Information Architecture: Susan's Blog
A compulsive writer, I started this blog as an outlet for my endless ideas. Your feedback is very welcome!
Political messages need IA? (11-21-2016)
In the wake of the 2016 elections, I thought to review what the parties and candidates had to offer.
HillaryClinton.org has a page for "issues." First of all, I think that is the wrong word. The page opens with about 40 issues contained as text within rectangles, four across. I originally was disgusted by the lack of organization, but later saw that there were (1) six categories plus an "all" itemized in a horizontal line at the top of the "list" and (2) that within each category the issues were ordered alphabetically by the first words. I was not impressed. Then I read a few of the issues and was deeply disappointed. They were composed of a short title (in bold text) followed by one sentence. Now I thought the sentence might actually provide some useful detail for the brief title, but no . . . it proved to be a value statement with no metrics and no plan to achieve them.
DonaldJTrump.com has "positions", which seems a much better word than Clinton's "issues." "Positions" is located in the top "menu" bar. It has a drop-down list of 16 items, none of which is "jobs" (which I expected because of his speeches). The list of 16 items is ordered alphabetically. The detail for most positions is a single page with bullet items grouped by "vision," "key issues", and "contrast with Hillary Clinton." Some of the visions are detailed, while some are vague. There is also a sidebar for graphs and links to videos.
Home page has a link to "Party Platform" in the small print at the bottom of the page. In the same location is a list of 12 "Issues". This location is clearly of low value.
The Party Platform page has 12 "ideas and beliefs" in no particular order. Each has subordinate items, also in no obvious order. One idea/belief has 16 subordinate items, with no easy way to refer to them (such as a letter); instead they have a hollow square bullet.
The "issues" link opens a page with text and image that presumably explains the party's history on the subject and a reason why they are still addressing it.
GOP.com has a list of 17 "Issues" at the bottom of the home page in small print, in alphabetical order; visually an afterthought. There are 11 "principles for American renewal" with links on the home page. Also on the bottom of the home page in the group titled "Our GOP" is an item "Republican Platform"; this opens a page with lots of graphics, some with links to 7 subjects and a "Preamble." I clicked on "Restoring the American Dream" to see a bulleted list of 15 items, in no observable order. A lot of text follows this list consisting of a title for each list item and several paragraphs of explanatory text. The one that I read reminded me of "motherhood and apple pie" and was full of erroneous claims (like "competitiveness equals jobs").
So, what are the issues for IA? And is it at all appropriate for a political site?
The starting point in any site planning is clarifying the desired/expected audience and their needs. Admittedly, political messages rarely contain clear, cohesive explanations. This is likely intentional. So is the site a way to "go on the record"? Or pacify voters? Even mislead voters, if the candidate expects to benefit from it?
The goals of IA are to clarify subject matter and speed access to the desired content. I suggest this is antithetical to a political site.
I was planning to launch into an explanation of choosing a good word as a category for positions (I liked that word the best of the ones I encountered). And then explain how to subdivide a position to a maximum of seven items, each of which can be further subdivided. And providing a way to refer to an item, such as by a letter (as opposed to the popular bullet). But what's the point? I now realize that IA is not appropriate for a political site.
Design by committee (11-21-2016)
My analysis of the political sites suggests that each was designed by a committee, each member representing an objective and a job. I think that each site's content, navigation, and structure is a compromise of several-to-many conflicting points of view. The UI designer clearly had their hands full balancing the conflicting demands.
Other sites may want to present a coherent, consistent presence to their readers. Design by committee will not accomplish that. Instead, the various requirements need to be worked out before UI design work begins. UI designers should exercise caution in their evaluation of the consistency and coherency of the requirements before beginning the actual design work. Diplomacy seems a useful skill in this case!
Is there a gradient of IA? Say, from effective to ineffective? (11-21-2016)
I think not. If the result is not effective, then it is NOT IA. It is simply UI design.
Use IA to tell history (7-16-2016)
You can use IA to improve the telling of history. I've elaborated on this in a two-page PDF file.
Do not choose the tool first (1-10-2016)
You've started a new project, one that will use a website to provide access to a document repository. The pressure to choose a tool/framework with which to implement the website is quickly becoming urgent. Should we use X or perhaps Y?
Often the choice boils down to a from-scratch site or one based on a framework like WordPress. WordPress offers an appealing choice—you can be up and running in no time. And the cost to go-live is expected to be less than that of a hand-coded, from-scratch site.
Pause here, take a deep breath. Remember the software and UI design process. Optimally you do the functional design, including the UI details, BEFORE the technical design.
The risks of choosing a framework before the functional design are many. They include (1) choosing a UI design feature (like navigation design or interface design) to fit what the framework allows—not what the application needs and (2) just plain not doing one or more of the UI design planes at all. And even if you do rigorously adhere to the UI design planes, it may be that in the end you will have spent more time than if you had not committed yourself to the framework at the outset.
Choosing the tool first is putting the cart before the horse.
So what is that I as an IA do? (11-14-2015)
- Identify the domain(s). A domain is a subject matter area shaped by its audience and their uses of the subject.
- Discover taxonomies and metadata for each domain.
- Design repositories of informational documents.
- Refine requirements and specify functionality for application scope. Document with data flow diagrams, process flow diagrams, use cases, data dictionary, wire frames (page layout and interaction specifications).
- Design navigational schemes for information. Document with navigation diagrams.
- Design interaction on multiple platforms including desktop, web, and mobile. Document with low-fidelity protopyes and/or wireframes.
Not all collections of things have a taxonomy (5-11-2015)
Consider my collection of electronic files about my family's genealogy. Whereas they have commonalities, there is no decomposition in kind. What they do have is metadata. Some of these metadata have sub-divisions.
Example of metadata for a family history collection:
- Primary surname (my grandparents).
i. Spousal surname.
- Document type: vital record (birth, marriage, death), military, census-directory, travel, photograph, legal (e.g., will), etc.
- Date of record/document.
- Source of document.
Taxonomy vs. metadata (5-10-2015)
Together these comprise a scheme of classifying things. For IA, those things are usually text documents.
Let's consider the following items that describe a particular text document:
(a) published 3-3-2013
(b) 53 pages
(c) geography of North America
(d) college text
All of these items are useful in a reader's decision to read or ignore the document. They are all independent of each other, that is, a publication date and the number of pages have nothing to do with each other.
The subject of the document is clearly (c), geography of North America. This is what fits into a taxonomy. The rest are metadata.
Depending on the collection of text documents, a geography of North America may fit near the top of the subject hierarchy, in the middle, or near the bottom.
Now let's consider a group of related subject matter taxonomies.
- Domain: Science.
- Hierarchy of taxa: science by type (physical and behavioral). Physical science subtypes could be life science, earth science, environmental science, engineering and technology, medicine and health. Behavioral science subtypes could be anthropology, archaeology, economics, philosophy, sociology. Earth science includes geography.
- Domain: World geography.
- Hierarchy of taxa: geography by continent, geography by region, geography by nation, geography by state, geography by county.
- Domain: Geography of North America.
- Hierarchy of taxa: geography by region, geography by nation, geography by state, geography by county.
- Domain: Geography of California.
- Hierarchy of taxa: Geography by region, geography by county, geography by park.
The example above illustrates a hierarchy of taxa that varies by domain. It also illustrates two kinds of subdivisions, one by type, the other by location.
DMOZ Building a Geographic Taxonomy is an interesting discussion of how to define a taxonomy of geography. You will see that it depends on the domain of the subject matter and the needs of the users. There is no one-size-fits-all for taxonomies.
The plant taxonomy that Linnaeus compiled is but one kind of taxonomy. In this case, each plant exists only at the lowest level in the hierarchy. Geography is different.
Undoubtedly the same kinds of rocks will exist in many places and those places will fit into different parts of a taxonomy. If the document collection features both locations and kinds of rocks, then location and kinds of rocks can be separate but equal dimensions of the taxonomy.
The subdivisions in a taxonomy represent a decomposition in kind. A taxon in level 3 represents a subdivision/decomposition of its parent taxon. That taxon exists independently of taxa with different parents.
Consider the following example posed by a friend. The domain is my collection of electronic files about my family's genealogy.
. . . Year
. . . . . . Events
. . . . . . Family
. . . . . . . . . Person
. . . . . . . . . Activity
. . . . . . . . . Siblings
. . . . . . . . . Spouse
. . . . . . . . . Beliefs
The indentation suggests hierarchy, but is not. Year is not a subset of location. Event is not a subset of year, etc. All those things, except the people things, are independent of each other in the sense that they exist independently of each other. Yes, time flies when you are on vacation as opposed to being in jail (year and location), but each exists independently of the other. Events by their nature have a time and place, but you cannot say that either is always superior; they are different dimensions.
This proposal is arbitrary. You could just as well order things differently. The ability to correlate two or more things, such as Cromwell and the archibishop is outside of taxonomy. This is where metadata can be very useful. In the following example, the metadata follow the colon:
event: date, location, people, cause, effect.
Software supply and demand (6-23-2014)
I first wrote about this in December 2010. Here is a shorter version:
We don't have a problem designing and building software. We've been doing that for over 50 years. What we do have is a demand problem—the demand exceeds the supply. That is not a software problem, it is a political problem. One which software managers have failed at for 50 years.
Universal Decimal Classification (UDC) (6-5-2014)
I just learned about the Universal Decimal Classification (UDC), first published in 1905 by Belgian bibliographers Paul Otlet and Henri La Fontaine and currently used extensively world-wide. They proposed a single classification system for the domain of all knowledge. Classes are organized in tables of common concepts (context-free concepts) and in tables of main knowledge subdivisions (subject classes). UDC numeric codes can describe any type of document or object to any desired level of detail. Whereas the UDC subject classes are hierarchical, more than one can be applied to the same object. Thus the UDC accommodates objects with more than one class in the same domain. This allows the useful classification of documents with a primary subject and several secondary subjects, each of which with its own class in the same taxonomy. This is being called a faceted classification system. The UDC is managed by the UDC Consortium, www.udcc.org.
A goodly number of consultants have been hyping "content" in the past few years, referring to whatever is presented on web pages. Content Management Systems (CMSs) emerged a little before 2000 as the need arose to enable lots of people of varied technical knowledge and expertise to add, edit, and publish content on large, complex websites. Now there are job titles like Content Manager and Content Producer. It's a new growth industry.
I recently read a book review bemoaning the loss of analytical news reporting. It made me remember the disappointment I had yesterday reading a New York Times "news" article. News is now a story, with the same attention to story telling employed by film makers. Consequently I am forced to read six paragraphs when the actual news could have been conveyed in one.
News as content?
It begins to look as if much content is much ado about nothing.
What I find missing in commercial websites is any presentation of who we are, why we are, and where we are. How we are affected by and respond to the world around us. Both the identity and the personality of the business organization is hidden.
Commercial websites are trying to sell products and services, no surprise. But how am I, the potential customer, expected to choose between the products of competing vendors? After verifying that both the functionality and usability of the product/service meets my needs, I want to know about the company that created it. All things being equal, I want to choose the vendor over the product.
What I miss are stories about the people who created the company and the product.
What can information architecture do to improve this situation? I have long believed that IA is more than just a content classification scheme — that it must also specify/prescribe subject matter, types of content, and writing styles. It must address the use and style of written introductions and transitions. It must design a whole that exceeds the parts.
Origins of graphical user interfaces (5-5-2014)
A user interface is the set of techniques and mechanisms that a person uses to interact with an object.
The earliest computer user interface was a command-line interface in which a user typed commands on a single line. Alternatives emerged in the 1960s with the oN-Line System (NLS) developed by SRI International which used a mouse-driven cursor (but no point-and-click), multiple windows, and context-sensitive help. A GUI with windows, icons, and menus was developed at Xerox Palo Alto Research Center (PARC) in the 1970s and was first produced in the 1973 Alto computer. The 1980s saw the MEX (short for Multiple EXposure) user interface of the IRIS 1000 Series made by Silicon Graphics, Inc. (SGI), the X Window System developed at MIT which became the standard for UNIX, the OSF/Motif GUI specification for X Window System applications on UNIX, and the Apple Lisa and Macintosh.
Open Software Foundation (OSF) was founded in 1988 to create an open standard for an implementation of the UNIX operating system in opposition to the work done by Sun Microsystems and AT&T. The original sponsoring members were Apollo Computer, Groupe Bull, Digital Equipment Corporation, Hewlett-Packard, IBM, Nixdorf Computer, and Siemens AG, sometimes called the "Gang of Seven." Later sponsor members included Philips and Hitachi with the broader general membership growing to more than a hundred companies. Motif, a graphical user interface Toolkit (aka Motif widgets) and specification for the X Window System (the Motif Style Guide) was designed to compete with the earlier Open Look GUI specification defined by Sun and AT&T in 1988. Motif was initially released about 1989 and adopted as a formal standard by the IEEE in 1994. In 1993 Sun and AT&T, facing growing competition from Microsoft, became OSF sponsor members.
In 1987 IBM, addressing the growing diversity of its computer products, began developing a conceptual framework for the design of graphical user interfaces that would rationalize and harmonize all user interfaces, from mainframes to microprocessors. They called it Common User Access or CUA. The third and current version is CUA 91, published in 1991. CUA 91 incorporated input from OSF/Motif and Microsoft, was the basis of OS/2 Presentation Manager, and was adopted by Microsoft in Windows 95.
CUA declared what a keyboard looked like, what a window looked like, the appearance and operability of the various window controls, and what all the common keybindings were. It also prescribed principles of GUI design. Its main objective was to render applications invisible so that users could perform the same actions in the same way on any object, regardless if it was a purchase order or memo or spreadsheet.
Today (2014) CUA remains the foundation of most application user interfaces, but few designers and programmers are aware of it. Its ideals are still valuable contributors to software design.
The IBM manuals on CUA can be found at publib.boulder.ibm.com/cgi-bin/bookmgr/Shelves/ceesl002
Evolving user experience and IA (4-12-2014)
Scientists know that the act of measuring something can affect it. Similarly a user's experience with a tool gained over time changes their ideas about what they want to do with the tool and how. They often become sophisticated users.
You can see that IA should evolve as the interests and needs of a site's users evolve.
Question: What does an information architect do? (3-2-2013)
Answer: An information architect uses organization, navigation, and typography to devise an efficient and pleasurable access system to information — documents, images, and/or sounds.
Taxonomy and metadata revisited (7-18-2012)
This discussion is a separate document due to its length. It introduces the origins of taxonomy before elaborating on taxonomy and metadata for document libraries. It develops the ideas to far more detail than previously on this page. It explains one-dimensional taxonomies and multi-dimensional taxonomies.
Quality of design (1-12-2011)
Too many IA and UI job descriptions spend a lot of words on the need for wireframes and user interaction specifications. Certainly these are useful and even necessary. They are one means of documenting a design. They are the basis for implementing the design by programmers.
They are NOT the design.
The word "design" itself rarely appears in job descriptions. It is invisible. Perhaps because real design and creativity are mysterious artistic forces, and forces vaguely discomforting to hiring managers and recruiters.
While it is pointless to talk about the "process" of design, it is highly desirable to talk about the quality of design.
A clever business systems analyst can include in the Statement of Requirements many characterstics of and criteria for design quality. Thus design quality can be specified in advance. It remains to the project managers to make sure the design is thoroughly evaluated for adherence to the requirements—before it is accepted.
Information architecture and narrative (1-12-2011)
Let's not forget that IA is in service to the website and its role as a business tool of its owner.
It's my theory that narrative can play a powerfully effective role in establishing cohesiveness, unity, and community. It can convey attitudes and values. It can set up an internal dialog with site visitors that engages them. It can facilitate understanding.
The books you prefer to read, and enjoy reading, have a strong, consistent narrative. Yes, there's lots of content—that's why there are chapters. The narrative can both sustain the division of content into chapters and persist through them. It keeps you engaged and informed.
It would be a mistake to see IA as a tool for organizing existing content pages and slapping on a navigation menu and a few landing pages. You won't create a narrative that way.
Narrative has to begin in the vision for the website and extend into the IA, writing, and layout.
Growing demand for software (12-24-2010)
We need to imagine information site designs that continue to attract users while those users become more sophisticated consumers. Perhaps we can build in some artificial intelligence that can analyze each user's needs and preferences and adjust the site design accordingly. Certainly we need to avoid a design that keeps us buried in the supply-and-demand quagmire. This is discussed in detail in a separate essay.
Title is a special case (12-15-2010)
The title of a document/unit of content is a special case. It is the intersection of content, use, metadata, and taxonomy. I think title plays a strategic role in the overall classification scheme and navigation scheme. The title should avoid repeating both taxa and metadata. Instead it should complement the classification, be meaningful, and read as a complete, well-written phrase.
Identifying metadata (12-15-2010)
Some metadata applies to the document, e.g., author, date created, file type, version.
Some metadata applies to the use of the document, e.g., document type, audience, keywords.
So "finding" metadata in a population of "content" involves (1) an analysis of the documents to determine their intrinisic properties and (2) an analysis of the audience and their needs for the site (content as a whole).
Is taxonomy "scope" or "structure"? (12-5-2010)
This question came to me as I was revising my portfolio. The categories belong to Jesse James Garrett. His "scope" is functionality and content, "structure" is high-level design. I think that taxonomy, because it is a property of content, belongs to "scope."
So the answer is—scope.
This idea differs from the polar bear book. Its chapter 5, Organization Systems, jumps into the subject of organizing a website. The subsection "Organization Structures, The Hierarchy: A Top-Down Approach" begins with "The foundation of almost all good information architectures is a well-designed hierarchy or taxonomy." Now that I see it (NTISI) this claim is incorrect: taxonomies are not "designed," but they are discovered by careful analysis of the content. On page 222 the book introduces the notion of faceted classification or multiple taxonomies. Frankly I think this notion is wrong, there is one taxonomy and then there is metadata. 5-15-2014: see also Taxonomy and Metadata Revisited written 7-18-2012.
What is taxonomy? Taxonomy reflects the relationships the various units of content have with each other. And taxonomy does not have to be hierarchical, it can be relational.
How might the essential content classification scheme (taxonomy and metadata) differ from the site organization and navigation? The correlation of units of content with taxa is not necessarily the same as the presentation on the site by an organization and navigation scheme. And the actual labels used in the site are specified by controlled vocabulary, which again may differ from the taxonomy. One example of differences between a taxonomy and a website organization is the high-level division of content by audience—user. In this case user is really a metadata, but for purposes of maximizing access to site content, user is treated as the top level in the organization design.
The scope of information architecture (11-30-2010)
[I am not completely sure of the following idea. Please comment!]
Practitioners of IA seem content to limit their work to an organization scheme for a site's content and a navigation design based on it. They seem equally content to leave the "visual design" and copywriting to others.
Richard Saul Wurman, in his book Information Architects, presented complete designs as examples of IA. Thus I tend to think of IA as the result, not an interim design artifact. Just as a building architect is judged on the finished building, so should website designers, whatever their title, be judged on the finished website. You can say "I like the facade but not the massing" or "I like the shape but not the way I feel when I walk in the front door." But—it is all one, for better or worse.
Why would you design only part of the site and leave the rest to others? What if they misrepresent your ideas? How can you claim credit for the site when your work focused only on one element of the design? If the completed design fails, is it your fault?
Turning over rocks (11-15-2010)
A metaphor for information search.
Scenario 1: Imagine a rockfield with rocks about the size of 2 russet potatoes separated from each other by about 18 inches. The solution to a simple problem is written underneath each rock. You want to find the solution to a particular problem. You have two choices: (1) walk away or (2) turn over each rock in sequence until you, hopefully, find the solution.
Scenario 2: Same rockfield as scenario 1 with the addition that a text label is inscribed on the top of each rock with a clue to the nature of the problem whose solution lies underneath. You can now walk through the rockfield and turn over only those rocks whose label seems to match your needs.
Scenario 3: Same rockfield as scenario 2 with the addition that the rocks are colored to reflect some classification scheme. You do not know the scheme, but after reading a few rocks of each color you can theorize that one or two colors are your best choices. After that you restrict your search through the rockfield to only rocks of those colors, i.e., you still walk through the rockfield, but only leaning over to read the rocks with the "best" colors.
Scenario 4: Same rockfield as scenario 3 with the addition that the rocks of the same color are grouped together. Now once you decide which colors are your best choices, you can restrict your search to rocks that are near each other.
Scenario 5: Same rockfield as scenario 4 with the addition of a sign erected near the front of the rockfield with an alphabetic list of the rock labels grouped by color, the entry for each rock includes its geographical coordinates. You can search the sign for the most likely rocks by color and label, then walk to their location where you turn them over. You may have to return to the sign for the location of another candidate rock if you do not remember it.
Scenario 6: Same rockfield as scenario 5 with the addition that each entry includes a synopsis of its solution. You can search the sign for the most likely rocks by color and label, then read each synopsis. When one or more seems to fit your needs, you walk over to the rock and turn it over. As with scenario 5, you may have to return to the sign for the location of another candidate rock if you do not remember it.
Scenario 7: Same rockfield as scenario 6 with the addition that each rock includes the coordinates of another rock having a similar classification. You can search the sign for the most likely rocks by color and label, then read each synopsis. When one or more seems to fit your needs, you walk over to the rock and turn it over. Now you do not have to return to the sign for the location of another candidate rock if you do not remember it, as the current rock has the location of a related rock.
The quality of information architecture in each of these scenarios ranges from non-existent (scenario 1) to very high (scenario 7).
Designing distraction—NOT (6-1-2010)
The article "Chaos Theory" by Nicholas Carr in the June 2010 issue of Wired magazine offers an explanation of the disadvantages of distraction in visual design.
Use hypertext links in linear text with care: "Research continues to show that people who read linear text comprehend more, remember more, and learn more than those who read text peppered with links." Other "research suggests that links surrounded by images, videos, and advertisements could even be worse."
Cognitive overload interferes with the transfer of information from our working memory to our long-term memory. "We can't translate the new material into conceptual knowledge. Our ability to learn suffers, and our understanding remains weak." Frequent switching from one task to another and the automatic interruptions like "You've got mail" contribute to cognitive overload.
One school of visual design holds to the "more is more" philosophy. Their products overflow with visual stimulation. Now we see that while they can attract our attention, their design works against our real understanding of the subject matter.
Information architecture, whose goal is to promote the reader's conceptual understanding, is best served by avoiding cognitively distracting design elements.
Defining IA for friends and family (4-21-2010)
Still looking for a succinct definition that non-experts can understand. Try this:
IA is (1) the organization of content into a form that is relevant to its audience and use and
(2) the design of the navigation to the content that reflects its organization.
Again, what is IA? (3-30-2010)
IA is about distributing content among web pages so that readers can: (1) see what content is available and (2) access the desired content efficiently.
The solution is going to be different for a website with four pages of content than for a website with 400 pages of content.
I adhere to the following principles:
- Every page has either an overview of the entire site content or a navigational link to a page with such an overview.
- A navigation network can reflect one or more of the following: classification scheme, relationships, index, special collections.
- Accept the target audience's attention span, ability to "see" the site at a glance, manual dexterity, and tolerance (or intolerance) for busyness.
- Classification system has a consistent level of detail.
- Accessing content is more than navigation, it also includes reading and understanding.
- Writing and words matter: Vocabulary is consistent without using several different words for the same thing. Employ wordiness only when it furthers the site goals.
The complexity and style of the site content overview is a function of the volume of content. It is common practice these days to employ a hierarchical menu as a site overview (this is modeled after the hierarchical menu system adopted by desktop applications beginning about 1989). But this is not the only method and, as it has become ubiquitous, it has also become a little boring, so be sure to reach for a method that is both effective and engaging.
Resisting entropy (3-2-2010)
Entropy, the Second Law of Thermodynamics, was officially described by Rudolf Clausius in the early 1850s. Entropy is a state-function that explains why thermodynamic systems tend to progress in the direction of increasing entropy. This concept is easily generalized to other subjects including software. Entropy in software is the inevitable tendency towards disorder.
For software, entropy applies to the code and the functionality. For websites and help files, it also applies to content, organization, and navigation. In each case as entropy increases integrity decreases.
Entropy gets its foot in the door at the time of the first system change and grows with each succeeding change. (Changes are rarely approached with the same thoroughness of the original software design and development.) When content is involved, entropy can appear when content is not added or changed—when it should be.
Entropy can transform a high quality application into one that warrants the disparagement heaped upon it by its users. I doubt entropy can be prevented, but it can be minimized. I offer the following tips for minimizing entropy:
- Write an Administration Guide for the ongoing maintenance of the application and content.
- List the design principles which shaped the application and which adherence to will minimize disorder introduced by changes.
- When content is important, write an Editors Guide and a Style Guide.
- List the events and/or conditions which you expect will warrant application and/or content changes.
- Identify the people and/or roles who can be assigned the responsibility to recognize when content changes are needed and to see that the changes are made properly. This is unlikely to be the department secretary!
Foundational principles (3-2-2010)
Design is not arbitrary.
Quality software doesn't get that way accidentally. It is designed by application architects based on tried-and-true principles.
Similarly a high quality information architecture should be based on principles. These principles should be deliberately chosen, analyzed, and applied.
I want to bring Common User Access (CUA) to your attention. CUA is a conceptual framework for the design of graphical user interfaces; it was developed by IBM, beginning in 1989, and adopted by Microsoft Windows, OS/2, and Mac OS. CUA relies on design principles organized into four groups: Principles to Put the User in Control, Principles to Reduce the User's Memory Load, Principles that Support Consistency, and Principles that Create Aesthetic Appeal. CUA further advocates the organization and meaning of the application be a natural extension of the user's conceptual model. I have summarized CUA in an 11-page document which I use as a reference. You are welcome to it.
The design principles of CUA are just as valid today as when they were formulated in 1989. They are quite general and need to be extended to more specificity on each project: rules can be specified for the particular application of the principles. The information architect/UI designer can consider each CUA principle and consider how best to apply it to the project at hand. They are also free to formulate other principles.
As an example, here are window operation principles I adopted for a recent project:
- User doesn't have to learn the rules the hard way.
- Limit actions to action controls. (CUA differentiates between types of controls).
- When a window is opened, the cursor is positioned to enable the primary use.
- Never assume.
- User cannot leave a window with unsaved changes, either by closing it or switching to another open window.
- Drill-down is done by double-clicking the lower level "record" in a list.
- Drill-down opens a window that presents the lower-level "record" only if that window is not already open. If it is open, the selected record is made current. The window on which drill-down is done remains open.
Tagging and IA (2-26-2010)
Tagging is not IA. It is keywords assigned by authors and, sometimes, by readers. It has meaning only to the person who assigned it. Any relevance to tags on other content is accidental. It is no substitute for a classification scheme. At best it can invite involvement by readers. To take this involvement further, let other readers rate the tags and, by extension, the taggers.
Is user experience (UX) relevant to user interfaces? (2-26-2010)
Think of a bar with a band, a dance floor, and a disco ball. These elements set a baseline for UX. Think of too few, surly bartenders. Minus UX. Think of hot-looking guys who dance well. Plus UX.
So what comprises your UX? Mood, atmosphere, possibilities, reliable service and products. You go to this bar to meet guys, dance, and have fun.
Think of a dress shop with attractive decor and good hours. Now think of inattentive salespeople. Minus UX. Think of gorgeous clothes priced affordably. Plus UX.
What comprises your UX? Mood, atmosphere, possibilities, reliable service and products, affordability. You go to this dress shop to get perfect clothes for most occasions. You also go for fashion inspiration to help you stay up to date, or even a little ahead of the trends.
Would a string quartet at the dress shop affect your UX? Probably not, because it is irrelevant. Unless it is a dress shop cum entertainment venue, still unlikely.
Think of an ATM machine in your local grocery store. Now think of it out of service when you want to use it. Minus UX. Think of it responding way too slowly to your commands. Minus UX. Think of it with an unreadable display screen. Minus UX. Think of withdrawing cash easily and quickly with an accurate legible receipt. Plus UX.
What comprises your UX? Reliable, consistent, accurate, and efficient service. You go to this ATM to get cash in a hurry.
In these examples UX is about getting what you expect. In the first two examples it's also about getting something more, something unexpected perhaps, but relevant. The first example is primarily experiential. The third example is strictly a business transaction. The second is a little of both.
Designing for a satisfying UX requires an understanding of the business, the users, and the nature of the uses. What is relevant, what is not.
How does this differ from straight user interface design? Why should it? When would a serious design effort ignore the user? To me the only point to all the hype about UX is its reminder to not forget the users. Good design never did.
Information architecture and typography (2-26-2010)
IA seeks to (1) establish a content classification scheme based on taxonomy and metadata and (2) design a visual presentation of that content with an effective navigation scheme. The visual presentation must include a typographic language composed of a layout grid, typefaces, and color.
Is it possible to do IA without specifying the typography? I don't think so.
How does the information architect document their design?
- The classification scheme: Specified completely with hierarchy, relationships, and phrasing (vocabulary). The choice of words is an important element, too important to ignore. Identify types of navigation in text and diagrams.
- The visual presentation: Wireframes can be used to illustrate the layout and where the various elements, especially navigation, are placed. Mockups can specify the layout grid, typefaces, and color.
- The interaction: Swimlanes and/or flow charts can be used to describe each use case. A prototype can be used to illustrate the actual operation.
- Site file organization: In a hierarchical list name the directories and specify their contents.
- Content management: Text and diagrams can outline roles and activities. Swimlanes and/or flow charts can illustrate procedural sequences and dependencies.
Yes, but what is information architecture? (1-26-2010)
Practitioners usually have an "elevator speech" on hand, but these are often not very understandable to non-practitioners. Dan Klyn, Lecturer at University of Michigan School of Information, created a 4 minute video to explain IA that is pretty good.
For Dan, an IA project has
- a subject
- an audience
- a technology for implementation
The role of IA is to present the subject, within its domain, so that it is clear to its intended audience, regardless of its complexity. The clarity of presentation is facilitated by a deep understanding of the nature and meaning of the subject, a classification system, and a design of user movement through and interaction with the subject that achieves utility and delight.
Catalog vs. index (8-6-2009)
These terms do not refer to the same things. They are often misused in common speech. Both have value for consumers of information widely available on the internet. They are described in a separate essay, their differences elaborated, and suggestions for future use are described.
What is a reference library? (6-2-2009)
An application focused on providing access to a library of documents about one or more related subjects and having a known audience. Most of my reference libraries were created for software developers, the subjects included the software application design as well as development practices, processes, and tools. "Tools" can be a large group, embracing programming languages, source control, external APIs, document templates, etc.
I have had difficulty in my resumes labeling these libraries so non-technical readers (the recruiters and all-too-often the hiring managers) would understand. I started out calling them "document libraries," which seemed limited. "Information library," which meant everything to me, drew a complete blank with others. My latest effort is "reference library." What do you think?
Components of a reference library (6-17-2009, revised 2-26-2010)
1. Document repository. A place to put those documents that comprise the library. A primitive form still in use is a file system on a data storage device—hard drive or file server. It's where you put your document files.
2. Catalog. A place to store references to the documents so that the user can understand the contents of a document sufficiently to decide to read it. A catalog entry identifies the location of the document, for electronic documents this is usually its complete filename.
3. Classification scheme. Each document is classified according to a predefined scheme. It is the document's classification that is held in its catalog entry. It is the document's classification that enables the user to choose to read the document or pass.
4. A user interface that provides access to the documents via the catalog.
5. A content management strategy and tools to enable the addition of content so that it adheres to the classification scheme, fits in the user interface, and is reflected in the catalog.
Functionality of a reference library (6-2-2009)
The two most obvious functions are:
• storing documents
• finding documents
Since finding documents is the driving force for a library, storing documents must include:
• classifying documents
• adding a catalog entry for each document
In order for users to quickly find the right document ("the right document at the right time"), the catalog must have a user interface that makes the classification scheme obvious and lets users find documents based on their classification. This is often called a
• navigation scheme
• maintain the classification scheme
• audit document classification
Lastly a library needs alternate ways for users to find documents, for use when the catalog is just not enough:
• full text search (of title and contents)
• bookmark (to save the identity of a document once the reader has found it, to speed finding it a second time)
What is a classification scheme (6-2-2009)
A classification scheme includes a taxonomy and metadata. At least for the purposes of this discussion; it could be argued that a classification scheme is only a taxonomy, but I feel that a taxonomy is incomplete without the additional information provided by metadata.
Taxonomy: Taxonomy (from the Greek words for law, nomos, and order, taxis) may refer to either a classification of things, or the principles underlying the classification. Almost anything—animate objects, inanimate objects, places, and events—may be classified according to some taxonomic scheme. Taxonomies are frequently hierarchical in structure, but can also be relational.
Taxonomy is defined as:
1) The classification of objects in an ordered system that indicates natural relationships.
2) The science, laws, or principles of classification; systematics.
3) Division into ordered groups or categories.
A taxonomy that we are familiar with is the hierarchical organization of organisms which includes species, genus, family, order, etc. More recently taxonomy is being used to classify information.
Taxon is a single taxonomic category or group. Taxa is the plural form.
Metadata: Data about data. Properties of items in a taxonomy. Examples: author, document type, revision date.
Characteristics of metadata, taxonomy, indexing, and a relational model (06-2-2009)
1) Taxonomy is a multi-level hierarchy of property-value pairs, e.g., genus = Felis, species = Felis sylvestris.
2) Metadata is a collection of property-value pairs, e.g., document type = agenda.
3) An index is composed of index entries that can be multi-level and can have cross-references to each other, e.g., Ferrari. see cars.
4) A relational model is characterized by an explicit link between two objects where the link itself has an identity and a grammar, e.g., A owns B, B lives in C.
A classification scheme applies to a domain. A domain is composed of subjects and users, and implies a particular purpose.
1) A taxonomy typically has more than one level, i.e., it has more than one taxon.
2) A multi-level taxonomy is hierarchical.
3) The values of a single taxon can be prescribed.
4) The values of a lower level taxon can be specific to its parent taxon, i.e., only some values of a lower level taxon may be applicable to a particular parent. Said a third way, for a given pair of taxon, where the second exists at a level immediately below the first, there may be a prescribed set of values. In the following example of valid values, there are two taxa, A and B, where A is the parent and B is the child:
A1 - B1
A1 - B2
A1 - B5
A2 - B3
A2 - B4
A2 - B5
5) The contents of a domain are listed in a catalog with their classification data. The catalog and the classification of each document must always be in sync.
Design challenges (6-2-2009)
1) How to establish a domain?
2) How to establish a set of valid values available for classifying a single document within a given domain?
3) How to administer the scheme, i.e., provide oversight and control?
4) Where to store classification data for a document?