Foundations of Human-Computer Interaction - Book Review - The Design of Everyday Things by Don(ald) Norman

Don Norman is a Computer Scientist, Electrical Engineer, a Doctor of Psychology and a professor of Human Centered Design. He has written The Psychology of Everyday Things (later The Design of Everyday Things) to deal with frustration: the frustration of being unable to operate the most simple of devices. This inability, Norman realized, was not a consequence of his own perceived stupidity, but was a consequence of bad user interfaces, in other words, bad design.

In The Psychopathology of Everyday Things, Norman discusses that objects shouldn't need instruction to be usable. Interfaces should have visibility, that is, we should be able to see what an object is capable of. These visible messages are natural signals and the design that takes advantage of those is natural design.

Natural mapping is an expansion of the concept of mapping, a direct relationship between what is perceived as possible and intention. Natural mapping uses physicality and culture as guidelines for those mappings, making this relationship understandable.

Affordances, without a doubt the most interesting concept discussed in this chapter (and maybe the book) are properties of a given object. They describe what is possible to do with an object and are the attributes that make objects tools: they give purpose. Constraints on the other hand are like anti-affordances: they describe what we cannot do with a given object or the limits of what we can do. Putting all of these concepts together gives us conceptual models, the mental image of an object, all you can and cannot do with it and the relationships between intention and possibility.

He concludes saying that designing for people is all about making good conceptual models (therefore, understanding mappings, affordances and constraints) and making things visible.

In The Psychology of Everyday Actions we learn the 7 stages of action: forming a goal, forming an intention, specifying an action, executing an action, perceiving the world, interpreting its state, evaluating the outcome; rinse, repeat. We also learn of the two Gulfs, the gulf of execution, the distance between the intentions of the user and the actions possible with the object (see mapping) and the gulf of evaluation, the distance between the physical (visual, perceptual) presentation of the object and the intentions of the user (see visibility, affordances).

Norman also speaks of naïveté through Aristotle, mentioning that we tend to hold very simple, and often wrong, explanations for the phenomena around us. Coincidence leads to perceived causality, making us see relations between actions and results when they are not there. He also discusses silence, the insistence in not reporting errors that we believed are caused by ourselves. This conspiracy of silence leads to helplessness, the perception that we are simply unable to complete a task. This can lead to its own cycle, of taught helplessness, where we generalize our inabilities (and our fear of discussing them) leading to the habit of not trying, which leads to a self-fulfilling prophecy (you prove your own perceived inability by failing to even try to succeed).

Knowledge in the Head and in the World is about memory. Memory in the head is that efficiently, easily retrieved knowledge that we acquire by learning. Memory in the world is information like mappings and reminders, the information that is external to us but, after retrieval (seeing, hearing) and interpretation, becomes content that we can also use to act.

These two types of memory allow us to have precise behavior without precise knowledge:

- We don't need all the information to be within our heads, we can retrieve it from the world.

- Precision is unnecessary; as long as we have enough knowledge to distinguish right from wrong, so we can interpret the external information to complete a task.

- The world limits possible actions through natural constraints, so we do not need to know all that is possible/impossible, we can infer through interaction

- Additionally, cultural constraints are carried with us through learning, so we know what is acceptable to do with a certain object.

A design model is the ideas of the object's designer on what can and cannot be done with it. A user's model is that which the user develops through observation of this object and a system image is the totality of information available on a given object (affordances, constraints, instructions, appearance). In an ideal world, the design and user model are identical. The system image should also perfectly align the appearance and instruction set of an object with the real affordances and constraints of an object.

Knowing What to Do expands on constraints by attempting to assemble a Lego motorcycle: physical constraints are true limitations of size, weight, mobility; semantic constraints are limitations based on meaning (the windshield goes in front of the driver, the driver points forward, etc); cultural constraints are about accepted conventions (the red light means stop, goes in the rear; text is to be read, so it must be right-side-up, etc) and logical constraints are limitations in logic (all bike parts must be connected to work, no parts can be missing, etc).

Constraints are as important as affordances to tell us what to do. If constraints are deceiving, we are unable to understand what is really possible with an object. If you can perceive more than one possibility from a single part, you are a victim of bad mapping.

To Err is Human is about making mistakes. The most common type of error is a slip, when we end up doing something when we intended to do another thing. Slips can be capture errors (two sequences start the same and you unconsciously move from one to the other), descriptive errors (we confuse two look-alike objects and do to one what we should do to the other), data-driver errors (we retrieve wrong data and work on this equivocal information), associative activation errors (the event gets a similar, but wrong, response), loss-of-activation errors (we forget what we were doing mid-action) and mode errors (objects with several available modes are used without first selecting the correct mode).

Good design uses feedback to prevent or warn about slips by showing there was a discrepancy between intention and result. 

Different levels of complexity are acceptable for different types of tasks:

- Deep tasks cannot be wide, but shallow tasks can (there can be many flavors of ice-cream to choose from without risk).

- Narrow tasks can be deep, but wide tasks can't (when cooking by following a recipe, there can be many steps because there are very little decisions to be made).

Good design deals with human error by understanding and minimizing causes for error, making it possible to undo actions, making errors easy to discover and fix and understanding that tasks are imperfect simplifications of users actions. And if all else fails, designers should resort to locks (forced sequences that prevent users from making mistakes).

The Design Challenge is about designers as human beings. We are at a constant balance between usability and aesthetics and the iterative process of design, analyse for errors and modify (repeat until time/budget/resources run out). We also have limited mental resources and cannot predict everything. Our attention is too focused and too limited. We also are not end users, and most of the times, neither is our client. Lastly, we have a tendency to ignore convention in the name of innovation, sometimes breaking affordances and constraints.

We must also accept that sometimes there are no perfect answers and the best solution is a balance.

Last but not least, the most common design sins: complex is cool or the worshiping of false images and creeping featurism, adding more and more features that add more complexity than utility.

User-Centered Design, the final chapter, is about the philosophy of good design. The chapter revisits every chapter of the book in order to posit what good design actually is.

A good design makes possible actions clear, makes everything visible (actions, results and the conceptual model itself), makes evaluation easy and follows natural mappings between intention and action, action and result and visible information and interpretation. It uses both knowledge in the world and knowledge in the head, it simplifies structure, it bridges the gulfs of execution and evaluation, it gets mappings right, it exploits constraints and it allows for dealing with error.

This was the first time I've read The Design of Everyday Things since my early days of Industrial Design college in 2007. It is one of my two favorite books by Don Norman, together with Emotional Design. Contrary to the opinion of most of my Industrial Design colleagues, I do not think this book is offensive to the profession, but rather a fantastic guide of good, human-centered design. Designers should be less egocentric and understand they are not artists, they are craftsmen who trade in human needs and whose handicraft are useful objects. Re-reading this book made me decide I must read something else Norman wrote between these two of my favorite books, probably Turn Signals are the Facial Expressions of Automobiles.

Foundations of Human-Computer Interaction - Final Concept Map, A.K.A. The Behemoth.

This thing is huge.
Gargantuan.
Confusing.

H. P. Lovecraft would write about it, terrorizing Rhode Island, domineering the Ancient Ones.

Cthulhu has nothing on The Behemoth.

I present thee "The Integrated Concept Map for Foundations of Human-Computer Interaction"

he com̡e̶s, ̕h̵i​s un̨ho͞ly radiańcé destro҉ying all enli̍̈́̂̈́ghtenment, concepts lea͠ki̧n͘g fr̶ǫm ̡yo​͟ur eye͢s̸ ̛l̕ik͏e liq​uid pain,  can you see ̲͚̖͔̙î̩́t̲͎̩̱͔́̋̀ it is beautiful t​he final snuffing of the lie​s of Man ALL IS LOŚ͖̩͇̗̪̏̈́T ALL I​S LOST  he comes the ich​or permeates all MY FACE NO NOO̼O​O NΘ stop the an​*̶͑̾̾​̅ͫ͏̙̤g͇̫͛͆̾ͫ̑͆l͖͉̗̩̳̟̍ͫͥͨe̠̅s ͎a̧͈͖r̽̾̈́͒͑e n​ot rè̑ͧ̌aͨl̘̝̙̃ͤ͂̾̆  H̸̡̪̯ͨ͊̽̅̾̎Ȩ̬̩̾͛ͪ̈́̀́͘ ̶̧̨̱̹̭̯ͧ̾ͬC̷̙̲̝͖ͭ̏ͥͮ͟Oͮ͏̮̪̝͍M̲̖͊̒ͪͩͬ̚̚͜Ȇ̴̟̟͙̞ͩ͌͝S̨̥̫͎̭ͯ̿̔̀ͅ

You have been warned.

Foundations of Human-Computer Interaction - Module 7 - History and Vision

I tried very hard not to use a timeline structure... and failed. I guess I'm hardwired to think of History as linear. One of those things David calls my tendency of being too aseptic, I guess.
It was very cool to revisit this timeline in my head, as I've written extensively about the oN-Line System (NLS) and Doug Engelbart's unsung parenthood of so many things we now take for granted in Human-Computer Interaction: the pointer, the collaborative document, tele-conferencing, hypertext etc.
So here it is, in a timeline-ish shape, the final Concept Map before integration:

Design for All - Main Assignment - Public Transport and the Disabled

Background

This study was made with a personal friend, whose identity I chose to preserve. He is a foreign national living in Estonia, like me, but he has been here for a few years and is an amputee who needs a wheelchair for mobility on a daily basis. He is an otherwise healthy man in his late 20s to mid 30s. He chooses to only move around by car and one day we decided to take a bus ride together so we could discuss the challenges of using public transport when you have a disability.

We chose to stack the deck in our favor and picked a non-busy weekend's day to move between two locations downtown. As he explained to me, Tallinn is a nice place mobility-wise for the disabled, as long as you only look to the streets and sidewalks. Most if not all street corners have dropped curbs that make climbing up and down with a wheelchair fairly easy. Buildings are another story completely, as most buildings have flights of stairs, raised steps or a combination of both to access the interior, including rather modern ones. He lives in a modern building complex that has no elevators whatsoever, and he has to climb up or down several flights of stairs, something he does using his wheelchair as a makeshift crutch in a rather ingenious way, moving it up a few steps and leaning on it to go to the next step, then repeating the procedure until he is up. The process of doing it for several flights of stairs is rather painful and takes an insurmountable amount of time, as I've witnessed.

Once we hit the streets he is quite ambulatory and his youthful strength keeps him moving as fast as everyone in the group. Some dropped curbs are easier to negotiate than others, but on average he has no trouble at all moving within the city. We reach the bus stop and check the timetable. I am completely unable to read Estonian, but he is capable of negotiating a few words of the language (an ability that would come useful soon enough, as we were about to learn). He tells me that the buses with wheelchair accessibility are marked by a blue line under the minutes on the timetable. Luckily for us, Tallinn buses come regularly and the next scheduled bus passing has a blue line under it, so we wait and lo and behold, a bus.

The Problem

The bus driver stopped and, as you can see in the picture above, the low floor on the bus was not enough to enable you to negotiate the distance between the curb and the bus without additional help. We had no idea how to access the bus and were about to give up and let the bus go when my friend rolled to the front door and asked the driver how to board using a wheelchair. The driver stops, opens his personal door, gets a pair of gloves, goes to the middle door of the bus and pulls a handle from the floor. There is a very well hidden, unsignalized ramp on the floor! I pushed him up the ramp, as it looks quite steep, but he said that, for the record, he could do it himself with some effort.

We took the bus for a few blocks, staying in the designated wheelchair space. It has a few seatbelt-like appendages that we assume are to keep the wheelchairs strapped in case of accident, but he jokingly explained that nobody else is wearing a seatbelt, so he does not feel it is a fair option to be the only one strapped to the bus. When we arrived at the destination, I (who cannot speak a word of Estonian) took it upon myself to pull the handle and unfold the ramp so he could climb down. We negotiated a few more dropped curbs on our way to a restaurant to discuss the experience we'd just had.

Discussion

He explained that riding his own car he retains a level of independence that he could not have with the public transport. Since the ramps are not automated and pushing the button inside the bus simply informs the driver that he has to come to the middle of the bus to unfold the ramp, we agree that it breaks the freedom of mobility expected from a true accessible experience, and the fact that we could not find a button to do the same thing from the outside makes us assume the only way to do it is to actually speak Estonian and bother the driver.

We drew the following capability levels scale comparing his actual abilities to the abilities needed to use the accessibility ramp. It is surprisingly difficult to do it, considering you need to communicate your need for it. It also made me feel quite helpless if I were in his place, because I would most certainly not be able to ask the driver to operate the ramp for me nor do it myself from a chair. It is disconcerting to see an accessibility feature that matches so much the actual abilities of the person who needs it that any slight difference might make or break the usability scale.

Conclusions

In further discussion, we reached together the conclusion that the fact that the ramp is not automated makes a lot of difference, but what really makes this option a kludge instead of a real solution is the terrible signalization and communication features. Not having a way to communicate, from the outside, that you need the accessibility ramp makes it borderline useless. It is the only high-demand point in the whole scale, and it needs addressing. I myself felt extremely unable to do the exact same thing if I were on a wheelchair and alone and although I'm a strong proponent of using public transport on a daily basis I could not find any fault in his decision to only move around by car.

Design for All - Essay - Universal Design for Automobiles

As in many other areas, automotive designers are often squeezed between conflicting demands: safety, structural integrity, performance, aesthetics, fashion, marketing, accessibility. It is often very hard to equally satisfy the needs for all of these demands, and economic forces will push designers into prioritizing one against the others: looks over safety, fashion over accessibility. As automobiles are often the object of fetishistic desires from their consumers (and are expensive status symbols), it is often very hard to convince people to prioritize safety and accessibility when purchasing automobiles, making the market a powerful force against universal design.

Around half of older, frailer folk have difficulty entering and exiting personal vehicles. This is not helped by the trend for sportier designs, either of low ceilinged, shallow seated sporty coupés or sedans that aspire to look like coupés, or of high floored, high seated sports utility vehicles (SUV), vans and minivans. The latter two categories are very often the vehicle of choice for people with limited mobility because of the generous space for wheelchair and other walking aids, but the push for SUV-like looks causes most of these vehicles to have a much higher floor than actually needed in their true usage scenarios.

The back seat of most vehicles (the seat of choice for many people who are being transported by their progeny or next-of-kin) is especially troublesome; aerodynamics push the ceilings down, comfort for the driver (the purchaser of the vehicle most of the time) pushes the seats back and down, engineering dictates the existence of a B-column between the front and rear doors, etc. That makes accessing the back seat very difficult for people with reduced mobility, dexterity and stretch. The best-case scenario for these passengers is the use of the front passenger seat on two-doored vehicles. These are also very sough-after by users of mobility aids capable of transferring themselves from chair to seat. For these users, the ability to access the leg-space between front and rear seat using a single door is especially useful for stowing away a foldable mobility aid.

These two-doored vehicles are often of a sporty variety, though. That makes their front seats usually low and shallow, which limits field-of-view and makes accessing controls farther away, like automatic rear-view mirrors, ventilators and even some less commonly-used lighting controls like headlamp angle regulators, as well as external features like toll booths, parking-slip emitters, intercom buttons and other drive-through features especially difficult to operate for people with limited reach and stretch ability. In these cases, the higher-seated SUVs and vans make for better seating, both from a security/visibility point of view as well as for reaching external road features and farther-removed controls within the car. Some of these constraints can be addressed by optional, remote-operated devices such as wireless toll tags, but those solve a single problem by addressing a symptom instead of trying to solve the positioning problem as a whole.

One class of vehicle that has been approaching the accessibility/positioning problem from the right angle are sub-compacts and electric vehicles. Perhaps because their target groups are more open-minded and less focused on sportiveness, most of these vehicles get some or all of these features right: low floors, high ceilings, upright seating, big front doors (usually the only door in the vehicle, but sometimes associated with special, “suicide” rear doors that open in the opposite direction without a B-column in-between). These features usually make for less sporty looking cars, but make for very universal features.

Staying in the area of visibility, aging takes away lots of sensory sensitivity: low light vision, dynamic acuity, visual field, hearing of higher frequencies and overall, etc. This makes for very dangerous situations where drivers cannot assess distance or speed of obstacles, cannot see approaching obstacles from the sides or the rear, cannot hear sirens, horns or other warning signs of approaching obstacles, etc.

There are several aids being developed and some already deployed for these issues: HUDs, night-vision systems, parking aids, rear-view cameras, collision-avoidance sensors, collision-warning systems and other crash prevention techniques.

Night-vision systems are usually deployed in luxury vehicles and employ one or more special cameras to absorb road information in ways the human eyes are incapable like LIDAR (light detection and ranging) and infrared or heat-sensing. Different systems process this information in different ways, but in most cases the information is displayed on a screen in the dashboard. This can be very useful, but also very distracting. It also lacks any kind of depth of view, making judging distances and avoiding detected obstacles very difficult.

Some collision-warning and crash prevention systems build on top of these sources of information by further analysing them and generating warning on assumptions made by image analysis. They can, for instance, infer that an obstacle is human-shaped and warn of pedestrians in the driveway, or use reflected light as a way of measuring distance and warn of impending collisions. Some of these systems use mixed visual and auditory feedback, which makes them more accessible. Some also use haptic feedback, but some drivers cannot tell the difference between a warning vibration and a mechanical defect or vehicle quirk, so proper combination of visual, auditory and haptic feedback is needed.

Some systems go one step further and connect this collision-warning information to the cruise-control and dynamic braking systems, effectively taking decision faster than a human driver could (even a very capable one with perfect vision and hearing). These vehicles can decelerate, shift gears and even apply selective braking to one or all wheels to prevent accidents. This combined collision-warning and cruise control system is called adaptive cruise control.

Another way of giving visual feedback is in the form of HUDs or Heads-Up Displays. This military-born technology uses images reflected on the windshield to give vehicle information to the driver, diminishing the need for moving the eyes from the road. These systems can be used for common visual information (speed, current gear), warning systems (collision warning, reckless driving, impending mechanical failure) and navigation (turn-by-turn instruction), among others. These systems must be used with care, though, because this information still makes drivers shift focus from the road to the windshield (people with limited fields of view can have difficulty adapting to different depths), can cause information overload (especially for older drivers whose sensory sensitivity is diminished and therefore cannot cope with too much information at the same time). It is best to combine HUDs with auditory warnings and to be very selective of what information must be displayed on HUDs.

Speaking of HUDs and sensory overload, one common on-board technology nowadays is navigation systems. Older drivers lose confidence in their ability for path-finding, choosing to only drive in familiar roads. The ubiquity of navigation systems brings confidence back to these drivers, restoring locomotive capacity by instructing them on unfamiliar paths. Most of these systems rely on stored maps and path-finding algorithms, making them extremely limited when it comes to alternate routing and hazard avoidance, though. More modern systems use external sources of information (internet connectivity, RDS Traffic Message Channel) to adapt to situations in real time. But these systems can also cause information overload, generate misunderstanding and cause even more unfamiliar situations. Some drivers will blindly follow information given by navigation systems disregarding surrounding environmental warnings, like shifting lanes when prompted without checking their rear view mirrors first or making illegal U-turns or conversions attempting to follow the system prompts. Some will be put in unfamiliar and uncomfortable situations like driving in highways, making left turns and other driving situations they usually avoid. Adaptive navigation algorithms can learn of these preferences and generate less direct but more familiar driving experiences for inexperienced or older drivers.

Another problem very common to navigation systems is shared with on-board entertainment systems (or infotainment systems, as most integrated, user-faced on-board computing is called nowadays): distraction. These systems take a lot of user input to operate and generate a lot of output, causing distraction, taking the eyes off the road for long periods, causing confusion, etc. Most mitigation techniques are a balancing act between these various perils: simplified input usually generates longer operation periods with higher distraction, shortened distraction periods require multiplexed input techniques that can be confusing. Alternative communication techniques such as voice input and output are very desirable, but are culturally and linguistically limited and are a long ways ahead in development before they can be considered universal. The only safeguard that can really guarantee safety when operating these infotainment systems is to limit their usage by passengers or when the vehicle is not mobile, but both are simply workarounds that generate more frustration from drivers (who are the decision-makers and market drivers, slowing the adoption of these mitigators).

A last area where universal design and vehicle design have still to catch-up to each other is legislation. Many laws can and should be implemented in this area, as leaving marketing forces alone to dictate vehicle design can significantly slow down adoption of very well established academic knowledge on universal vehicle design. Most of the factors analysed above, especially those regarding external and structural design, are caused by marketing efforts nullifying engineering and design efforts to integrate universal design and safety decisions. Like the impact that Ralph Nader’s Unsafe at Any Speed book had on highway safety legislation and vehicle design regulations in the US in the 1970s, universal design could and should have a significant impact on design decisions for automobiles, but without supporting legislation, there is no way to counter marketing forces pushing in the opposite direction.

Another area where legislation could become more flexible is creating adaptable licensing requirements. Just as most countries have special licensing regimes for younger drivers (like forcing them to signal their learner’s status or drive only on special periods of the day), creating special licensing regimes for older drivers could increase safety not only for them but also for other drivers around. It would also increase older people’s access to mobility by allowing them to retain their license to drive in restricted situations that cover most of their needs (like allowing for urban driving but forbidding access to highways), not forcing them to relinquish their license when no longer fit for dangerous driving situations.

Conclusion

The fast cycle of automotive renovation (although the automobile is a durable good, it has a rather short effective life cycle compared to housing, for instance) allows for very quick adoption of safety and universal design features. Marketing pressures, though, push these adoptions further away. The Trojan horse for the adoption of these design techniques can be the electric and hybrid vehicles. Their status as a niche product and their fresh legislation needs can be used as a way to push universal access requirements to their design features. The smart combination of legislation and fresh market status can push these new vehicles into a status of safe, accessible vehicles that can pull the other, more mainstream automobile categories towards these adoptions as a way to not lose their “cutting edge” when compared to electrics and hybrids.

References

Hakamies-Blomqvist, L., “Research on Older Drivers: A Review,” International Association of Traffic and Safety Sciences (IATSS) Research, 20:91–101, 1996.

Steinfeld, E., M. Tomita, W. Mann, and W. DeGlopper, “Use of Passenger Vehicles by Older People with Disabilities,” Occupational Therapy Journal of Research, 19(3):155–186, 1999.

National Highway Traffic Safety Administration, “Vehicle Backover Avoidance Technology Study—Report to Congress,” Washington: U.S. Department of Transportation, November 2006.

Preiser, W.F.E, Smith, K. H., “Universal Design Handbook”, McGraw Hill, 2011