Present: Books and Managing Complexity (with computers) Date: , Page ©Copyright 1996, Rainer von Königslöw. All rights reserved.
15 York Valley Cr., North York, Ont., M2P 1A8Tel.: (416) 410-2202, Fax: 489-4688
Books and Managing Complexity Draft: for review only Rainer von Königslöw Version 0.1: June 14, 1997 Copyright 1997, All rights reserved. Diaspora and convergence: the current state of books Types of books: Books as entities Company books - business plans, accounts & financial reports, "book keepers" electronic books, databases as books lab books - scientific evidence courts: collecting the paperwork civil service - managing policy and complexity with written records advertising, persuasion, hate mongering, misrepresentation Description - record keeping Encyclopedias Reference texts books of accounts, financial statements lab notes of experiment results medical histories court case evidence reports architectural & engineering drawings documentary movies anthropological - ethnological recordings Story telling, reporting, persuading, entertaining novels, romances, legends, children's books religious legends movies Instructions - Constraints Laws - acts, regulations, bylaws, codes religious commandments "how to" books, user guides listings of computer programs Hidden books tables in many databases files in government archives (not accesible by Freedom of Information, or not known to exist) some medical record adoption records Usage scenarios for books Application of book-related skills & technologies accountability, communication, reasoning, public scutiny (sharing) How much of the moment to moment behaviour is captured by written rules: could we capture and automate the process with expert systems or transaction oriented system? Accounting Lists, inventory, order processing, money & balance & value, financial planning & budgets & futures, evaluation of an organization, shares bank account books (still used for saving accounts) Management, administration HR job descriptions management reports memos, judgements, evaluations expressed on paper Science "publish or perish" -- extend duality with new measurement schemes, new theories, new observations, new experiments, etc. public documents allowing incrementality, requirement for novelty, review & judgement of peers - juried justice legislation, court system, rules of evidence, precedence cases very much dominated by the paper trail, linking arguments to legislative code, previous documented interpretations, precedence cases architecture moving into paper mode with models (pictures) of future views, engineering drawings meeting constraints medicine slowly moving into paper mode, but not fully yet history, tests, images, treatment plan & schedule, prescriptions billing for small job components & processes, insurance forms, advising & warning patients (clients) to avoid liability Administration, civil service managing paper trail, data bases, flow of paper information with forms & lists inventories, order processing exception: driving a car managing a car in complex situations: traffic laws mostly constraints, owner manual covers routine, constraints, exception events -- usual combination of looking, steering, braking, accelerating -- not covered by book rule -- cannot be automated easily by expert system exception: politics not yet govered by paper flow? - (Liberal red book) primarily argument ad hominem (and ad valorem?) "vote for me, I am this type of person and have the following general values -- no detailed promises or accountability for past actions Where we get into more accountability we tend to switch parties -- can't properly compare in power vs out of power candidates or parties; can't compare past actions & promises Role of writers, publishers, advertisers, ... Role of schools, other training, ... schooling through highschool How much is learning to deal with symbolically expressed information, paperflow How much is memorizing specific information to actually use later in life exactly as learned - unchanged and unreviewed How much as learning language skills (including math) other skills? How do we use a book. speed: reading, writing, typing, looking up references, finding things reading printed book flexible in where to use easy to read sequentially electronic book requires computer equipment to use not as flexible in where to use not as comfortable to read sequentially The electronic book requires a computer and other peripherals, such as a monitor and a CD-ROM so that usage costs are higher and flexibility of usage is restricted. This is less of a factor if computers are amortized over other usage, and if notebook computers are used for travel access. There is a weight trade-off for large books: 1 million pages = 0.5 million sheets of paper vs. notebook computer plus CD-ROM player plus CD-ROMs. searching for information printed book it is fairly easy to find specific information that is listed in an index or TOC - by looking up pages it is very difficult to find specific information that is not listed in an index or TOC it is fairly difficult to follow a reference to another book electronic book The main advantage of electronic books is that they let users "bounce around" in the book easily. There are three main methods of navigation. First, the table of contents and the index work like menus, so that one can jump directly to the content. Second, one can also jump directly from a references to the text being referred to. Third, all of the content is indexed automatically so that one can search for words or phrases and then step to the places where they occur. it is easy to find specific information that is listed in an index or TOC by following an electronic link it is very easy to find specific information that is not listed in an index or TOC by starting a search, obtaining a list of places where the information is found, and finding the information which has been highlighted. Powerful search capabilities give more flexibility for information retrieval, since the search does not have to be preconceived and structured in with a TOC or index. The search is structured at the time of use and is individualized for the user. reference: it is very easy to follow a reference to another book by following an electronic link The electronic book may allow for personal profiling for faster access to frequently used information. sharing & interacting with other readers printed book it is difficult to share the content with another user simultaneously & interact about the content electronic book it is fairly easy to share the content with another user simultaneously & interact about the content (via modem or on a LAN) responding to & interacting with the author printed book it is very difficult to respond to the author about the content it is nearly impossible to interact with the author electronic book response: it is easy to respond to the author about the content (with E-Mail) interact: it is possible to interact with the author (on a Bulleting Board, or via E-Mail) annotating & editing printed book it is slightly awkward to annotate & edit the content for private use it is difficult to share annotations with others & jointly edit the content for group use (e.g., model agreements) electronic book Users may mark up the text by leaving bookmarks, highlighting the text, or adding annotations. Inserting or extracting text from or to a word processing package such as WordPerfect is straightforward. it is easy to annotate & edit the content - by inserting notes or by attaching files it is easy to share & jointly edit re-using the material in the book printed book re-use: it is awkward and difficult to reuse the content, even if permitted electronic book re-use: it is fairly easy to reuse the content, if permitted, by extracting the text into a file The electronic book provides more support for information re-use, by allowing selective retrieval of information in printed or in file formats. This information may be integrated into other applications that require the information. skill levels required for usage printed book Basic literacy skills of reading in the appropriate language is sufficient for many books Some books may require background information for comprehension Some books, e.g., mathematics, may require literacy in language extensions, e.g., reading equations, maps and schematics, etc. Basic writing skills are required for annotating and editing electronic book The same basic skills are required as for reading the equivalent printed book. Basic computer usage skills are also required. The electronic book requires some training to install the system & use all the facilities. This time may be amortized over several usages. basic entrance costs for usage printed book Access costs for the book, e.g., purchase, rental -- generally very low electronic book Access costs for the electronic book, e.g., purchase, rental -- generally low, but higher than for printed books The electronic book requires a computer and software to access the information. The cost can be amortized over multiple usage (or it may already be available) Extended usage topics Maintenance printed book Books wear out, can get lost, have to be shelved. electronic book Electronic books have to be maintained as files on a computer, adapted to computer upgrades, can be modified and re-released, can be copied. Enhancement printed book Books cannot be modified, except through markings in the margins. It is difficult to share any such markings, except through serial lending. Manuals in 3-ring binders can be updated by replacing pages - which is a laborious effort. electronic book Electronic books can be updated by being edited, with sections being added and other sections being replaced. Depending on the SW support, they can be annotated, and the annotations can be shared. (FOLIO: not with the runtime version, but with the 'personal edition'.) Security printed book Books can be kept under lock & signed out. electronic book Individual access to an electronic book can be traced. Integration with other tasks printed book Books are highly portable and can be used in most environments, with few exceptions, e.g. wet, underwater electronic book Because of the requirements for a computer, it is harder to integrate the use of electronic books into many tasks. The electronic book can be linked with expert systems, form filling packages, etc. to semi-automate a task requiring both information retrieval and information providing. Access to books Electronic books in an electronic library (CSALT '94) There are now a number of electronic books available in CD-ROM or diskette form. Many firms are producing and "publishing" their own electronic books, generally for internal use. Electronic books are files on computers that can be viewed on a monitor with special software. Once one has many electronic books they should be housed in an "electronic library". We shall start off with the questions "what is an electronic library?" and "what functionality should it provide?". The next question is "how can we develop and maintain an electronic library?". The last question is "how can we train users to use an electronic library to best advantage?". What is an electronic library? Just as a traditional library provides and supports shared access to physical books, an electronic library provides and supports electronic access and sharing of electronic books. In practice this means that users can use their local computer to access electronic books stored centrally on a file server or on a CD-ROM reader attached to a central facility. A catalogue is provided in electronic form, and access is managed on a first come first served basis, or with multiple simultaneous shared access. To simplify finding titles, the files for the electronic books can be arranged in standardized directories and with standardized file names which mirror the stacks and call numbers of a traditional library. As new titles are acquired and added to the collection or older ones removed, the catalogue is updated. What functionality should an electronic library provide? In general, an electronic library should provide all the functionality of a traditional library, with additional benefits due to distributed and simultaneous access: Books are rarely inaccessible, do not get lost, and do not need to be reshelved. An electronic library can also include access to remote collections of electronic books through wide area networks or through links to online databases. Additional benefits can be realized by adding functionality to the catalogue, to the electronic books, and re-engineering the work habits of users. We will illustrate below. How can we develop and maintain an electronic library? An electronic library requires hardware, software, and titles (electronic books). The most common hardware configuration is PCs (or Macs) in the user's office, connected with a local area network to a fileserver. Another PC in the librarian's office provides management access. CD-ROM players may be attached to the network to provide access to less used titles. Simple CD players require manual intervention to load the CDs, but "jukeboxes" are available which automatically load the CD. Modems provide dial-out capabilities to databases and to remote electronic libraries. The electronic book software ("viewer") such as Folio Views must be purchased for DOS, Windows or the Mac as required. Communications software may be required to provide access to on-line services or to remote libraries. Electronic book titles can be licensed from commercial publishers such as Carswell, CCH, and Prentice Hall. Single user, multiple user, and unlimited use licenses are available. Electronic book titles may be developed in-house to support litigation for a complex case, to provide access to internal research memos, to support re-use of the contents of previous documents in crafting new documents, or to publish material such as internal policies and procedures.. The electronic catalogue ("directory of books") is an electronic book which is developed in-house. It provides automatic access to electronic books in the collection. Table of contents and other catalogue information can generally be extracted automatically from electronic books and online databases to provide the core content of the catalogue. It is hoped that in the future vendors of electronic books and database services will provide catalogue information in electronic form to simplify adding their material to the library. The loading instructions for CDs in "jukeboxes" can be embedded in the catalogue. Table of contents for the databases and remote libraries can be added to the catalogue. References to traditional books and to word processing documents may also be added. Further extensions to the catalogue will be discussed below. How can we train users to use an electronic library to best advantage? In my experience generic training in the technology doesn't work. One has to focus on a specific application, on usage of the technology in context. There are three elements to this question. How can we train users to use the existing applications? How can we develop a more integrated, user-friendly electronic library? How can we re-design and re-engineer some of the work to take best advantage of electronic libraries? How can we train users to use the existing applications? The most common applications are: - support for legal research - support for document drafting - generally based on internal material - litigation support (fast internal publication of transcripts & supporting documents) The major benefits from use of an electronic library are: - the speed of navigating through voluminous material - the ability to find quickly all relevant chunks of information The major disadvantage of current implementations of electronic libraries arises from the lack of integration of existing titles. Generally more than one software package has to be mastered. Even where there is standardization, e.g., using Folio Views for the catalogue as well as for most of the internal and commercial publications, the catalogue does not support rapid navigation. Generally there is a three step approach: look up the book in the catalogue, then look through the TOC or index for a selected book, then look for the information. Using a word or phrase search still involve three steps: select the book from the catalogue, formulate a word or phrase search within the book, then scroll through the "hits" to find relevant chunks of information. Automation with hyperlinks and searches generally is supported only within books but not across books. Going through several books or following indirect references involves repeats of the above scenarios. How can we develop a more integrated, user-friendly electronic library? The goal is to be able to navigate quickly and easily to narrowly defined chunks of information. Narrowly defined chunks means text which is easily visible, i.e., is visible within one window or with little scrolling. A more tightly integrated electronic library would extend the catalogue so that rather than pointing to entire books it could point to smaller meaningful chunks of information. This may involve concatenating call numbers with chapter, section and paragraph numbers that are internal to the material in the book. With hyperlinks one can then jump automatically from the catalogue directly to the paragraph within the book. The same approach can be used for following embedded references from one electronic book to another. (This approach is being beta tested for a "mini-library", a collection of primary and secondary materials dealing with tax.) A second concept involves time stamping individually named chunks of information, as well as the links to the information. This allows keeping time variants of information chunks in the library, such as revisions of sections of legislation. One can then ask the catalogue, say, to reconstruct a statute as it was current on a specific date by reassembling the appropriate chunks according to their time stamps. (This approach has been successfully tested with a pension plan.) The library should also support applying a word or phrase search across multiple electronic books. It would be useful to then have the same search be applied automatically to an online database. The internet as a world-wide library Content of books "Atomic" components, organized by sensory modality sight text letters, katanji (syllables), Chinese, pictogram, runes? numbers, tables, mathematical equations secret codes, encryption Text is associated with language and communication, i.e., transferring information from one person to another. It is not the only form of communication, but it is the most common. It is artificial, i.e., it does not occur in nature without man. It has two main forms, an alphabetic representation and a pictographic representation. It is the written form of language and speech, and it is highly standardized. English, using the roman alphabet, and the Oxford dictionnary, can be seen as one of the standards that it applied to communication. Text, unlike speech, has a life independent of the originator. It is organized into objects that persist over time and can be distributed over space. To show parallels, tape recording and the radio allow sound, including speech to persist over time and be distributed over space. The video recording and TV has done a similar thing for sight as well. Two predominant examples of text objects are books and letters. Other examples are newspapers, magazines, manuals, and documents. Forms and tables are on the boundary between text and data. Comic books are on the boundary between text and pictures. Books and letters are visual objects, in the sense that they depend on sight to be read & deciphered. Books and letters can also include other, non textual visual information in the form of static pictures. We shall focus on books, and discuss variants later. The main form we are familiar with is a printed book. format of text text structures: paragraph, chapter, heading text boxes: page, margin, header, footer, footnote, sidebar anchored & floating frames for graphics, text, images, tables, equations, video, sound hyperlinks drill-through applications, active objects web browser simulation spreadsheet (& simulation) database (& active data IO) lists - enumeration pictures, images colour planes, photos, rasterized images (pixels, palettes, colour planes) line drawings patterns, representations, cartoons, engineering drawings & schemas, calliography? vectorized images (lines & points) video movies sculpture sound recorded speech recorded music touch Braille text sculpture? 3 dimensional fold-outs, textured smell perfumed pages rub & smell in children's books other not components of books sight Theatre performances, dance, mime signs & signals semaphores (on ships), sign language for deaf, heliophones & blinnking signal lights (Morse code) sound speech music "Molecular" components, organized by ? data descriptions, stories procedures & instructions programs metaphors, poetry structure of intellectual & artistic content common title TOC List or tables & illustrations cover art, blurb copyright notice, acknowledgements, introduction, etc. Chapters, headings, paragraphs Appendices printed book Index footnotes & references running headers & footers, pagination electronic book automatic TOC automatic & complete Index automatically linking footnotes & references limited cover art, blurb no running headers & footers, pagination format of intellectual & artistic content printed book fonts, sizes & other attributes of the text pictures, drawings - colour, gray scales tables, foldouts, forms electronic book limited fonts, sizes & other attributes of the text limited resolution pictures, drawings - colour, gray scales - linked by special viewer that zooms & scrolls for display and prints tables - scrollable horizontal & vertical no foldouts automatic link to forms SW, to display, print, and fill interactively intellectual & artistic content: common ideas, factual material, persuasive arguments, fantasy, commitments Content 1. Scale-up capability: up to 2 gigabytes (almost 4 gigabytes of compressed text). In comparison, a book @50 lines per page and 80chars per line = 4,000 chars per page. 4 gigabytes = 1 million pages. 2. More flexibility in structured access: multiple table of contents, tabs, hyperlinks 3. The electronic book has less flexibility in layout & design and the use of visual presentation devices such as fonts, point size, bolding & italics, etc. This disadvantage might disappear with future versions with SMGL or other markup languages to tag the text, and fast GUI displays. The electronic book has the potential for more flexibility with multi-media presentations, including videos etc. Smell and feel (texture) are more likely to be easier for books. books as physical objects printed book size weight can be damaged - break spine, rip pages, water, dirt has a cover has pages electronic book electronic files on HD, on LAN electronic files on CD-ROM, on diskette Developmental lifecycle of books: How do we generate/use/retire a book. authoring (writing) printed book long delay between writing & use (sales) difficult & slow to improve the product (until a second edition) high business risk electronic book with electronic distribution, the delay between completion of the writing and use can be very short since there are no inventories, new releases can be issued very quickly, and transparent to the user. production (editing, designing & printing) printed book mass production relatively slow turn-around very expensive for small volumes fairly long delay to recover investment through sales fairly high business risks electronic book reduce the cost for production reduce the time for production need for printing depends on media for distribution distribution (distributing & selling) printed book mass distribution through retail outlets easy to store & transport requires inventory may have out-of-stock & waiting for reprint problems electronic book as file faster distribution no large inventories never out of stock & need to wait for reprints electronic book as CD-ROM or diskette less weight per amount of content lower cost to ship, store per amount of content access (buying, storing & retrieving) printed book inexpensive to acquire fairly easy to store & retrieve subject to loss & damage electronic book via electronic distribution almost instant acquisition requires computer equipment to acquire, store & retrieve may allow multiple access to single copies subject to loss & damage as file electronic book as CD-ROM like printed books, but more compact, requires less space subject to loss & damage Environmental impact printed book Books require paper, which is mostly based on trees, a renewable resource. The paper can also be recycled, but not easily outside of the major metropolitan centers. electronic book Electronic books require fewer resources and produce little or no garbage. Generation & access to printed books Generation book selection access: content selection Access to electronic books with FOLIO SW Generation book selection access: content selection Electronic references to electronic books with FOLIO SW reference: content selection reference: presentation & user interface Project: definition & requirements Players (interested parties) 1. Users a) classes b) representative(s) 2. Info providers a) classes b) representative(s) 3. Info management (librarian role) 4. Project management Objectives 1. Strategic a) productivity increase b) usability & access to info c) info distribution d) timeliness of info 2. Tactical a) acceptance & satisfaction criteria Constraints 1. cost, timing, people access 2. target platform, performance Trade-offs 1. priorities Methodology 1. Rapid prototyping vs. one-shot 2. Quality assurance (manual vs automated) 3. Technology transfer, training, support 4. Documentation (formats, content) User access 1. finding info a) TOC - multiple, hierarchical b) searches c) key words 2. retrieving info a) look only b) extract & retrieve for re-use 3. modify, add, delete 4. Security a) all, selected parts b) multi-level c) type of use 5. Ergonomics a) usability b) skill level required 6. Equipment accessability a) at work b) portable c) at home 7. Support a) hot line b) training 8. Motivation, career development Info management 1. Quality assurance - info integration 2. Version control 3. Distribution 4. Update projects a) information addition b) enhancements c) redesign 5. Support & training Info providers 1. Models a) structure (multiple infobases, segmentation) b) content (char & attribute mapping) 2. Quality assurance for info provided a) methodology b) support utilities 3. Technology transfer, training, support 4. Process for entering & adding information a) ease of use b) efficiency & effectiveness c) skill requirements d) error tolerance & recovery 5. Responsibilities for content, etc. 6. Constraints a) adherence to procedures, sign-offs b) timing 7. Motivation Project management 1. Schedule 2. Budget 3. Tasks & manpower loading Project: architecture & design Structure of information See design section above. Approach to distribution HW & OS base SW tools & utilities publish, maintain: Folio, Windows Help, Acrobat process & edit: Awk, Perl, Rexx, etc. Design for info access & usage Design for info-providing methodology & projects Project: (prototype) development participants 1. apprenticeship training review process stages Project: Acceptance testing Test plan Testing project 1. Content 2. Access 3. Distribution 4. Updating process 5. Configuration for target platform 6. Performance Project: training & support Training 1. User 2. Author (info provider) 3. Info manager (librarian) 4. Instructors & support personnel Support 1. User hotline 2. Technical support 3. Design support Business case Production 1. Lower production time & cost especially for updates. Once the content is prepared, preparing the electronic book is faster and more inexpensive than printing. Maximum infobase: 2 gigabytes of infobase is approx. 3 to 4 gigabytes of raw text. At 10 pitch & 1 inch margins, there are 65 chars per line (78 for 12 pitch). At 6 lines per inch & 1 inch margins there are 53 lines of text on a page, or 3,445 bytes of a fully covered 10 pitch page (4,134 bytes for 12 pitch). A more standard page, with formatting and blank lines between paragraphs usually contains around 2 megabytes per page. This means that a single infobase might contain over 1 million pages. CD-ROM maximal infobase: up to 600 megabytes can be put on a single CD-ROM. Before compression, this could be approximately 1 gigabyte or raw text, or 0.5 million pages (as estimated above). Single copies of CD-ROMs can be produced for approx. $1,000. In volume, production costs drop to under $10. For a single copy, this is 0.2 cents per page. For higher volume this drops down to approx. 0.002 cents per copy. CD-ROM break-even point for a single copy: If one estimates copying costs of 20 cents per page, one can calculate that the break-even point for a CD-ROM is at 1/100 maximal size, or at 6 megabyte size of infobase, or at a document containing approx. 5,000 pages. CD-ROM break-even point for multiple copies: Three CDs cost $1,350 or $450 per disk. The break-even point is now at approx. 2,500 pages. For slighly larger volumes, the cost goes to around $100 per disk. Now the break-even document size is 500 pages. For high volumes, the cost per disk goes below $10, or a break-even point of 50 pages. Diskette break-even point: A 3.5 in diskette can contain over 2 megabytes of raw text, including the run-time & a user manual. This is equivalent to 571 pages, for a cost of $1.50. Even including the cost of a run-time license at approx. $25, the diskette version is likely to be less expensive than a paper copy. Even if photocopied at $0.05 per page, the paper copy would cost $28 without factoring in document assembly and packaging as well as labour. Shipping costs are lower for diskettes and CD-ROMs at even quite low page counts. 2. Design & layout costs are probably about the same, possibly slightly higher for low-end items (no professional design & layout - just type & reproduce) Distribution 1. Fast, inexpensive distribution over long distances via electronic communication. For larger items, electronic items are lighter than paper (e.g., CD-ROM vs. large book) and therefore would have lower shipping costs. 2. Better control over distribution access - security etc. Our first CD, a cautionary tale Chronology - Discussion started summer 1992, with anticipated start of the project in Sept. 1992 - Sept to April was spent with business plans and marketing forecasts. - May 1993 the development started, with documentation of usage scenarios, project definition, and initial content specification and design specifications. - The first alpha prototype started in time for the developer to go to the hospital for a double hernia operation -- since development continued during convalescence, meeting had to be scheduled in the hospital room. Praise be to notebook computers. - A fair bit of effort was spent to get a complete model of the content, especially since much of the content was to be based on existing publications which mixed together inhomogeneous material from different sources and with different authors. - A model was built of the content, with separate buckets for each homogeneous type of information (i.e., from a single source / author / editorial process. - Each bucket was analysed for the struture of information within that bucket, so that information could be separated into separately meaningful chunks. Chunks may be embedded in other, higher-level chunks. They may even be nested. - Each chunk of information was assigned a unique name, so that it could be referenced and jumped to. (This was a major design principle underlying the basic architecture of the system.) Each such chunk could be described and named based on natural criteria such as sections, clauses and paragraphs for legislation, and "style of cause" for court cases. A "library" model was developed for combining the buckets of information. An index analogous to a central annotated catalogue was developed. Each individually named chunk of information was entitled to an entry in the catalogue. Each entry in the catalogue referred back to the source as well as to other relevant chunks of information. These references were also used as hypertext links to directly jump to the relevant entries in the source text. (This library + catalogue model was a second major design principle underlying the system.) The library model was also used to organize the content. Homogeneous material would be found in a single book. Closely related material would be found together on a library stack. Each stack might contain a list of books, where each book contained all the material from a separate bucket of information. Equivalent parts of books would be represented similarly in the Table of Contents. (The hierarchichal tree structure of the table of contents was a third major design principle underlying the system.) - While the design was progressing though a series of prototypes, the discussion on content was continuing. Content acquisition proved to be a major challenge. For the beta test prototype we would include only content which was available in electronic form. A small percentage of the content was coded in SGML. Other content wqs available only in various typesetting codes and had to be transformed. Another large chunk of content (the full text of relevant cases) was available only in an on-line data-base system at a remote location. Another large bucket of information (forms) was available either to be downloaded from a government bulletin board in one format, or as transformed files which were a component of another electronic product. The SW underlying the government files was not available in a run-time version so that it could not be licensed at reasonable cost for inclusion in the product. The second version, the transformed files, were incomplete, many were out-of-date, and lacked systematic quality assurance. (Eventually it was decided not to include the forms.) Context - The CD-ROM product design and development was taking place while the company wqs undergoing a second technological transformation, the adoption of SGML to represent the information to be published. The SGML was meant to be the basis for maintaining and updating the information for an anticipated monthly re-release of the information. - Interestingly enough, the SGML project was quite separate and independent, and with a different design philosophy and architecture. In both cases most of the design was done by external consultants. However, despite some informal exchanges, there was no coordination between the two projects. Management seemed indifferent if not hostile to design and architectural coordination, despite some efforts by the outside consultants. Management seemed to be desirous to keep the projects separate. Outcome It seems quite likely that all personnel associated with the project will have left the company, so that all design experience will have been lost to the company. The company of course still has the design as instantiated by the product, but without a clear representation of the vision underlying the design and anticipated future extensions. Interestingly enough the company seem to be applying the same design to other products. Process automation Since the product involves around 120,000 pages of information, automation in the transformation stages was very important. More chronology The first public exposure of the product was at the tax convention in Nov. 93 in Montreal. It was quite difficult to make final changes and do the integration in time. Several planes had to be missed while making final changes and error corrections. We eventually ended up taking a portable computer and a large external hard drive and doing the final integration in the hotel room in Montreal.