the language: The message must be acceptable to the Readership! Meeting face-to-face only when essential, readers red-mark their copies of the author’s drafts, which cycle back to each reader with the author’s blue-mark response and a revised draft, until no more red is generated. There is only one requirement for "good-citizen" Readership: Accept only the acceptable (being ready to "live with" others’ views).
The Result
In the field of functional programming (an offshoot of LISP-based Al) there is the recent concept of "lazy evaluation" of functions, in which the full context is remembered, so that the individual function is evaluated only on demand for a new value required by another function. Since functions may produce other functions (as values), the power of this concept can be appreciated. Such programs literally "are ready for anything" they know how to handle, but continually have expended only the resources necessary to achieve that state. The "new twist" of AUT provides this potential for its ubiquitous subject-coverage. It is up to the team to control how much to do, when, to ensure the best outcome.
When the method is maximally applied, the definition of the graphic algebra allows arbitrary precision of expression and the reading rules ensure unique meaning. But the language and methodology rules are so simple that they can be used very loosely, as well. Because of the shared responsibility that follows from the Reader/Author Cycle in use, a skilled team of experts can sketch and proceed rapidly to consensus — with no loss of assurance in the quality of the result, as long as each member fully exercises professional responsibility of "good citizenship" and doesn’t shirk the hard questions that deserve solid answers before signoff. Fluff shows and doesn’t survive. The visible structure stimulates deep insight. Hard knots are worked through to a conclusion (if not a solution). Productivity is in the hands of the practitioners, throughout.
The format of presentation also is familiar. When only box names are listed, an ordinary, indented, outline-form Table of Contents of a model results. But those box names are only descriptive — organizing, summarizing, orienting, and guiding the interpretation of the definitive terms that appear in the arrow labels that meaningfully link the boxes together. Being inherently attached to the structure-imposing, graphical "punctuation", only the labels truly are definitive — with prescriptive power. Those connecting arrows are constraints, not flows, in general. They express the potential for happenings, like roles for actors in a well-drawn plot. The boxes supply the actions needed to suit those definitions, making real various concurrently-happening parallel combinations of the allowed possibilities. The modeling presents graphically what hypertext browsing relations provide.
Viewing the AUT graphic-algebra modeling technique as a natural extension of the standard outline form shows why the information content and information density of an AUT model goes far beyond that of ordinary outlining, document, and thought-organization presentation methods. Every reader senses it; every reader is drawn in to deeper, more effective involvement and focus by a well-expressed model. The new "chain-mail", linked, MultiView models also provide the benefits of object-orientation — but extended to the much broader domain of arbitrary engineered subjects, not just programming system design and structuring.
Complete documentary modeling calls for model diagrams to be accompanied by well-structured text that "tells the story" of the diagram, by concisely weaving together the patterns that show in the structure. Glossary terms introduced in the modeling also are summarized across diagrams, but the entries (like all dictionary definitions) really only concisely summarize the meanings that actually become definitive through proper use. The deep meanings of the terms come from reading the model, where all the nuances are brought to the reader’s attention. "Terminology models" have only key words or phrases throughout, and are the ultimate in subject-dependent dictionaries.
Where appropriate, modeling is directed into actual system and program design (the final stages of the "DT" Design Technique) with the benefit that all inherent parallelism is naturally discovered for possible exploitation. Many forms of media presentations can derive their factual base from (and can be rigorously "tied" back to) the modeling. "Animation" of model meaning (through to actual simulation at appropriately detailed levels) can incorporate cost, quantity, and timing data appended to the analytic structure of the models — like trimming (pruning) a tree, and then trimming it with baubles.
In general, the full strength of AUT understanding technology application should be kept behind the scenes — as is usual for most professional disciplines. It is the completeness, integrity, robustness, and assurance of reliability of the delivered product that is important. AUT is product and production oriented. Its direct users are analysts, designers, engineers, and business and marketing professionals who desire the same technological qualities brought to bear in their own related disciplines. It is up to those, its users, to package its results and put them at the service of the ultimate end users in their terms — the top-level decision-makers and executives at one pole, and the customers at the other.
IDEF — A Significant Start
In its best-known form, called "IDEF0", the SA language is well-established as a de facto standard in both manufacturing technology and business reengineering areas. Currently it is a Federal Information Processing Standard (FIPS) through the National Institute of Standards and Technology (NIST), with work under way toward a more complete treatment as a prospective IEEE/ANSI/ISO standard, which will replace the initial FIPS, once adopted.
A nonprofit independent IDEF Users Group with individual, corporate, and government members exchanges experiences, considers extensions, and supports the development of standards for Enterprise Systems Integration Methods. Both IDEF0 and its companion IDEF1X information modeling, database design technique enjoy capable tool support supplied by vendor members of the Users Group. Both standards use the "DT" of the SADT Reader/Author Cycle (for ongoing peer review of work in process) to ensure a level of quality and project productivity suitable to the stated purpose of the model-development team.
"SA" names the box-and-arrow analysis language portion of the method. "SADT" restores to IDEF0 the full set of SA features omitted when it was simplified out to function-only modeling. "RSA" (Rigorous Structured Analysis) extends the SA graphic language to document design itself, through triangle-and-circle objects and operators expressing multi-model interrelationships. The spectrum of RSA/SADT/IDEF0, from most general to most specific, forms the base for the AUT approach, providing a seamless foundation for a technology for understanding that covers the complete sweep of technological needs, whatever the enterprise.
APPENDIX:
For those already familiar with IDEF0, the following technical summary is offered for comparison. At the time when IDEF0 was simplified out from SADT (in 1981) no use was made of the Structured Analysis Maxim and the metaphors that are definitive of the design of the SA language. Also, SA Data modeling was omitted entirely from IDEF0, and call/support were included without full description, among other differences. (RSA came even later.) Therefore much of what appears here will be new to most readers.
Basics
The SA language of SADT allows the "things" and "happenings" of subjects to be modeled in terms of "data" and "activities", respectively. The actual discriminator between these terms is whether or not they are described by noun phrases or verb phrases, respectively. Thus physical, logical, abstract, concrete, real, or imagined domains, etc., all are equally accessible to modeling.
SA has only a few primitive concepts from which all of its properties derive, the first being the definition of modeling:
The sole graphic primitive is the rectangular SA Box which has (clockwise from the left side) Input, Control, Output, and Mechanism sides to which directed arrows attach. Only Output arrows point away from their side (the right side of the box), and arrows may branch (fork and join) as they interconnect box sides. The semantics of the SA Box are defined by the Formal Saying "Under Control, Input is transformed into Output by the Mechanism, which augments the inherited means" (to carry out the transformation).
Boxes are named and arrows are labeled. In Activity modeling, box names are descriptive verb phrases and arrow labels are prescriptive noun phrases. In both A and D modeling, the arrow labels (via the Formal Saying) prescribe the definitive requirements that each box meaning must satisfy. Hence mnemonic box names only describe. In Data modeling, verb and noun roles interchange, so A and D modelings are a duality, and there is no syntactic difference at all between them. There is, however, a real semantic difference — but it is only an option that need not be exercised.
In Data modeling, Input activity creates and Output activity uses the data (represented by the box), and Control effectively says when — i.e. the circumstances under which those events that change or use data take place. The option distinguishing between A and D modeling is that Control may be omitted entirely from a Data box. Such D modeling is strictly definitive because there is no time constraint saying when it does not apply! The modeled facts cannot be ruled out. By duality, it is Input that may be omitted entirely in A modeling, because since Output is required to satisfy Control, the Mechanism itself must supply any ingredients not supplied by Control (or Input). An example is a copier that produces a Mechanism-supplied paper copy unless (in addition to the Control master to be copied) a transparency Input also is supplied manually. In either A or D modeling, Output never may be totally omitted, but Mechanism arrows often are. Note that optional omissions of this sort do not weaken models; they make them more general by easing constraints. The Reader/Author Cycle determines the best balance in each case.
Modeling
The hierarchic "top-down" modeling of SA (common to SADT and IDEF0) arises from the Structured Analysis Maxim:
The pieces are modeled by SA Boxes, of course. A model always has a top box that models the "Everything" about the subject, and the Maxim applies to each still-worthy child box which itself becomes a parent box, with details on a deeper child diagram. Output-to-Input and Output-to-Control connections specify the interfaces between the child parts that compose the parent-box whole (which therefore decomposes into those parts). The boxes of a child diagram effectively are "plugged in" into their parent-box hole in the parent diagram (replacing its name by visible, structured details) at each node of the hierarchic decomposition.
Each model has an intrinsic node number coordinate system allowing reference to any meaningful point in its structure relative to the top box (its origin) which is "Mn/AG0" (or "Mn/D0", for a Data model), where "Mn" is a unique model name abbreviation, and "A0" (or "D0") is the node number of that parent box and also of its child diagram. The A0 diagram has children "Al", — and from there on, box numbers extend the system "A21", "A22", etc. The coordinate system extends to individual arrow segments, by appending "ICOM codes". The arrows bounding any box are imagined to be numbered (left to right or top to bottom) to provide a means of reference, relative to that node. Thus "INT/A23.IC2note2" is the complete node reference to the second remark about the meaning of the second Control on the first child of node A23 — itself the third grandchild of the A0 ancestor of the "Do International Business" model —in some Reader/Author interchange. (The embedded "." period indicates that the "INT/A23" diagram should be in view — the standard "q.v." meaning "which see".)
The structural integrity of the "plugging in" of decomposition detailing at a node is specified by tagging the free end of each child diagram boundary arrow with the matching ICOM code derived from the parent box. The functional role of an arrow (indicated by the box side where it is attached) may differ between parent box and child diagram. Thus a child-diagram boundary arrow tagged "I2" from the role of that arrow at the parent box may be "C2" on one child box and may branch to be "I1" on another (on the same diagram!).
Sharing
Boxes with no children of their own are atoms, and collectively, even though they are at various depths in the hierarchy, they comprise the "atomic level" of the model. Always, the atomic level is the current viewpoint of the model — all there is to be seen. It starts out as the "Everything" of the SA Maxim for the top box, but may be augmented here and there by (possibly branching) mechanism arrows (which supply support from other overlapping viewpoints).
Being atomic does not preclude deeper detailing. It just can’t come from the current limited viewpoint. Any atom can acquire subatomic detailing from any node in another sub-model if that sub-model supports (by a complete node reference) some ancestor of the atom (so the atom is in the supported sub-model). Support merely supplies access to detailing, not the detailing itself, which comes through explicit calls between the sub-models. If the viewpoints of the two sub-models are imagined to actually overlap, like layers of cloth spread out on a board, a pin stuck into the board would pierce and connect two atoms, like a dumbbell with an end in each viewpoint. Such a call has bounding arrows on each end — arguments and parameters for caller and called, respectively.
The bounding arrows of the caller box are tagged (next to the box side) with matching ICOM codes of the called box to specify caller-box arguments that are to replace parameters of the called box. (Those not replaced are default parameters.) . In this way, the meaning of the arrow structures of the two viewpoints can intermingle through the Formal Saying semantics of the called box’s details. This is the way that the "chain-mail body armor" coverage actually is constructed. With branching, any sub-model can support any number of other sub-models, and within any such interconnected region of support, any number of such calls can be made.
Skipping nonessential further details, only one item remains that is of importance to the appreciation of Applied Understanding Technology, and that is inheritance. When a parent-box’s meaning is decomposed into child-box parts, only the portion of the parent’s "substance" (the Everything of the Maxim) that satisfies the requirements of the child’s bounding arrow meanings driven by the Formal Saying for that specific box becomes the meaning of that child box. This is inheritance, proper. But the inheritance of any box for which there is an upward-pointing support mechanism arrow potentially is augmented by subatomic calling details.
Specifically, when such a supported box is detailed, a support arrow may branch and support only a few of the child boxes, and this selective branching may continue in further detailing. Only when support stops at a box (and does not show on any child of that box) is the inheritance of that box actually augmented — and even then, the actual shared link(s) may not show up immediately. But once the sub-model to sub-model support is made actual, in this way, then any caller offspring box of one can call any called offspring box of the other to make a specific connection between the sub-models (— actually overlapping the views).
None of the current Object-Oriented (OO) approaches reflect this precision of understanding of inheritance, because none have the mathematical underpinnings to address the topic adequately. The details of this aspect of pure Understanding Technology and their influence on the OO community may be a significant output from the AUT Project.