History of DITA
The history of DITA is the history of its many powerful characteristics - modularity, structured writing, information typing, separation of content from presentation, single-sourcing, minimalism, topic-based, task-orientation, content reuse, conditional processing, localization-friendly, multi-channel, component publishing, usability, consistency, object-orientation, inheritance, specialization, simplified XML.
If you don't understand all these DITA characteristics, you may not have analyzed the DITA Business Case properly - for your organization, or for yourself if you are a professional writer.
You don't have to know how to do all these things to use DITA, but if there is no one in your organization who knows why you should use them, you may have a problem. If you have already been doing some of these things, you will want to know how DITA incorporates them.
- Before 1960
- The 1960's and '70's - Programmed Learning, STOP, QRC, Advance Organizers, Information Mapping
- The 1980's - SGML, IBM Task Orientation, Desktop Publishing, Macintosh Documentation Guidelines, and Information Typing
- The 1990's - Windows Help, IBM Minimalism, and XML
- The 2000's - IBM DITA and the Open Toolkit
- OASIS DITA
The historian of technical communications, R. John Brockmann, researched efforts to document products going back centuries. He finds that some of today's hottest new documentation ideas were present in the work of those creating, documenting, and selling the technology of manufacturing just after the revolutionary war.
( From Millwrights to Shipwrights to the Twenty-First Century: Explorations in a History of Technical Communication in the United States)
Today's computers, with their spectacular graphical interfaces, allow us to present animated visual images, even 3-D models to illustrate complex machinery. But this is not the work of the everyday tech writer. Flash animations and computer-aided design (CAD) demand skills more like those found in a game design team than a lone tech writer and wordsmith.
Brockmann found that two-dimensional images were a key part of 18th century technical documents. And modern ideas like modularity were there in the form of documents which were as often a set of cards as a book. He also found that early work was very user-centered and task oriented, and that it took advantage of knowledge already available to the user.
It seems that much of the change in today's technical documentation is the direct influence of the computer, and for some obvious reasons:
- Software documentation, including online help, had become the paradigm for today's tech writers.
- Since desktop publishing appeared, we rely heavily on the computer to assist us in the preparation of our documentation.
- Theorists of computer documentation thought they were in a new field and simply reinvented practices long in use for explaining ordinary machines.
The 1960's and '70's - Programmed Learning, STOP, QRC, Advance Organizers, Information Mapping
At Harvard in the 1960's, computers were enlisted to become "teaching machines" by the behaviorist B.F.Skinner. His ideas of "programmed learning" still have influences in today's eLearning models. His work required knowledge to be broken down into chunks.
Hughes STOP - (Sequential Thematic Organization of Publications) advocated a storyboard approach with two-page spreads. A large graphic on one page, with clear labels, faces the main explanatory text on the opposite page.
The U.S. Navy published the Quick Reader Comprehension (QRC) method in 1961. It explicitly called for modular documentation that could be reassembled and reused for different purposes, perhaps the first mention of Reuse.
David Ausubel first proposed Advance Organizers in 1960. They are formal versions of the teacher telling the students what will be said (then saying it, then telling them what was said - a summary, in the classic three-step teaching method). Ausubel advocated images and clear titles and subtitles that revealed the structure in a document.
In the mid-'60's Robert Horn (winner of an ACM SIGDOC Lifetime Achievement Award for Documentation) developed Information Mapping techniques and founded the company by that name. Common "Information Types" were identified in dozens of standard document types like user manuals, policy and procedure manuals, annual reports, etc. Identifying standard information types is at the heart of DITA (Darwin Information Typing Architecture).
In the late '60's, Charles Goldfarb, Edward Mosher and Raymond Lorie (whose surname initials were used by Goldfarb to make up the acronym GML) created IBM's Generalized Markup Language for documents. In 1974, GML became SGML, with the help of Yuri Rubinsky and others. SGML was the standard for many years of structured documents in the military, aerospace, and large computer companies. It became the basis of DocBook.
In 1973, David Wooley at the University of Illinois developed PLATO Notes, a kind of message board. Posted topics were the basis of an online community supporting the PLATO timesharing system. Ray Ozzie used PLATO Notes as a student at Illinois and in the 1990's created Lotus Notes, including some features of PLATO notes.
The 1980's - SGML, IBM Task Orientation, Desktop Publishing, Macintosh Documentation Guidelines, and Online Help
In 1980, the ANSI standard committee for Information Processing published the first working draft of the SGML standard. SGML was the standard for many years of structured documents in the military, aerospace, and large computer companies. It became the basis of DocBook.
In 1981 a team at IBM led by Fred Bethke called for a new "task orientation" in computer software documentation. Their report, IBM Improving usability of publications (1981), contrasted documents that reflected the software systems architecture. They found that a user had to already understand the software to find the help they needed. Inexperienced users got lost. Another approach had been role-based documentation. But the new idea for Bethke was task orientation, which deals with the tasks people commonly perform with computer programs, regardless of their job titles, and focuses on the information needed to perform the tasks.
In 1981 Interleaf introduced technical publishing software for document authoring and composition. It included word processing, graphics, charts, tables, equations, image editing, and automatic page layout. Interleaf automatically generated indexes and tables of contents for books, and featured conditional processing of content.
In 1983 IBM's Santa Teresa Laboratory published Producing Quality Technical Information
(now unavailable), guidelines for technical writing, mostly within IBM
and mostly for software documentation. The team of writers included
Fred Bethke, whose earlier IBM Publishing Guidelines has established
the importance of task orientation. They identified seven quality
characteristics as task orientation, organization, entry points,
clarity, visual communication, accuracy, and completeness.
In 1984, the new Apple Macintosh was a revolution in computer user interfaces and a similar revolution in computer documentation. The user interface for documents was WYSIWYG (what you see is what you get - when you print the document). Affordable Desktop Publishing was born. The first DTP program, the $99 MacPublisher, was created by Bob Doyle, in the year of the Mac. Aldus (later Adobe) PageMaker followed in 1985. These tools led technical writers to style their documents and even arrange the content layout on the page. To this day DTP thinking is the most important inhibitor of content reuse, mixing presentation with content.
The new Macintosh Documentation Guidelines called for three sections. A Learning overview with tutorials that introduce new concepts and functions, an extensive Using section that spells out how to accomplish tasks, and a program Reference section. To this day, well written books on computers (for example those from O'Reilly) have Learning (e.g., Learning PHP), Using (e.g., Programming PHP), and Definitive Reference volumes.
Note how Learning, Using, and Reference map perfectly onto the three DITA information types specialized from the basic DITA Topic structure - Concept, Task, and Reference. And note that the Macintosh "Using" section was task-oriented, just as IBM was recommending.
In 1984 Lotus introduced their spreadsheet program 1-2-3, which was later the first software to use the F1 key to invoke context-sensitive topic-based online help.
In 1986 FrameMaker was introduced on the Sun OS. This DTP program was designed for long-form documents like books. It became very popular among professional tech writers and at $2500 was a major competitor for the much more expensive Interleaf system.
In 1987 Ralph Walden wrote Microsoft 's first online Help system, QuickHelp, for MS-DOS. He would later develop WinHelp and HTML Help.
In 1986 R. John Brockmann published Writing Better Computer User Documentation. Brockmann described the changes needed to move from paper docs to online. He reported on the new task-based approach, which limits information to that needed to perform a single task, assuming that the user can find general information elsewhere, or very likely already knows it.
In 1988 SoftQuad founder Yuri Rubinsky gave his high-school friend Peter Sharpe the task of developing Author/Editor for SGML, the first specialized SGML editing application.
In 1989 Bob Horn published Mapping Hypertext, an extraordinary book with fantastic illustrations - all drawn by Horn himself - exhibiting the kind of structured writing that Information Mapping was proposing for all documentation. This is still one of the most important books in the history of documentation in general (it's not about computer docs). The book described the seven information types of a structured document - classification, concept, principle, procedure, process, structure, and fact. Horn was inspired by Harvard Professor George Miller's famous work on the Magical Number Seven (plus or minus two) as the number of things easily learned at one time.
Learning theorist Dr. Ruth Clark would trimmed these down to five - concept, principle, procedure, process, and fact - her information types for Training and eLearning - in her workshops and book Developing Technical Training: A Structured Approach for Developing Classroom and Computer-based Instructional Materials. But Clark says the idea for these five types came from instructional theorist M.David Merrill and his "Content-Performance Matrix," not from Information Mapping.
The 1990's - IBM Minimalism, Windows Help, and XML
In 1990 MIT Press published the research results of another IBM team led by John M. Carroll. Carroll's book, The Nurnberg Funnel introduced the idea of minimalism in technical writing. It was task orientation carried to an extreme. Minimalism meant small non-linear chunks readable in any order. It emphasized reading To Do, not reading To Know or To Learn, a phrase first introduced by Ginny Reddish. It attacked the standard systems approach to learning of Gagne and Briggs, with its hierarchical decomposition of learning objectives, which remains to this day as a standard in learning systems. And it emphasized handling errors when the user could not accomplish a task.
Minimalism included the earlier IBM task-based approach, and it limited instructions to the bare minimum needed to perform a single task, assuming that the user can find general information elsewhere, or very likely already knows it.
William Horton published Designing and Writing Online Documentation: Help Files to Hypertext in 1990. It contains clear references to many of the most important concepts in technical writing - task orientation and topic-based content, single sourcing and reuse, and conditional processing, Horton called for new names for tech writers - "document weavers" and "topic writers." Horton's topics had topic sentences, smooth transitions, and summaries. These are difficult to accomplish when online topics are written to be read in any order, and there is no beginning, middle, and end (p.216).
Also in 1990 Microsoft introduced WinHelp 1.0, developed by Ralph Walden, Cheryl Zubak, and others.
In 1991 Ed Weiss produced a book - How To Write Usable User Documentation - on structured and modular documentation that was itself an excellent example of structured and modular documentation, following closely the STOP methodology developed in the 1960's.
In 1991 Sun Microsystems introduced FrameBuilder, a version of FrameMaker with added support for SGML.
In 1991 Arbortext released Adept, their SGML editor, later to be known as Epic, and finally simply the Arbortext editor, when Arbortext was acquired by PTC.
In 1992 Blue Sky software released RoboHelp, a task-oriented, topic-based, online Help Authoring Tool (HAT). Later they changed the company name to eHelp. eHelp was acquired by Macromedia, prinmarily to get the Flash-based tool RoboDemo (now Captivate). RoboHelp was discontinued, but after Macromedia was acquired by Adobe, RoboiHelp became a strong part of Adobe's Technical Communication Suite, including FrameMaker.
In 1992 at Lotus a team of user assistance specialists started to design a single "Working Together" common help design for Lotus' office package SmartSuite and Lotus Notes. John Hunt (1-2-3), Janet Smith (Freelance Graphics), Bryan Steh (Word Pro), and Susanna Doyle (Notes) created a core design with six topic types: overview, context-sensitive, steps, details, examples, glossary, and reference. It used Notes for content management and WordPro templates for editing, with single-source/multiple output to a variety of delivery formats.
In 1992 the HyTime standard for Hypermedia and Time-based content (an application of the SGML architecture) identified problems with linking document types that was to inform the specialization mechanism in DITA.
In 1994 JoAnn Hackos published her landmark Managing Your Documentation Projects, revised and republished as Information Development: Managing Your Documentation Projects, Portfolio, and People by Wiley in 2006. Fully in tune with task orientation, Hackos book described only three information types - concept, procedure, and reference (p.236). This seems to be a combination of Information Mapping's seven types, Ruth Clark's simplification to five types, and Apple Macintosh Documentation Guidelines three components.
In the mid '90's, Yuri Rubinsky's team at SoftQuad (creators of one of the first and most popular HTML editors, HoTMetaL, became involved in the development of a compromise markup language somewhere between the extraordinarily complex SGML and the popular new HTML (Hypertext Markup Language) for web pages. (HoTMetaL was the precursor to today's XMetaL from Justsystems.) HTML was a disaster from the point of structured reusable component documentation, not least because it combined presentation markup with structural markup. The new markup language was XML (eXtensible Markup Language), and SoftQuad developed one of the first XML editors, XMetaL.
In November 1995 John Carroll convened a workshop, sponsored by the Society of Technical Communication (STC), to evaluate Minimalism in the years since the Nurnberg Funnel. Carroll invited his major colleagues - R. John Brockmann, David Farkas, JoAnn Hackos, Hans van der Meij, Janice C. (Ginny) Reddish, and others.
In 1995 Adobe acquired FrameMaker and FrameBuilder, which was to become FrameMaker + SGML, and eventually the more affordable Structured FrameMaker, now included with every copy of FrameMaker, though used by a small percentage of tech writers. Most writers continue to prefer unstructured documents.
In 1995 CNET Founders Halsey Minor and Jonathan Rosenberg built their own Web content management system and it introduced a number of today's core capabilities, like content reuse and personalization. Page templates assembled the content dynamically from a relational database. They sold the system to Vignette for a share in that new company. It was the first "content management system (CMS)."
In Toronto in 1996, an IBM documentation team including Michael Priestley, Bob Fraser, Dennis Bockus, and Jamie Roberts was developing a Help system for IBM's new line of VisualAge software. Roberts had just returned from graduate study at University of Waterloo and attended a brainstorming session to define some basic information topic types for the new Help. He was inspired by Lotus' online help in 1-2-3, which then had the reputation of being very good at user experience. Lotus Help included procedures (called "steps") and overview (concept). Roberts scribbled "concept, task, and reference" on a napkin, handed it to Bockus for implementation, and a new help document architecture was born. There is not much unusual about a Help system that is task-based and assembled from topics. What was new was that this was to become the simplified form of XML known as DITA, with very significant contributions from Lotus, which IBM had just acquired. And it was to be both web-based and delivered as a PDF.
After the release of Windows 95 and WinHelp 4, in 1996 Scott Boggan, David Farkas, and Joe Welinske wrote Developing Online Help for Windows 95. It had a strong task orientation and was topic-based. But "concept/overview" was only one of ten standard topics, which did not include "reference," but did wisely include errors and troubleshooting.
In 1996, IBM signed a long-term agreement to use the Arbortext Adept editor for internal SGML document creation.
In 1998, JoAnn Hackos and Ginny Reddish published the definitive reference on task analysis, User and Task Analysis for Interface Design, and John Carroll published the edited proceedings of his 1995 workshop, Minimalism Beyond the Nurnberg Funnel, with major contributions by Hackos and Reddish.
In 1998, IBM revised their 1980's PQTI tech writing guide, retitling it Developing Quality Technical Information. The team of writers was led by Gretchen Hargis and included only one from the PQTI team, Polly Hughes. They now cited nine quality characteristics - accuracy, clarity, completeness, concreteness, organization, retrievability, style, task orientation, and visual effectiveness. Note that task orientation had slipped from first to eight out of nine quality characteristics.
The 2000's - IBM DITA
In March 2001, IBM introduced DITA as a series of developerWorks articles about a new simplified version of XML for documentation. It was intended to replace IBM's IBM ID Doc, an internal version of SGML for IBM's technical software support. While XML was enjoying great use as a data exchange method (RSS and SOAP protocols), it had little traction as a document markup language. DITA was an attempt to make a simplified XML starter set for documentation markup, one designed from the outset to encourage reuse of small content components. The key ideas were to be simpler than the complex SGML and also be usable online.
The goal of DITA was to formalize information typing practices, both print and online, and also enable an extensible typing architecture through specialization of base topics. DITA maps were a way to standardize collection publishing and information architecture/outlining models. DITA was initially known as MITA, for Mendel Information Typing Architecture, to emphasize the object orientation of the new architecture, with its "inheritance" and evolution of topic structures via specialization. Since MITA was already a somewhat proprietary acronym, IBM switched to Darwin and DITA.
In May 2002, IBM added domain specialization to topic specialization, and demonstrated these in the Open Toolkit, a reference implementation of DITA publishing, with a starter set of XSLT stylesheets. IBM encouraged authoring tool vendors to integrate the Open Toolkit as a means of publishing DITA, and most have done so.
JoAnn Hackos' Content Management for Dynamic Web Delivery was published in 2002. She described creating an information (content) model and developing information types, such as procedures, concepts, warnings, specifications, and others. Michael Priestley, the DITA specialization architect, and Dave Schell, then the principal DITA evangelist for IBM, wrote a 3-page vignette on DITA, perhaps its first mention in a book. They stressed topic orientation, information typing, specialization, inheritance, and two architectures, one for information and one for specialization. Hackos briefly mentions the AutoCAD Learning Assistance software she and learning guru Wayne Hodgins developed for Autodesk. It was cleverly dubbed CPR for its three-tabbed interface to concept, procedure, and reference.
In 2003, IBM published a second edition of Developing Quality Technical Information, by Gretchen Hargis and others. Now the nine quality characteristics were rearranged once more, putting task orientation first again. More importantly, they added an introductory chapter that called for content to be structured as separate information types, specifically task, concept, and reference. (Note the correct order of the three basic DITA information types.) This book is all about DITA without mentioning the name, probably because IBM was using DITA internally but not yet sharing it with the world when the book was drafted.
In April 2004, the Organization for the Advancement of Structured Information Standards (OASIS), formed a Technical Committee to explore a DITA Standard. The TC included XML tools vendors, consultants on Information Architecture and Content Management Systems (CMS), and end users of the DITA Document Type Definitions (DTD) and Schemas needed for the new DITA Standard.
In February 2005, IBM donated the Open Toolkit, a limited version of their internal Information Developers Workbench, to SourceForge. IBM continues to develop the OT, which is not a part of the OASIS DITA Standard efforts.
DITA 1.0 was approved as an OASIS Standard in June 2005
DITA 1.1 was approved in August 2007, adding a new Bookmap specialization.
DITA 1.2 is expected sometime in 2008. It will add structured learning, creation of Learning Objects with DITA, which will be compatible with eLearning standards such as SCORM.
How To Write Usable User Documentation, by Edmond Weiss, Oryx, (1991)
Managing Your Documentation Projects, by JoAnn Hackos (Wiley, 1994).
Developing Online Help for Windows 95, by Scott Boggan, David Farkas, and Joe Welinske, (Solutions, 1996).
Standards for Online Communication, by JoAnn Hackos (Wiley, 1997).
Two approaches to modularity (1999). Robert Horn compares structured writing to Hughes STOP.
Review of the Nurnberg Funnel (1999) Robert Horn compares structured writing to Minimalism.
The Impact of Single Sourcing and Technology, Ann Rockley, 2001.
Content Management for Dynamic Web Delivery, by JoAnn Hackos (Wiley 2002).
Managing Enterprise Content, by Ann Rockley, New Riders, 2003.
Single sourcing: Building Modular Documentation, by Kurt Ament, Andrew Publishing, 2003.
Developing Quality Technical Information: A Handbook for Writers and Editors (2nd Edition) , by Gretchen Hargis, Michelle Carey, Ann Kilty Hernandez, Polly Hughes, Deirdre Longo, Shannon Rouiller, Elizabeth Wilde (IBM Press, Information Management Series, 2004).
Information Development: Managing Your Documentation Projects, Portfolio, and People, by JoAnn Hackos (Wiley, 2006).