π Lecture 42 β Communicating Users
π Overview: This lecture explores how to design interfaces that eliminate or minimize error messages, notifiers, and confirmatory dialog boxes. It argues that these elements are often abused and can be replaced with smarter, more flexible software that prevents errors from occurring in the first place, fostering a more positive and efficient user experience.
ποΈ Topics Covered
This lecture begins by examining why error messages are so prevalent and problematic, arguing they are a sign of inflexible software design. It then proposes methods for eliminating errors, such as making errors impossible through bounded inputs and making programs smarter. Next, it discusses the importance of positive feedback and provides guidelines for improving unavoidable error messages as a last resort. The lecture also covers notifiers and confirmations, explaining why they often stop proceedings unnecessarily and how alerts announce the obvious while confirmations pass the buck. Finally, it introduces rich modeless feedback as the superior alternative to dialog boxes.
π Lecture Summary
42.1 Eliminating Errors
Bulletin dialog boxes are frequently used for error messages, notifiers, and confirmations, but they are often overused and represent poor design. The lecture posits that with proper design, these dialogs can be largely eliminated. Users do not want error messages; they want to avoid the consequences of errors. The tradition of error messages stems from early computing where machines were expensive and operators were scientists tolerant of machine-centric communication.
Error messages are problematic because they stop the proceedings with a modal dialog, often informing the user of the program's inflexibility rather than a user error. Users hate error messages because being told they are wrong by a machine is humiliating. The assumption that users need to be told when they are wrong is false, as programs can often make reasonable substitutions for invalid input. Most "errors" occur when the program gets confused by input that is in an unfamiliar sequence, not when the user makes a substantive mistake.
π Definition β Blocking modal bulletins: Error messages that must stop the proceedings with a modal dialog box.
π Example: When a user enters an invoice with an invalid customer number, most programs stop and demand a valid number. A better approach would be for the program to accept the transaction and remind the user to fix it before the end of the session. This mirrors how human assistants work; they don't stop everything for missing information.
π‘ Why this matters: Error messages are often a failure of program design. The program should be flexible enough to handle user input in various sequences, rather than forcing the user to adhere to the program's strict rules. The program should assume the user is correct because the human is more important than the code.
The best way to eliminate error messages is to make errors impossible by using bounded gizmos like lists and drop-down menus instead of free-form text entry. Another way is to make the program smart enough to deduce information without asking, such as getting the date from the internal clock or remembering recently used files. These solutions require more work from programmers, but the programmer's job is to satisfy the user, not the other way around.
π Example: Instead of having a user type a state code, present them with a list of valid state codes. Instead of asking for a file name, have the program remember which files were accessed in the past and allow selection from that list.
42.2 Positive feedback
Software is often hard to learn because it rarely gives positive feedback. People learn better from and are motivated by positive feedback. They need to be rewarded or acknowledged for success, not just slapped on the wrist for failure. While negative feedback is effective for preventing people from doing things they shouldn't (like speeding), positive feedback is best for helping them do what they want to do.
To be given negative feedback by software is an insult, as the computer has no genuine experience or merit. The lecture concludes that there is nothing inside a computer that justifies humiliating or degrading a user. Negative feedback is used only out of habit.
Improving Error Messages: The Last Resort
If an error message is truly unavoidable, it should be used as a last resort and must conform to these requirements:
- Be polite: It must never even hint that the user caused the problem. The customer is always right. The program should protect the user even when they take inappropriate action.
- Be illuminating: It must explain the scope of the problem, alternatives, the program's default action, and what information was lost.
- Be helpful: It should offer to implement at least one solution directly on the dialog box, such as offering to defer a print job if a printer is missing.
π Example: A well-formed error message for a missing printer should be polite, explain that no printer was found, and offer buttons for "Try Again," "Select Another Printer," or "Defer Printing." It should not imply the user made a mistake.
42.3 Notifying and Confirming
This section discusses alerts (notifiers) and confirmations. Like error dialogs, they stop the proceedings with idiocy but do not report malfunctions. An alert notifies the user of a program's action, while a confirmation gives the user the authority to override an action. Both should be eliminated in favor of more useful idioms.
42.4 Alerts and Confirmations
Alerts occur when a program feels uncomfortable with an action it has taken, so it informs the user. This violates the axiom: "A dialog box is another room; you should have a good reason to go." A program should either have the confidence of its convictions or should not take action without user guidance. An alert is unnecessary when the user has explicitly requested the action.
Confirmations are dialog boxes that ask the user for approval when the program feels unsure about its actions. They pass the buck from the program to the user. The proper solution is to make the action easily reversible with a robust Undo facility.
π Definition β Alert: A dialog that notifies the user of the program's action. π Definition β Confirmation: A dialog that gives the user the authority to override an action, reflecting the program's lack of confidence.
π Example: A dialog that appears after adding an entry to a phone book, saying "The entry was added," is a bad alert. The user just told the program to do it, so the announcement is obvious and annoying.
π Example: A confirmation dialog asking "Are you sure you want to delete this style?" passes the responsibility for a potentially risky action back to the user. A better solution is to provide clear visual feedback about which styles are in use and a powerful Undo feature.
π‘ Why this matters: Confirmations suffer from the "boy who cried wolf" effect. When they appear routinely, users dismiss them without reading, rendering them useless when a truly dangerous situation arises. For confirmations to work, they should only appear when the user is almost certain to click "No" or "Cancel," which makes them pointless.
42.5 ELIMINATING CONFIRMATIONS
The best way to eliminate confirmation dialog boxes is to Do, don't ask. The program should act with confidence. To compensate for potential mistakes, every action must be reversible. Instead of asking in advance with a confirmation, let the user issue a Stop-and-Undo command on those rare occasions when the program acted out of turn.
π Example: The Recycle Bin in Windows is a good implementation of an undo facility for deleted files. Files are moved to a suspense directory instead of being immediately deleted, so there's no reason for a confirmation box for every deletion.
For truly irreversible actions, provide modeless warnings built into the interface. For example, instead of a dialog box saying a document is larger than the printable area, show permanent visual guides on the page indicating the printable region.
42.6 Replacing Dialogs: Rich Modeless Feedback
Modern computers have high-resolution displays and high-quality audio, but most programs still use the blunt instrument of dialogs to communicate. What users need is constant feedback through a different channel. Rich modeless feedback is information provided in the main displays that doesn't stop the flow of the program, effectively eliminating pesky dialogs.
42.7 Rich visual modeless feedback
Rich visual modeless feedback (RVMF) is a key type of modeless feedback. It is:
- Rich: Giving in-depth information about the status or attributes of a process or object.
- Visual: Using pixels on the screen, often dynamically.
- Modeless: Always displayed, requiring no special action from the user to see it.
π Example: In Windows 2000 or XP, clicking an object in a file manager window automatically displays details (title, size, author, modification date) and a thumbnail. If the object is a disk, it shows a pie chart of used space. This eliminates the need for a separate properties dialog to display this information.
π Lecture 43 β Information Retrieval
π Overview: This lecture explores how to communicate effectively with users through audible feedback, identity systems, and help mechanisms. It also examines the critical distinction between storage and retrieval systems in digital interfaces. Understanding these concepts is essential for designing interfaces that provide clear, modeless feedback and enable users to find information efficiently without relying solely on memory.
ποΈ Topics Covered
The lecture begins by examining positive vs. negative audible feedback and how silence can be a powerful indicator of success or failure. It then explores various channels for communicating with users, including program identity through names, icons, about boxes, splash screens, online help, wizards, and intelligent agents. Finally, it analyzes storage and retrieval systems in both physical and digital worlds, introducing positional, identity, and attribute-based retrieval methods.
π Lecture Summary
43.1 Audible feedback
In data-entry environments, clerks need both auditory and visual feedback when entering data, as they often examine source documents and type by touch rather than looking at the screen. The proposed auditory indicator for a problem is silence, not a beep.
Negative audible feedback announces user failure. Users are offended by computer sounds because computers have always accompanied error messages with alarming noises, and those noises have been loud, monotonous, and unpleasant. Error message boxes with beeps are a public announcement of the user's failure. It is the negative aspect of the feedback that presents problems, not the audible aspect. Negative feedback is issued at a time when a problem is discovered, naturally taking on alarm characteristics β alarm sounds are designed to be loud, discordant, and disturbing. People choose no noise versus noise for negative feedback, and no noise versus soft, pleasant noises for positive feedback.
π‘ Why this matters: The negative feedback paradigm teaches users that success is greeted with silence, but humans like to know when they are doing well β which is what positive audible feedback provides.
Positive audible feedback occurs when success with tools yields a sound. Almost every object outside software offers sound to indicate success: closing a door produces a click when latched, silence tells us it's not secure; conversation partners say "Yes" or "Uh-huh" to register what was said. Our keyboards provide positive audible feedback through clicks when keys are pressed. The true value of positive audible feedback is that its absence becomes an extremely effective problem indicator β the software doesn't have to insult the user. Software should give constant, small, audible cues like keyboards, issuing sounds when user actions are correct, and remaining silent when input is not understood.
π Definition β Negative audible feedback: Emitting noise when something bad happens, usually through loud, shrill, unpleasant beeps accompanying error messages. π Definition β Positive audible feedback: Sound produced when success with tools occurs, indicating correct action has been taken. π Principle: Success = sound, Failure = silence β more effective and user-friendly than the opposite model π Example: When a user drags an icon, the computer issues a low-volume sliding sound; when dropped on a meaningful target, a cheerful "plonk" is heard; if the drop was not meaningful, there is silence β immediately informing the user without embarrassment.
43.2 Other Communication with Users
This section discusses communicating with users in ways that are helpful to them while also asserting brand and identity. The modern desktop screen is crowded with many programs running, and each must assert its identity through several conventions.
Your program's name is displayed in the title bar of the main window. The title string is a single text value usually owned by the main window. Since Windows 95, the title string plays a greater role β it is displayed on the program's launch button on the taskbar. As more buttons are added, they shrink and title strings are truncated. Adding company names to program names can cause truncation to the company name only, making adjacent buttons identical. Many programs also use the title string to display the active document name. A better approach would be adding a separate title string for launch buttons, allowing tailored shorter strings.
Your program's icon is the second biggest component of identity. Windows uses standard 32x32 pixel icons and miniature 16x16 pixel icons. Mac OS X icons can be up to 128x128 pixels. The 32x32 size is used on the desktop, while 16x16 is used on title bars, taskbars, and Explorer. Icons must be readily identifiable from a distance, especially the miniature version. Icon design is a specialized craft β Susan Kare's original Macintosh icons set the industry standard.
Ancillary application windows are windows not part of core functionality but provided by convention. They are available on request or offered once, forming channels of communication that help users and communicate brand.
About boxes identify the program to the user and serve as credit screens. On Macintosh, they're summoned from the Apple pop-up menu; on Windows, from the bottom of the Help menu. Microsoft's approach treats the About box as an Identity box β displaying name, version number, copyrights, user identification β useful for customer support but not for users. A true About box should tell the user the program's scope, what it can and cannot do, and provide a one-lesson introduction for new users. It should also list the product team's names, as credits are important motivation for programmers.
π Definition β Identity box: A dialog that identifies the program by name, version number, copyrights, and user information, used mainly for customer support purposes rather than helping users understand the program.
Splash screens are displayed when a program loads, giving users something to view while the program prepares. They should disappear automatically after a few seconds, and immediately if the user presses any key or clicks. Splash screens create a good first impression, establish visual brand, and can direct first-time users to training resources. For shareware, splash screens can inform users of terms of use β but should not guilt-trip customers.
Online help is a reference tool for perpetual intermediates β users already successfully using the product who want to expand horizons. A complex program needs a reference document, either printed or online. The key to effective help is the quality of the index, which must have as many synonyms as possible. The list of index entries is more important than the entry text itself. Missing entries are worse than poorly written entries. Help systems should also offer a shortcuts option showing all tools and keyboard commands, and overviews explaining scope, effect, power, upside, and downside of features.
π Definition β Perpetual intermediates: Users who are no longer beginners but not experts, wanting to expand their horizons with a product's features.
Modeless and interactive help like ToolTips are incredibly effective. Standard help systems in separate programs cover up the main program and cannot point to objects on screen. Apple's innovative help system directs users step-by-step by highlighting menus and buttons in sequence.
Wizards are an idiom that attempts to guarantee success by stepping users through a series of dialog boxes. Programmers like them because they treat users like peripheral devices. Wizards use interrogation tactics rather than informed conversations. Users learn to just click "Next" without critically analyzing. Wizards are appropriate only for very rarely used actions like installation. A better approach is a simple automatic function that asks no questions and lets users modify results using standard tools later. Wizards give programmers license to put raw implementation model interfaces on complex features.
π Definition β Wizard: A step-by-step dialog system that guides users through complex procedures, often using interrogation tactics rather than providing true choices.
"Intelligent" agents like Clippy are problematic because animated anthropomorphism raises user expectations of the agent's intelligence. If the agent cannot deliver, users become furious. These constructs become cloying and distracting, interfering with users trying to accomplish tasks.
43.3 Improving Data Retrieval
In the physical world, storing and retrieving are linked β putting an item on a shelf gives us the means to find it. In the digital world, only faulty thinking links these concepts. Computers enable sophisticated retrieval techniques if we break traditional thinking patterns.
Storage and Retrieval Systems A storage system is a method for safekeeping goods in a repository β a physical system of container and tools. A retrieval system is a method for finding goods in a repository β a logical system allowing location according to abstract values like name, position, or contents.
Storage and Retrieval in the Physical World The "everything in its place" model works by location β knowing where something is and going to fetch it. This is a coupled storage and retrieval system, where the storage system equals the retrieval system. This works at home but is limited by human memory at larger scales.
For large collections like the Library of Congress, we use indexed retrieval with the Dewey Decimal system β giving each book a unique number based on subject and title, shelving in numerical order. Three index cards (author, subject, title) are created and filed alphabetically. The serial number bridges storage and retrieval systems.
Storage and Retrieval in the Digital World Ironically, where easily implementing dynamic retrieval mechanisms is possible, we often don't implement any retrieval system. Most computer systems have no retrieval system other than the storage system β users must know a file's name and place to find it. This puts 100% of the retrieval burden on user memory while the CPU idles.
π Definition β Coupled storage and retrieval system: A system where the storage method (location) is identical to the retrieval method β you find something by remembering where it is stored.
Retrieval Methods There are three fundamental ways to find a document on a computer:
- Positional retrieval β remembering where you left the file in the file structure
- Identity retrieval β remembering its identifying name
- Associative or attribute-based retrieval β searching based on inherent qualities of the document itself (e.g., red cover, discussing light rail, containing photographs)
Positional and identity retrieval also function as storage systems. Associative retrieval is the only method that is not also a storage system. When retrieval is based solely on storage methods, users must know what information they want and where it is stored.
π Definition β Attribute-based retrieval: A method of finding documents based on inherent qualities of the document itself, such as content, type, creation program, or usage patterns.
An attribute-based retrieval system would enable finding documents by their contents or attributes. The system would automatically know where all documents are, allowing searches without specifying directories. It would enable browsing by synonyms or related topics and assigning multiple attributes to documents. The system could automatically learn attributes like:
- The program that created or last opened the document
- Document type
- File size
- Time since last use
- Duration of last edit session
- Amount of information added/deleted
- Whether the document has been edited by multiple programs
- Whether it contains embedded objects
- Printing and faxing history
- E-mail recipients
The Enfish Corporation product provides this functionality β dynamically and invisibly creating an index of information across the computer system, LAN, and Web.
π Example: A user has proposal letters for different clients. The storage system requires each letter to be in one folder. An attribute-based system would allow finding all proposal letters (by type/function) while each letter retains its association with its particular client β impossible in a purely location-based file system.
β Key Takeaways
Positive audible feedback (sound for success, silence for failure) is dramatically more user-friendly than negative audible feedback (beeps for errors), which publicly announces user failure. The About box should be redesigned as a true About box that tells users what the program does, rather than just an Identity box with legal information. Wizards should be reserved for very rare tasks and replaced with automatic functions that let users modify results later. The most critical insight for data retrieval is that current computer systems have no retrieval system separate from storage β users must remember file names and locations. Attribute-based retrieval systems that automatically learn document characteristics and allow searching by content and usage patterns are essential for modern computing.
π§ Quick Revision Questions
- What are the key differences between negative and positive audible feedback, and why is silence more effective for indicating errors?
- What is the difference between an Identity box and a true About box?
- What are the three fundamental methods for finding documents on a computer?
- Why are wizards problematic for user interfaces, and when are they appropriate?
- What attributes could an automatic attribute-based retrieval system learn without user setup?
π Lecture 44 β Emerging Paradigms
π Overview: This lecture explores two critical paradigms in modern web design: information architecture through metadata and controlled vocabularies, and web accessibility for people with disabilities. It explains how metadata systems and thesauri create more navigable websites while examining the ethical, legal, and practical imperatives for making web content accessible to all users regardless of ability.
ποΈ Topics Covered
The lecture first addresses metadata and controlled vocabularies as fundamental components of information architecture, exploring how thesauri model semantic relationships between terms. It then examines web accessibility in depth, covering the opportunities the web offers people with disabilities, current shortcomings, legal requirements, and comprehensive solutions including organizational commitment and technical training.
π Lecture Summary
Metadata
A web site is a collection of interconnected systems with complex dependencies. A single link on a page can simultaneously be part of the site's structure, organization, labeling, navigation, and searching systems. Metadata and controlled vocabularies present a fascinating lens through which to view the network of relationships between systems. In many large metadata-driven web sites, controlled vocabularies have become the glue that holds the systems together. A thesaurus on the back end can enable a more seamless and satisfying user experience on the front end.
When it comes to definitions, metadata is a slippery fish. Metadata is definitional data that provides information about or documentation of other data managed within an application or environment. For example, metadata would document data about data elements or attributes (name, size, data type, etc) and data about records or data structures (length, fields, columns, etc) and data about data (where it is located, how it is associated, ownership, etc.). Metadata may include descriptive information about the context, quality and condition, or characteristics of the data.
Metadata tags are used to describe documents, pages, images, software, video and audio files, and other content objects for the purposes of improved navigation and retrieval. The HTML keyword meta tag provides a simple example: <meta name="keywords" content="information architecture, content management, knowledge management, user experience">
Many companies today are using metadata in more sophisticated ways. Leveraging content management software and controlled vocabularies, they create dynamic metadata-driven web sites that support distributed authoring and powerful navigation. This metadata-driven model represents a profound change in how web sites are created and managed. Instead of asking, "Where do I place this document in the taxonomy?" we can now ask, "How do I describe this document?"
π Definition β Metadata: Definitional data that provides information about or documentation of other data managed within an application or environment.
π Formula: N/A
π Example: HTML meta tags like <meta name="keywords" content="information architecture, user experience"> allow authors to enter words and phrases that describe content, not displayed in the interface but available for search engines.
Controlled Vocabularies
Vocabulary control comes in many shapes and sizes. At its most vague, a controlled vocabulary is any defined subset of natural language. At its simplest, a controlled vocabulary is a list of equivalent terms in the form of a synonym ring, or a list of preferred terms in the form of an authority file. Define hierarchical relationships between terms (e.g., broader, narrower) and you've got a classification scheme. Model associative relationships between concepts (e.g., see also, see related) and you get a full-blown thesaurus.
Classification schemes can be used in the context of searching. Yahoo! does this very effectively by presenting "Category Matches" in search results, which reinforces users' familiarity with Yahoo!'s classification scheme.
The important point is that classification schemes are not tied to a single view or instance. They can be used on both the back end and the front end in all sorts of ways.
π Definition β Controlled Vocabulary: Any defined subset of natural language used to improve information retrieval, ranging from simple synonym rings to complex thesauri. π Formula: Types: Synonym ring β Authority file β Classification scheme β Thesaurus (increasing complexity) π Example: Yahoo! search results show "Category Matches" like "Bizarre Humor > Hamster Dance" or "Rodents > Hamsters" for a search on "hamsters"
Thesauri
A thesaurus is a semantic network of concepts, connecting words to their synonyms, homonyms, antonyms, broader and narrower terms, and related terms. For web purposes, the thesaurus takes the form of an online database, tightly integrated with the user interface of a web site or intranet. While the traditional thesaurus helps people go from one word to many words, our thesaurus does the opposite. Its most important goal is synonym management β the mapping of many synonyms or word variants onto one preferred term or concept, so the ambiguities of language don't prevent people from finding what they need.
For the purposes of this book, a thesaurus is: "A controlled vocabulary in which equivalence, hierarchical, and associative relationships are identified for purposes of improved retrieval." β ANSI/NISO Z39.19-1993
A thesaurus builds upon the constructs of the simpler controlled vocabularies, modeling three fundamental types of semantic relationships. Each preferred term becomes the center of its own semantic network. The equivalence relationship is focused on synonym management. The hierarchical relationship enables the classification of preferred terms into categories and subcategories. The associative relationship provides for meaningful connections that are not handled by the hierarchical or equivalence relationships.
π Definition β Thesaurus (web): A controlled vocabulary in which equivalence, hierarchical, and associative relationships are identified for purposes of improved retrieval. π Formula: Semantic relationships: Equivalence (synonyms) + Hierarchical (broader/narrower) + Associative (related terms) π Example: A search for "automobile" might map to preferred term "car," with broader term "vehicle," narrower term "sedan," and related term "gasoline"
π‘ Why this matters: Thesauri convert the inherent ambiguity of natural language into precise, machine-manageable relationships, dramatically improving search accuracy and navigation.
44.1 Accessibility
Accessibility is a general term used to describe the degree to which a system is usable by as many people as possible without modification. It is not to be confused with usability, which is used to describe how easily a thing can be used by any type of user. One meaning of accessibility specifically focuses on people with disabilities and their use of assistive devices such as screen-reading web browsers or wheelchairs.
Accessibility is strongly related to universal design in that it is about making things as accessible as possible to as wide a group of people as possible. The disability rights movement advocates equal access to social, political and economic life which includes not only physical access but access to the same tools, organisations and facilities which we all pay for.
Various countries have legislation requiring physical accessibility:
- In the UK, the Disability Discrimination Act 1995 has numerous provisions for accessibility
- In the US, under the Americans with Disabilities Act of 1990, new public and private business construction generally must be accessible
- In Ontario, Canada, the Ontarians with Disabilities Act of 2001
π Definition β Accessibility: The degree to which a system is usable by as many people as possible without modification, particularly focusing on people with disabilities.
Introduction to Web Accessibility
The Internet provides access to information, news, email, shopping, and entertainment. Some have argued that no other single invention has been more revolutionary since Gutenberg's printing press. Now, at the click of a mouse, the world can be "at your fingertips"βthat is, if you can use a mouse, see the screen, and hear the audio.
The Web Offers Unprecedented Opportunities β The Internet is one of the best things that ever happened to people with disabilities. For example, without the Internet, blind people mostly couldn't read newspapers independently. With the advent of the World Wide Web, many newspapers now publish their content electronically in a format that can be read by text-to-speech synthesizer software programs (often called "screen readers") used by the blind. Suddenly, blind people don't have to rely on others to read to them.
Similarly, people with motor disabilities who cannot pick up a newspaper can access online newspapers using assistive technologies that adapt the computer interface. Adaptations range from crude (using a stick in the mouth to type) to sophisticated (eye-tracking software). People with tremors may use special keyboards with raised ridges between keys. Most of these adaptations make use of the keyboard or emulate keyboard use rather than mouse use.
People who are deaf can read online transcripts of important speeches or view multimedia content that has been fully captioned.
Falling Short of the Web's Potential β Despite the Web's great potential for people with disabilities, this potential is still largely unrealized. Where can you find web-based video or multimedia content that has been fully captioned for the deaf? What if content is only accessible by using a mouse? What if developers use all graphics instead of text, preventing screen readers from reading to blind people?
People with Disabilities on the Web β Though estimates vary, most studies find that about one fifth (20%) of the population has some kind of disability. Businesses would be unwise to purposely exclude 20, 10 or even 5 percent of their potential customers. Schools, universities, and government entities would be not only unwise, but in many countries, also breaking the law.
Each major category of disability requires certain types of adaptations in web content design. Most of the time, these adaptations benefit nearly everyone. People with cognitive disabilities benefit from illustrations, graphics, properly-organized content with headings, lists, and visual cues in navigation. Though captioned video is meant for deaf people, it also benefits those without sound on their computers or those in public places.
Occasionally developers must implement more specific accommodations, such as links allowing blind users or those with motor disabilities to skip past navigational links. In almost every case, these adaptations can be integrated with little or no impact to visual "look and feel."
From the perspective of people with disabilities, inaccessible web content prevents them from participating fully in the information revolution. It is a matter of basic human rights.
Comprehensive Solutions β There are two key components to any effort to achieve web accessibility:
- Commitment and accountability
- Training and technical support
Either of these by itself is insufficient.
Commitment and accountability requires: (1) Awareness β Most accessibility errors result from ignorance, not malice; developers who understand the issues are usually willing to help. (2) Leadership β If organizational leadership doesn't express commitment, chances are low that content will be accessible. (3) Policies and Procedures β The best approach is to create internal policy outlining specific standards, procedures, and monitoring methods, such as requiring compliance with W3C Web Content Accessibility Guidelines.
Training and technical support addresses the fear that accessible sites are more expensive. On a page-by-page basis, the extra time required by a knowledgeable developer is almost negligible. A developer can learn basics in a few days but may need months to internalize the mindset. Organizations should ensure access to training materials, workshops, books, or courses. Resources like WebAIM are available for free.
Ongoing technical support can be offered through outside consultants, discussion groups, internal workshops, or classes. Some organizations set up internal discussion groups for accessibility issues.
π Definition β Screen Reader: Text-to-speech synthesizer software programs that read text out loud so that blind people can use computers and access text content. π Definition β Universal Design: Making things as accessible as possible to as wide a group of people as possible.
β Key Takeaways
Metadata and controlled vocabularies form the invisible infrastructure of modern websites, with thesauri modeling three critical semantic relationships β equivalence, hierarchical, and associative β to enable powerful navigation and retrieval. Web accessibility is both a legal requirement under acts like the Americans with Disabilities Act and a moral imperative, given that approximately 20% of the population has some disability. The web offers unprecedented independence to people with disabilities through screen readers, keyboard-based navigation, and captioning, yet most sites fail to realize this potential. Achieving accessibility requires both organizational commitment with clear policies and accountability, plus ongoing training and technical support for developers. Importantly, most accessibility adaptations benefit all users, not just those with disabilities, making it a universal design principle rather than a niche concern.
π§ Quick Revision Questions
- What are the three fundamental types of semantic relationships modeled in a thesaurus, and how does each improve information retrieval?
- How does the metadata-driven model change the fundamental question web developers ask when adding content to a site?
- What percentage of the population has some kind of disability, and why is this significant for web design decisions?
- What are the two essential components of a comprehensive web accessibility solution, and why is each component alone insufficient?
- How do screen readers change the newspaper-reading experience for blind people compared to pre-Internet methods like Braille or audiotapes?
π Lecture 45 β Conclusion
π Overview: This concluding lecture explores emerging interaction paradigms beyond traditional desktop computing. It introduces students to ubiquitous computing, wearable technology, tangible user interfaces, and attentive environmentsβall representing the future direction of Human-Computer Interaction. Understanding these paradigms is essential for anticipating how computing will integrate seamlessly into everyday life.
ποΈ Topics Covered
The lecture presents four major emerging interaction paradigms: ubiquitous computing (integrating computation into the environment), wearable computing (computers worn on the body), tangible bits (physical objects as interfaces), and attentive environments (user and context-aware spaces). It traces the historical development and philosophical underpinnings of each paradigm, from Mark Weiser's foundational work at Xerox PARC to modern applications in military, medical, and consumer contexts.
π Lecture Summary
Ubiquitous computing
Ubiquitous computing (ubicomp, pervasive computing, calm technology, things that think) integrates computation into the environment rather than having computers as distinct objects. The goal is enabling people to move around and interact with information and computing more naturally and casually. Devices should sense changes in their environment and automatically adapt based on user needs and preferences.
The late Mark Weiser wrote seminal papers beginning in 1988 at the Xerox Palo Alto Research Center (PARC). The MIT Media Lab carries on research under the name "Things That Think." Writer Adam Greenfield coined the term Everyware to describe technologies of ubiquitous computing, pervasive computing, ambient informatics, and tangible media.
π Definition β Ubiquitous Computing: Computing that is invisible, everywhere, and does not live on a personal device but is integrated into the environment.
π‘ Why this matters: This paradigm represents a fundamental shift from "dramatic" machines we consciously use to "invisible" computing that blends naturally into our surroundings.
Early work in Ubiquitous Computing
The initial incarnation was "tabs", "pads", and "boards" built at Xerox PARC (1988-1994). Ubicomp helped kick off mobile computing research, though it is neither the same nor a superset/subset.
π Example β Weiser's vision: "Activate the world. Provide hundreds of wireless computing devices per person per office, of all scales (from 1" displays to wall sized)."
Weiser distinguished two paths: the "dramatic" machine (making computers so exciting we never want to be without them) and the "invisible" path (making computers so imbedded, fitting, and natural that we use them without thinking).
45.1 Wearable Computing
Wearable computers break the limitation of traditional personal computers by being worn on the body like clothing. They have been applied to behavioral modeling, health monitoring, information technologies, and media development. Government organizations, military, and health professionals use them daily. They are especially useful when hands, voice, eyes, or attention are actively engaged with the physical environment.
π Definition β Wearable Computer: A computer worn on the body that provides constant interaction, multi-tasking capability, and can act as an extension of the user's mind or body.
Main features: constancy (no need to turn on/off) and ability to multi-task (augmented into all other actions). These devices can be incorporated as a prosthetic extension.
π Example β Military application: The US Army's Land Warrior system (eventually merged into Future Force Warrior) is the most extensive military program in wearables.
Issues
Wearable computers have led to an increase in micro-managementβa society characterized by total surveillance and greater influence of media and technologies. There is concern about cameras affecting personal and private moments as a form of social control.
History
π Example β First wearable computer (1961): Mathematician Edward O. Thorp and Claude E. Shannon constructed a concealed cigarette-pack sized analog computer to predict roulette wheels. It used microswitches in shoes to indicate wheel speed, sending musical tones via radio to a collaborator's ear canal.
π Example β 1967: Hubert Upton developed an analogue wearable with eyeglass-mounted display to aid lip reading using high and low-pass filters.
π Example β 1981: Steve Mann built a backpack-mounted 6502-based computer to control photographic systems. He later created the Wearable Wireless Webcam (1994).
π Example β 1989: Reflection Technology marketed the Private Eye head-mounted display.
π Example β 1993: Columbia University's KARMA (Knowledge-based Augmented Reality for Maintenance Assistance) overlaid wireframe schematics on real objects being repaired, such as explaining how to change a paper tray in a laser printer.
Commercialization attempts by Xybernaut, CDI, ViA Inc, Seiko (Ruputer wristwatch), IBM (Linux wristwatch prototypes), and Fossil (WristPDA running Palm OS) met with limited success.
45.2 Tangible Bits
The development from desktop to physical environment has two phases:
- Xerox Star workstation (1981) β First generation graphical user interface establishing the "desktop metaphor" with "seeing and pointing" design principles.
- Marc Weiser (1991) β "Ubiquitous computing" paradigm displacing computers into the background.
Tangible User Interfaces (TUIs) make computing truly ubiquitous and invisible by changing the world itself to an interfaceβall surfaces (walls, ceilings, doors) and objects become interfaces.
π Definition β Tangible Bits: Coupling digital information to everyday physical objects, making digital information tangible and graspable.
π Examples:
- ClearBoard (1990-95): Changing a passive wall to an active dynamic collaboration medium
- Bricks (1990-95): Graphical interface allowing direct control of virtual objects through physical handles called "Bricks"
- Marble Answering Machine (Durell Bishop): Incoming messages represented by marblesβuser grasps and drops to play message or dial caller automatically
Goals and Concepts of "tangible bits"
Goals: Augment the real physical world by coupling digital information to everyday things, bringing cyberspace and real world together. All states of physical matter (solids, liquids, gases) become interfaces. Users should "grasp and manipulate" foreground bits and be aware of background bits. No distinction between input and output devices.
Three key concepts:
- Interactive surfaces β Transformation of each surface into an active interface
- Coupling bits and atoms β Seamless coupling of everyday objects with digital information
- Ambient media β Use of sound, light, air flow, water movement for background interfaces at periphery of human perception
45.3 Attentive Environments
Attentive environments are user and context-aware. IBM's BlueEyes research project explores and defines these environments. The goal is creating computational devices with perceptual abilities that people take for granted.
π Definition β Attentive Environment: An environment that senses user actions, extracts key information, and determines user's physical, emotional, or informational state to perform expected actions or provide expected information.
π Example β BlueEyes-enabled television: Could become active when user makes eye contact; user could then tell television to "turn on."
π‘ Why this matters: Adding perceptual abilities to computers enables humans and computers to work together more as partners, rather than tools requiring explicit commands.
β Key Takeaways
The most critical takeaway is that HCI is evolving from explicit, desktop-based interaction toward invisible, context-aware computing embedded in our environment. Ubiquitous computing aims to make computers "invisible in use" by integrating them into everyday objects and spaces. Wearable computing extends this by making computers part of our clothing and bodies, offering constant interaction and multi-tasking capabilities. Tangible Bits demonstrate a fundamental shift from graphical interfaces to physical objects as interfaces, coupling digital information with physical matter. Finally, attentive environments add perceptual intelligence, allowing computers to sense user states and respond proactively rather than requiring explicit commandsβtruly transforming computers from tools to partners.
π§ Quick Revision Questions
- What distinguishes Mark Weiser's "invisible" path of computing from the "dramatic" machine path?
- What are the two main features of wearable computers, and how do they differ from traditional personal computers?
- What were the three key concepts behind Tangible User Interfaces (TUIs)?
- How does the Marble Answering Machine exemplify the concept of "tangible bits"?
- What is the primary goal of IBM's BlueEyes project in creating attentive environments?