The Design of Everyday Things

The Design of Everyday Things

Author
Donald A. Norman
Year
2013
image

Review

This book has changed how I see the world. I’ve always been interested, amused and sometimes annoyed by bad design. Now I see it everywhere.

By focusing on the very fundamentals of design - I’ve found this book to be consistently and endlessly applicable to work and life.

The concept of affordances progressed my understanding of accessibility. Thinking about conceptual models helps me grapple with product messaging. Learning about feedback has left me dumfounded that the button to call my elevator proudly sounds before the press has actually been registered. You have to push a little harder to actually call the lift 🤯

Developing a shared vocabulary - to talk about these concepts with my colleagues in design has helped accelerate day to day product conversations.

You Might Also Like…

image

Key Takeaways

The 20% that gave me 80% of the value.

  • Norman doors (named after the author) are doors that don’t open the way you expect.
  • An affordance: A capability made possible by the relationship between the human and the object.
  • A signifier: Anything that communicates the purpose, structure or operation of a device to a user
  • Mapping: How controls and displays map back to actions and intended results
  • Feedback: How the system lets you know a request is being worked on, or is complete
  • Conceptual models: A simplified understanding of how something works
  • The Technology Paradox: Technology has potential to make life easier and more enjoyable. At the same time it risks adding complexity, increasing difficulty and frustrating the user.
  • The Design Challenge: Different disciplines (engineering, marketing, operations) often have different priorities (scalability, reliability, price, differentiation, production). The design challenge is to bring everyone together, to build a product that customers love.
  • 7 Stages of action:
    • Form the goal
    • Plan the action
    • Specify an action sequence
    • Perform the action sequence
    • Perceive the state of the world
    • Interpret the perception
    • Compare the outcome with the goal
  • 7 Fundamental design principles
    1. Discoverability. Determining what actions are possible and the current state
    2. Feedback. Full & continuous info about the current state. Particularly after actions.
    3. Conceptual models. Invoke a model of the system that enhances discoverability and evaluation
    4. Affordances. The proper affordances exist to make the desired actions possible
    5. Signifiers. Ensure affordances are perceived, increasing discoverability & evaluation
    6. Mappings. Make the relationship between controls and actions predictable
    7. Constraints. Trim possible actions, to ease interpretation. Physical, logical, semantic & cultural.
  • Usability is often not prioritised in the purchasing process especially when the purchaser ≠ user.
  • 4 Classes of Constraints:
    • Physical limitations to the possible operations.
    • Cultural. Cultures have a set of allowable actions
    • Semantic: Only certain combinations make sense.
    • Logical constraint. There is a logical relationship between the spatial or functional layout of components and the things that they affect. E.g If take something apart, put it back together again, and there's a part left on the table. You know you've made a mistake.
A forcing function
a physical constraint such that failure at one stage, prevents the next step from happening
Interlocks
Forces operations to take place in proper sequence. Example: Washing machine door doesn’t open unless its drained water
Lock-ins
Keeps an operation active, preventing someone from prematurely stopping it. Example: Warning that makes it hard to leave an unsaved word document
Lockouts
A lockout prevents someone from entering a space that is dangerous, or prevents an event from occurring. Example: The pin in a fire extinguisher that prevents accidental discharge
  • The Forcing Function Tradeoff: Make it too annoying and people will try to disable it. So minimise the nuisance value whilst retaining the safety feature.
  • Consistency in design is virtuous. People are great at transfer learning (lessons learned with one system transfer readily to others). On the whole, consistency is to be followed.
  • Skeuomorphic: incorporating old familiar ideas into new technologies, even though they no longer play a functional role.
  • Key Design Principles
    • Put information in the environment → Reduce the burden of needing endogenous knowledge.
    • Allow for efficient operations when people have learned the requirements
    • Use environmental knowledge to make it easier for non-experts. This will help infrequent journeys and infrequent users
    • Leverage natural and artificial constraints: physical, logical, semantic and cultural.
    • Exploit the power of forcing functions and natural mappings
    • Bridge the gulf of execution and the gulf of evaluation.
    • Make things visible, both for executions and evaluation
    • One the execution side, provide the feedforward information: make the options readily available.
    • On the evaluation side: make the results of each action apparent.
    • Make it possible to determine the system's status readily, easily, accurately and in a form consistent with the person's goals, plans, and expectations.
    • Embrace errors. Seek to understand their causes and ensure they don't happen again. Re-design don't reprimand
  • Depth and Breadth Research Tradeoff. Design research is deep insight on a small number of people, Market research is shallow insight on a large number of people.
  • There is no such thing as the average person
  • Complexity is OK. Confusion is bad.
  • Design is successful only if the product is successful (purchased, used and enjoyed). Design should pay attention to the total experience and the total lifecycle. Design should be concerned with function, usability and understandability.
image

Deep Summary

Longer form notes, typically condensed, reworded and de-duplicated.

Some useful definitions if you’re new to design theory
  • Discoverability is determining what actions are possible and how to do them
  • Understandability is determining what everything means (what is it? how should it be used? what do the controls and settings mean?)
  • Industrial design is the study of function, value and appearance of products and systems
  • Interaction design is the study of human-product interaction, usability and understanding
  • Experience design is the study of the quality and enjoyment of an experience (products, events, etc)
A Norman Door
  • Named after the author Don Norman
  • Norman doors are doors that don't open like you'd expect.
  • They're an example of failed discoverability.
  • Doors should be easy to understand, a push plate or grab handle in the right place is enough to signal how to open and close (you don't need to label them push/pull).
Human Centered Design (HCD)
  • Design is concerned with how things work, how they are controlled and the nature of the interaction between the people and the thing. We design things for people, so we need to understand people, technology and how they interact.
  • Machines are simple and rigid, making them unforgiving (if we provide the wrong rules or inputs). When working with machines we must be precise and accurate. We design knowing that users are humans and humans make errors. Designers need to focus on where things go wrong, not just the happy path.
  • The Human-Centered design philosophy focuses attention on understanding people and their needs (primarily through observation). Equal importance is assigned to the problem definition (what is the right problem) and the solution definition. It also relies on the rapid testing of ideas. HCD overcomes two common issues: You're often asked to work on the wrong problem. Often people aren't able to articulate their needs well.

Fundamental Principles of Interaction

Affordances: Capabilities made possible by the relationship between the human and the object
  • Affordances are things an agent-object pair can do, a list of possible interactions. A chair has no affordances (only properties). A 'chair & human' have affordances: e.g.sit on, stand on, pick up, pull out.
  • Designers have a habit of assigning affordances to objects, but they are relationships, an affordance mush have an object and an agent.
  • Object-Agent pairs can have anti-affordances. Take (person-glass), there's an affordance of transparency but an anti-affordance of being able walk through it.
  • For designs to be effective, affordances and anti-affordances have to be discoverable. If an affordance can't be perceived a signifier is needed. Visible affordances provide strong clues to the operation of things. When an affordance is but not present issues occur. Information pickup is the term given to what we can perceive from all our senses.
Signifiers: Anything that communicates the purpose, structure or operation of a device to a user
  • Signifiers communicate the structure and operation of the device to the people that use it. They come in many forms (marks, sounds, languages). and can be deliberate (handles) or unintentional (subtle hand marks on a door).
  • Signifiers can turn an affordance into a perceived affordance by signalling possible actions, and how to complete them. Signifiers must themselves be perceivable, else they're unable to function.
  • Signifiers are more important than affordances, as they communicate how to use the design.
Mapping: How controls and displays map back to actions and intended results
  • Mapping is important to the layout of controls and displays.
  • Designers should try to achieve a clear mapping between the control > action > intended result.
  • Mappings work best when aligned to a conceptual model. With mappings, accuracy is less important than understandability.
  • Principles of mapping include grouping and proximity. Grouping: related controls should be grouped together. Proximity: controls should be close to the item being controlled.
  • Example of natural mapping: light switches in the same pattern as the lights, seat adjustment controls in the shape of a seat. Cultures have their own mappings, they're not always the same between cultures.
Feedback: System letting you know a request is being worked on, or is complete
  • Feedback is the system letting you know it's working on your request or communicating the result of the action. Control and Information theory is an entire area of study in itself. Feedback can come in many forms and act through any sense (visual, audible, touch, etc).
  • Feedback is essential, but it has to be done correctly, appropriately
Principles of Useful Feedback
  • Feedback must be immediate (delays of 1/10th of a second are disconcerting)
  • Feedback must be informative
  • All actions should be confirmed in an unobtrusive manner
  • Poor feedback can be worse than no feedback
  • Too much can be worse than too little.
  • Feedback should be prioritised.
    • Unimportant should be unobtrusive
    • Important should capture attention
  • Avoid inappropriate or uninterpretable feedback
    • Distracts from the task at hand
    • Alert fatigue: too many announcements, so all are ignored
Conceptual models: A simplified understanding of how something works
  • A model is an internal explanation (often simplified) of how something works. It doesn’t have to be complete or accurate, just useful. Clues to how things work come from: Signifiers, affordances, constraints and mappings.
  • Simplified models are only valuable if their assumptions hold true. Good models help us predict how something will behave, and predict the effects of our actions. Models are often inferred, shared, or learnt by experience.
More on Conceptual Models
  • Conceptual models are a form of story telling, they help us understand our experiences and predict the outcome of our actions.
    • In the absence of information → People form their own beliefs and stories
    • They quickly jump to creating their own models.
      • Inaccurate ones often lead to frustration.
      • E.g: some people set thermostats to higher temperatures thinking it will warm a room faster
  • People have a tendency to repeat an action when the first attempt fails. Systems should give feedback within 0.1 seconds, longer waits should be estimated and communicated to the user.
  • People try to find causes for events. They assign a causal relation whenever two things occur in succession.
  • People often blame themselves when they can't use an everyday object, this can create a conspiracy of silence, with people unwilling to speak out.
  • Learned Helplessness: When people experience repeated failure, they conclude they can't do the task and they stop trying.
  • Positive Psychology: Just as we learn to give up after repeated failures, we can learn optimistic positive responses to life. When something doesn't work, it can be perceived as an interesting challenge, or perhaps just a positive learning experience
    • Remove FAILURE and replace it with LEARNING EXPERIENCE: to fail is to learn, science leads into failure, so does design thinking

Advice for Designers

  • Don't blame those that fail. Their difficulties are a sign of what can be improved.
  • Eliminate error messages, instead, provide help and guidance
  • Make it possible to correct problems
  • Don't impede a users progress, allow them to continue towards completing their task
  • Never make people start over. Assume people are partially correct, let them make corrections
  • Don't criticise unless you can do better!
  • The system image: When the system design meets the user interaction. We should be designing thinking about the interaction. Communication is an important part of design, good conceptual models can help.
  • The Technology Paradox: Technology has potential to make life easier and more enjoyable. At the same time it risks adding complexity, increasing difficulty and frustrating the user.
  • The Design Challenge: Different disciplines (engineering, marketing, operations) often have different priorities (scalability, reliability, price, differentiation, production). The design challenge is to bring everyone together, to build a product that customers love.
The Psychology of Everyday Actions
  • It's important to understand what a user is thinking when they interact with your design. However, much of human behaviour is subconscious. Often we are bad at predicting how people will behave.
  • Learning things often requires conscious thought, but many actions can be offloaded to the subconscious once learnt, this allows us to focus on higher level goals and planning (learning to ride a bike).
    • Cognition: is our attempt to make sense of the world, our understanding
    • Emotion: assigns value - is something safe or dangerous? desirable or not?
  • We can't separate cognition and emotions, actions have expectations and expectations trigger emotions.
The Seven Stages of Action - Goal, 3x Execution, 3x Evaluation
  • How people think when performing a task.
  • First they form a goal, then they move through 3 stages of execution, followed by the 3 stages of evaluation.
1
Form the Goal
What do I want to accomplish?
execution
2
PLAN the action
Reflective → What are the alternative action sequences?
execution
3
SPECIFY an action sequence
Behavioural → What action can I do now?
execution
4
PERFORM the action sequence
Visceral → How do I do it?
evaluation
5
PERCEIVE the state of the world
Visceral → What happened?
evaluation
6
INTERPRET the perception
Behavioural → What does it mean?
evaluation
7
COMPARE the outcome with the goal
Reflective →Is this what I wanted? Have I done it?
  • Each stage of execution and evaluation happens at a different level of processing:
    • Reflective: Goal setting and planning activity. Slow conscious reasoning and decision making.
    • Behavioural: Expectations and learned skills. Largely subconscious. Actions and controls.
    • Visceral: Sensing and muscle control. The unconscious lizard brain, emotional and fast.
  • The designers job is to bridge the Gulf of Execution (figuring out how to use something) with feedforward, and the Gulf of Evaluation (figuring out the current state of something) with feedback.
  • All three levels of processing work together to determine a persons cognitive and emotional state. Higher-level reflective cognition can trigger lower-level emotions and vice versa. The reflective level though is where memories are formed, and memories are what drives us to recommend a product (or not). For frequent or routine actions, much of the action cycle is subconscious.

Goals and Actions

  • A Goal can be nested in a hierarchy of goals. UXR interviews and root cause analysis are good at identifying goal hierarchies.
  • Often a single pass of the 7 stages of action won't be enough to complete a goal (multiple subgoals or loops are required).
  • Action cycles can be triggered in service of achieving an overarching goal, or by an event relating to a lower-level goal or task.
  • Some goals are opportunistic - when behaviour takes advantage of circumstance. Typically less precise and certain. Less effort and less inconvenience.
  • Reconsidering the goals of a user can result in new product categories and innovation

Fundamental Design Principles

  1. Discoverability. Determining what actions are possible and the current state
  2. Feedback. Full & continuous info about the current state. Particularly after actions.
  3. Conceptual models. Invoke a model of the system that enhances discoverability and evaluation
  4. Affordances. The proper affordances exist to make the desired actions possible
  5. Signifiers. Ensure affordances are perceived, increasing discoverability & evaluation
  6. Mappings. Make the relationship between controls and actions predictable
  7. Constraints. Trim possible actions, to ease interpretation. Physical, logical, semantic & cultural.
'Human error' is often the result of poor design - we should rebrand it to system error
  • Often the design is faulty, and others will make the same errors. Designers should aim to minimise inappropriate actions, through affordances, signifiers, mapping and constraints.If the user still makes an inappropriate action, maximise the chance its discovered and rectified. People get interrupted and disturbed whilst completing a task, expect them to leave things incomplete or lose their place.
  • When users take inappropriate actions. Ask which of 7 stages of action does it fail? Which design principles are deficient? Most interactions with products are actually with a complex system, good design requires consideration of the entire system.
Endogenous vs Environmental Knowledge
  • Endogenous knowledge requires learning - it can be ephemeral, here now, gone later.
  • Environmental knowledge doesn't require learning - but can be more difficult to use. It also requires on the continued presence of the knowledge in the environment.
  • We can design devices to guide behaviour, the guidance doesn't have to be precise.
  • E.g: constraints can limit behaviour by limiting the number of choices (natural, cultural conventions)
  • The most usable devices are understandable on first use, provide enough information in the environment and users will be able to learn how to operate the device.
  • Declarative knowledge: facts and rules → is easy to write down
  • Procedural knowledge: may be harder to articulate and is best taught through demonstration (e.g learning to play an instrument).
  • We can use cues to provide information to the user (signifiers, mappings, physical constraints). Even physical layout can convey information (people naturally organise things spatially: putting things in piles, grouping things).
  • We have to differentiate the desired action from the other choices a user can make. People naturally look for distinguishing features amongst options.
  • Confusion is often a function of history, when rules change people get confused.
  • What appears good in principle, can fail when introduced to the world.
  • Constraints reduce choices, making things easier to follow and learn.
Designing for Human Memory + Reminders
  • Often systems and schemas are defined without considering the limitation of humans memory.
  • Short-term memory (working memory) contains material currently being thought about. It is very fragile, limited. So don't count on anything retained in short term memory. Fades with both time and the number of things stored.
  • Long-Term Memory is your interpretation of past events. Takes time and effort to store something and retrieve it. Our memory is almost unlimited, although subject to bias and distortion.
  • It's difficult to learn knowledge with no underlying meaning or structure like the letters of the alphabet (arbitrary knowledge). There's often no way to sense check arbitrary knowledge. Good conceptual models can provide meanings to arbitrary things. Pilots don't rely on memory, they use checklists and quickly enter information or write it down.
  • Prospective Memory: the memory for intentions. Key Components:
    • Signal: knowing you have to remember something
    • Message: remembering the information itself
    • Trigger: time based or location based can help

Talking about memories with friends sparks more recollection.

Mapping of Controls + Natural Mapping
Good
controls arranged in the same spacial configuration as the objects to be controlled
Better
controls are as close as possible to the object to be controlled
Best
mounted directly on the item to be controlled
  • Usability is often not prioritised in the purchasing process especially when the purchaser ≠ user.
  • Cultural mappings can differ. Cultural mappings can change over time. Apple changed the default scrolling direction when it introduced touch devices.
  • It is possible to break convention and switch metaphors, but expect a period of confusion until people adapt.

Constraints Discoverability and Feedback

  • Environmental knowledge (perceived affordances, signifiers, mapping, constraints) and Endogenous knowledge (conceptual models, constraints, similarities to other situations we've face) help us operate things we haven't seen before.
  • Designers should provide critical key information to the user in the environment.
Four Classes of Constraints
  • Physical limitations to the possible operations. More effective if they're easy to see and interpret.
  • Cultural. Cultures have a set of allowable actions. Guidelines for cultural behaviour are represented in the mind by schemas, containing rules and information necessary for interpreting situations and guiding behaviour. Form scripts that we follow (or frames). Cultural constraints change with time
  • Semantic: Only certain combinations make sense. Given what is there already, the remaining options are limited. Study of meaning. Semantic constraints rely on the of the situation to control the set of possible actions. Rely upon our knowledge of the system and the world.
  • Logical constraint. There is a logical relationship between the spatial or functional layout of components and the things that they affect. E.g You take something apart, put it back together again, and there's a part left on the table. You know you've made a mistake.
On AA batteries

AA batteries are a bad design. They should make them so it's not possible to put them in in the wrong way. Alternatively, make it so the orientation doesn't matter. AA batteries though suffer from the legacy problem, they were too successful, there are too many batteries in circulation to change the design.

Cultures have their own norms, conventions and standards. Conventions are a form of cultural constraint: Given a situation, things that should be done and things that shouldn't be done

Sometimes cultural constraints are codified into international standards and laws. Traffic laws started as conventions.

Applying Affordances, signifiers, mappings and constraints can simplify encounters with everyday objects. Failure to deploy them leads to issues.

On Norman Doors
  • To operate a door you need to know what side that opens, and the part to manipulate it. There should be a signifier, something for the hand to grasp, telling us where to act. The next step is to determine what operations are permitted.
  • There's something wrong if the door needs a sign. For unlocked doors: Push doors: should have plates or bars. They tell you what to do and where to do it. Offer a visual affordance, a signifier, constrain the action space.
  • Car doors are good on the outside bad on the inside, cabinet doors are often the worst, transit doors are half automatic half manual.
On Switches
  • If you think about the implementation of a switch. You need to know what type of device it controls, and which one it controls (the instance).
  • A usable design starts with careful observation of how the task being supported is actually performed. This is called TASK ANALYSIS. The name for the entire process is called Human Centred Design.
Control Mapping Types
  • Spatial Mapping: orientate based on location
  • Activity Mapping: group by activity (movie night)
  • Device Mapping: grouped by device type

For complex controls, you might need to consider activity mapping, which can be great. Device mapping for complex tasks can be frustrating, as to achieve one thing you may have to control many devices. The challenge with activity mapping is handling exceptions, things that weren't thought of in the original design. You also need to be able to tweak and control with manual overrides.

Constraints, Conventions and Behaviour

Constraints can guide and force behaviour. They’re a key part of Safety engineering:

A forcing function
a physical constraint such that failure at one stage, prevents the next step from happening
Interlocks
Forces operations to take place in proper sequence. Example: Washing machine door doesn’t open unless its drained water
Lock-ins
Keeps an operation active, preventing someone from prematurely stopping it. Example: Warning that makes it hard to leave an unsaved word document
Lockouts
A lockout prevents someone from entering a space that is dangerous, or prevents an event from occurring. Example: The pin in a fire extinguisher that prevents accidental discharge
  • The Forcing Function Tradeoff: Make it too annoying and people will try to disable it. So minimise the nuisance value whilst retaining the safety feature.
  • Conventions can help a user go from perception of an affordance to understanding.
    • A doorknob has a graspability affordance, but the cultural convention is what helps understand we can open doors with them.
  • Conventions are cultural constraints - they can be different across cultures.
  • Going against a convention is difficult. People often object and complain if they have to relearn (e.g. the metric system). Just because something is different doesn't mean it's bad, if we never made changes we could never improve.
  • Consistency in design is virtuous. People are great at transfer learning (lessons learned with one system transfer readily to others). On the whole, consistency is to be followed.
  • Avoid mixed systems, they’re confusing. If there is to be a change - everybody has to change
The merits of the change need to outweigh the difficulties of the change
  • The new way has to be much better than the old.
  • If the new is only slightly better, then it's better to be consistent.
Norman on Plumbing - Controls and Conventions

People care about temperature and rate of flow. BUT water enters the sink from 2 pipes, a hot and a cold. There are a number of ways to configure controls to help the user. You can either have one control that does two things, or two controls that do a single thing. You can also choose to fix something, like flow rate or temperature.

  • When there are two controls there are 4 mapping problems...
    • Which knob controls the hot, which the cold? How do you change the temperature without affecting the flow rate? How do you change the flow without affecting the temperature? Which direction increases the water flow?
    • The simple one control faucet still has 4 mapping problems. What dimension is temperature? What direction is hotter? What dimension if flow? What direction means more?
    • The mapping problems are solved through cultural conventions or constraints. Left should be hot, right should be cold. Clockwise to shut off. Plumbing designers suffer from a lack of consistency though, which makes every new shower a challenge. Can you learn by trail and error? Just by evaluation. Well yes, but temperature can take 5 seconds to adjust, so if you go the wrong way, it can take 10 seconds to get back to a comfortable setting.

They should be simple, but they can violate many design principles... Visible affordances and signifiers, Discoverability and Immediacy of feedback.

  • The principle of desperation: If all else fails, standardise. When everything else fails, design all things the same way, so people only have to learn once. Standards simplify design for everyone, but they tend to hinder future development.
Random: Using sound as a signifier

Objects without speakers can still have great audio feedback (toast popping up, the click of light switch). When materials interact sounds are generated, so it feels natural to get some sound when doing an action. Getting sound design right: Sound can annoy and distract as easily as it can aid. When sound is expected, and silence occurs, then there can be problems.

The sound of a car signifies its presence, we've learnt that you can hear cars coming. Electric car makers are looking at adding the sound back for safety (especially on acceleration, at speed there's often enough tire noise).

Design considerations for electric vehicle sounds. Alert to presence, hint at orientation and speed, don't be annoying, don't contest with other street noises (sirens, horns) and should brands be able to have their own sounds as differentiators or should they be all standardised for consistency.

  • Skeuomorphic: incorporating old familiar ideas into new technologies, even though they no longer play a functional role.

Designing for Errors

Most industrial accidents are logged as human error when design is to blame.
  • 80% of industrial accidents are logged as human error
  • People are often punished for making mistakes
  • Errors occur for many reasons. They're more common when tasks require people to behave in an unnatural way (staying alert, being precise, being accurate, for long periods whilst multitasking).
The paradox of automation
  • Automation can is good for dull predictable tasks, but bad for complex ones. It can fail silently and without warning, as the human is out of the loop. The longer automation performs without error, the more we trust and rely on it.
We should avoid blame and instead look to find the root cause of failures
  • We can’t eliminate human error. So we should design for it. If the system lets you make the error then it is badly designed.
    • Don’t neglect designing things that aren’t on the ‘happy path’
    • When many people have the same issue, another cause should be found.
    • We can't fix problems unless people admit that they exist.
    • Problems can come from the system, procedures or social pressures. People aren't good at precision, monotony and time stress.
  • We should treat all failures the same way. Avoid assigning blame and find the underlying cause. We should workout why it happened and update the design of the product so it can't happen again, or to reduce the impact of it happening again.
Root Cause Analysis and the Five Whys
  • Don't stop at human error - try to understand what caused that error
  • Accidents often have multiple underlying causes - try to find all the major contributors.
  • Think about ways to prevent it AND/OR reduce the impact if it happens again
The Five Whys

When searching for a reason continue to ask ‘why that was the case?’ until you have uncovered the true underlying causes.

  • Example:
    • Why did the plane crash? Because it was in an uncontrolled dive
    • Why didn't the pilot recover from the dive? Pilot failed to initiate a timely recovery
    • Why was that? Because he might have been unconscious (oxygen deprived)
    • Why was that? We don't know, we need to find out.
A culture of error reporting is important: admitting and reporting errors and near misses
  • Admit → investigate → make changes.
  • Toyota Production System encourages reporting and root causing of errors by all
  • Poka-Yoke is an error proofing process (at Toyota)
    • Anything that makes it harder to make an error
    • Making sure the right conditions exist before a step in the process is executed
    • This makes it a preventative form of quality control, since errors are detected and then rectified before they occur.
    • Example: Microwave doesn't work if the door is open
  • NASA Aviation Safety Reporting System: pilots submit semi-anonymous reports about errors they'd seen in others. Neutral party.
The Swiss Cheese Model: How undetected errors lead to accidents (by James Reason)
  • Accidents often have multiple causes. Often errors or latent causes go unnoticed until a number of them line up and allow an accident to happen.
  • Well designed systems are resilient to failure.
  • Attempts to find the cause often fail → There can be multiple causes but also multiple actions that could have prevented the accident. You need to understand the system, and make the highest impact actions to increase reliability.
  • Ways to reduce accidents:
    • add more slices of cheese
    • reduce the number of holes (or make them smaller)
    • add alerts when several holes have lined up.
image
Defining different types of error: Slips and Mistakes
  • Human Error: Any deviation from appropriate behaviour
  • Error: General term for all wrong actions (Two categories of error: slips and mistakes)
  • Deliberate deviations: people knowingly take risks if they feel that guidelines slow them down
  • Routine violations: when noncompliance is so frequent it's ignored.
Slips: Correct goal, but required actions not done properly. Action-Based or Memory-Lapse.
  • You intend to do one action, but end up doing something else. Action performed is not the intended action.
  • Two types of slips:
    • Action-Based: the wrong action is performed.
    • Memory-Lapse: memory fails, so intended action is not done, or results not evaluate
  • Slips are a result of subconscious actions getting waylaid en route. Mistakes are from conscious deliberations. Slips occur more frequently to skilled people than novices.
    • Capture slips: Instead of the desired activity, a more frequently or recently performed one gets done instead. E.g: You start driving and automatically head towards work. They are partial memory lapse errors.
    • Description-similarity slips: Perform the correct action on the wrong object (usually similar to the target object). E.g: friendly fire. Designers should ensure that controls and displays for different purposes are significantly different from one another. In aeroplane cockpits controls for different operations are different shapes and sizes.
    • Memory-Lapse slips: Examples: leaving an original in the copy machine, forgetting a child, losing a pen, leaving a bank card in a machine.
      • Types of Memory Lapse: Not doing all the steps, repeating steps, forgetting the outcome of an action, forgetting the goal or plan. The Leading cause of memory lapse errors is interruptions.
      • Mitigations against memory lapse errors: reduce steps, provide reminders of steps that need to be completed, use a forcing function (cash returned before bank card, pen chained to desk). Often the interruptions are from outside systems, of which the designer has no control
Mistakes: When you have the incorrect goal or plan
  • The wrong goal is established, or the wrong plan is formed. Even if the actions are executed properly they are part of the error, because the actions themselves are inappropriate - they are part of the wrong plan.
  • Types of mistake:
    • Rule-Based: Appropriately diagnosed the situation, but decided on a wrong action.
      • Usually we identify the situation, select the proper rule and then follow it. Rule based errors are more common when normal routine is no longer applicable to the new situation. When IF-THEN doesn't work anymore.
      • Ways in which rule based errors occur
        1. the situation can be mistakenly interpreted, and choose the wrong rule
        2. the correct rule is invoked, but the rule itself is faulty
        3. the correct rule is invoked, but the outcome is incorrectly evaluated
    • Knowledge-Based: The problem is mis-diagnosed because of incomplete knowledge
      • The Situation is novel enough that there are no skill or rules to cover it. A new procedure must be devise. Knowledge based problems can take days to resolve. Things get harder when a false diagnosis is made. In commercial situations the pressure to keep things running is immense. Social pressures can be the dynamics between a junior and senior employee. We need to reward safety.
    • Memory-Lapse: Forgetting at the stages of goals, plans or evaluation.
    • Memory-Lapse Mistakes: Memory failure leads to the forgetting the goal or plan of action.
Mode errors: When a device has states, and controls do different things in different states.
  • Mode errors occur when the operator believes the device is in one mode, when it's in another.
  • It's tempting to use the same control to do multiple things, simplicity can mask complexity of operation though. Interruptions increase mode errors.
  • Mode error is really design error. avoid modes if you can.

Social and Institutional Pressures have a strong influence on behaviour, they lead to misinterpretation, mistakes, and accidents.

  • Novices are more likely to make mistakes than slip. Experts are more likely to make slips.
👉
Try to prevent errors before they occur ELSE detect and correct them when they do occur
Resilience Engineering Design Toolbox: Preventing and detecting errors

Resilience Engineering: Designing systems and procedures so they are able to respond to problems as they arise. Technique: deliberately cause errors in small tests and simulations.

Understand the causes of error. Design to minimise them.

Principles

  • Constrain actions to prevent errors
  • Make operations reversible (UNDO)
Make it harder to do actions which can't be reversed
  • Don't treat the action as an error, try to get the person to complete it properly. Think of the action as an approximation of what is desired
  • Show confirmation and error messages
  • Make the item being acted upon more prominent
  • Check inputs are in a sensible range
  • Minimise slips 1 → Make actions and controls as dissimilar as possible
  • Minimise slips 2 → Eliminate modes (or make them unmissable) to avoid mode error
  • Provide feedback - show the new state - show an undo button
  • Reduce interruptions where possible (consider pausing notifications whilst doing a task)
People easily spot and correct errors when they talk to each other - can we learn anything from that?
  • Clarifying questions can be asked and answered
  • Much harder to achieve with devices, mistakes are often ignored.
  • Make it easier for people to discover errors. Make them easier to correct
Checklists increase accuracy and reduce error → electronic ones can remember state
  • Tips for better checklists:
    • Better when one reads and once executes.
    • Better with a single owner for each item
    • Iterate on them regularly
    • Be wary when enforcing a sequence isn't always possible
On Warning Signals
  • They should attract attention
  • They should deliver information about the event
  • They should have a coordinated response (so a single event doesn't trigger multiple alarms)
  • The boldness of the signal should correspond to the seriousness of event
  • Make the follow-on action easy
  • Too many warning signals can be worse than no signals. People get alert fatigue quickly

Key Design Principles

  • Put information in the environment → Reduce the burden of needing endogenous knowledge.
  • Allow for efficient operations when people have learned the requirements
  • Use environmental knowledge to make it easier for non-experts. This will help infrequent journeys and infrequent users
  • Leverage natural and artificial constraints: physical, logical, semantic and cultural.
  • Exploit the power of forcing functions and natural mappings
  • Bridge the gulf of execution and the gulf of evaluation.
  • Make things visible, both for executions and evaluation
  • One the execution side, provide the feedforward information: make the options readily available.
  • On the evaluation side: make the results of each action apparent.
  • Make it possible to determine the system's status readily, easily, accurately and in a form consistent with the person's goals, plans, and expectations.
  • Embrace errors. Seek to understand their causes and ensure they don't happen again. Re-design don't reprimand

Design Thinking

Design thinking: places equal emphasis on problem definition (solving the right problem) and solution definition. Allows time to diverge and think widely before converging on a problem or solution.
  • Norman’s consulting law: never solve the problem you're asked to solve
    • Often we focus on the wrong problem. It takes work to discover the real problem
    • The secret to success in design is to understand what the real problem is.
HCD Human Centered Design and it’s four stages
  • Solving the right problem. Focusing on needs of individuals and jobs to be done. Building products that are understandable, usable and enjoyable.
  • HCD has a four stage model (arranged as a cycle)
    1. Observation.
      • The best way to understand the nature of the problem is direct observation and interaction with future users
      • Go to them, observe their actions and attempt to understand their true needs, motives and interests.
      • Applied ethnography: faster than traditional research and has an emphasis on building products to help. Studies focus on the activities that people do, and how they accomplish tasks.
      • Focus on the intended customer in the intended market. If you're targeting a sub-culture, then you need to study that group.
    2. Idea Generation.
      • Generating potential solutions (ideation)
      • Used in both phases of the double diamond.
      • Inducing creativity while brainstorming. Generate as many ideas as possible and avoid fixating on one or two too early.
      • Be creative without regard for constraints. Hold back from criticising ideas, even your own. Avoid premature dismissal. Question everything, stupid questions are great. Question the obvious. Dig deeper and profound insight can follow.
    3. Prototyping.
      • The only way to know if an idea is reasonable is to test it.
      • Build something quickly, get feedback as soon as possible.
      • Using tricks (like mimicking a backed system a.k.a. the Wizard of Oz technique) you can valuable insight before building a working solution. You can even use a competitor product as a prototype. They can be used in both phases of the double diamond, to help define the problem and the solution.
    4. Testing.
      • Gather a small group of your target population. Have them use the prototypes in a realistic way, in a realistic environment. Take video recordings. Ask them to think out loud or explain their thought process after the fact.
      • Study 5 people in each round, review the results and refine the process before going again. Don't do more than 5 without iterating on the script or product.
      • Product Andrew: My model is test until you're bored, and they become predictable.
      • CE marking and usability testing may require more participants and a more stable product and script.
    On iteration and the HCD cycle
    • The speed at which a team can learn determines its success. We want to be continually refining and enhancing the product, by rapid prototyping and testing. Fail frequently, fail fast. Reframe failure as learning experience, If everything works, little is learned.
    • Repeated observation and study can help get the requirements right (which is the hardest part of design). With each cycle ideas become clearer, specifications better defined, prototypes become more effective.
    • People aren't great at identifying and communicating their higher level needs. They often don't grasp fully the number of deviations and special cases they encounter.
The Double Diamond Design Process
  • Allows freedom to explore the problem and solution space. It doesn't guarantee success, you may converge on the wrong solution and have to repeat the process. Future iterations will be much faster though
    • When using the double diamond approach, visible progress only comes towards the end of the project, which can be unsettling.
  • The Design Problem: satisfying a number of constraints and concerns: shape, form, cost, reliability, effectiveness, understandability, usability, aesthetics, proud to own, joy to use
  • Phase 1: Problem Definition: Solving the right problem
    • Discovery: Diverging to examine fundamental issues, questioning the problem
    • Definition: Converging upon a single problem statement
  • Phase 2: Solution Proposal: Solving the problem in the right way
    • Develop: Diverging to explore a wide variety of potential solutions
    • Deliver: Converging on a proposed solution
Depth and Breadth Research Tradeoff (Design vs Market)
  • Design research → collecting deep insight on a small number of people. What people need and how they use it. Qualitative and observational.
  • Market research → collecting shallow insight on a large number of people. What people buy, how they make purchasing decisions. Quantitative & Surveys.
    • insight is shallow because what people say they do and what they do are different
Individuals vary - activities are constant (Activity Centered vs HCD)
  • HCD is a focus on the individuals, ensuring the product fits real needs and is understandable, useful and joyful to use.
  • BUT people vary widely - if you're making a product for the world that's a problem.
  • A focus on activity (vs people) helps define a product and its structure.
  • People have activities in common - and are willing to learn things that are essential to their lives
  • Task: A lower-level component of an activity
  • Activity: A collected set of tasks, all performed together to meet some goal.
  • Align your design to an activity gives the user motivation/ a reason to use your product
    • The complexity needs to be proportional to the task
What about big / huge scale projects?
  • Iterative methods are best suited for early design and smaller projects
  • It doesn't scale to 100's or 1000's of people.
    • Some waterfall methods have stage gates.
      • Stage gates can be useful if you don't waste time prepping for them. The best methods contain a mixture of iterations and stage gates.
  • The hard part of larger projects is communicating, organising and synchronising people across different departments. The longer the length of the project, the harder it gets (stakeholders and requirements shift). Often knowledge is implicit, in people's heads
In theory, there is no difference between theory and practice. In practice, there is
The day a product-development process starts, it's behind schedule and above budget (Don Norman's law of product development)
The Design Challenge
  • The fundamental principles of designing for people are the same across all domains. The design process must address numerous constraints.
  • Products have conflicting requirements: put work in to understand them, get representation for each in the room
  • Competing requirements is what makes design hard. It's easier to get to a viable solution if you have a team with representation from a mix of disciplines, and they work together as a single unit.
There's no average person
  • There is no such thing as the average person
  • You’ll exclude people even when you design for the percentiles
  • Design for interests and skill levels. Don't be trapped by overly general, inaccurate stereotypes.
The Stigma Problem

Most people don't want to advertise their imperfections or admit it to themselves.

  • The OXO peeler designed for those with arthritis was a better peeler for everyone (the curb-cut effect).
Flexibility helps with inclusivity and accessibility.
  • It’s often the best way to design for everyone’s needs. Allow people to adjust things. Fixed solutions will always fail for some, flexible solutions have a chance.
Complexity is OK. Confusion is bad.
  • Complexity is often essential. Don't make something more complex than it needs to be, tame complexity with a good conceptual model. Avoid confusion though
Technology Standardisation
  • Standardisation is a type of cultural constraint. Provides a breakthrough in usability!
    • Benefits from technological advances are either brought about by the technology itself, or by standardisation.
  • Establishing Standards is laborious. Easy in a company, much harder in an industry or international body. Standards are often a compromise and sometimes take so long to agree they can be irrelevant at launch. Standards are necessary, and help with interoperability.
Sometimes, you’ll want to deliberately make things difficult
  • Hide critical components
  • Use unnatural mappings
  • Make the actions physically difficult to do
  • Require precise timing and physical manipulation
  • Do not give any feedback
  • Use unnatural mappings

Design in the World of Business

On Competition
  • There are a limited number of ways to compete: price, features, quality, aesthetics.
  • The speed of product development is an important factor for companies in competition.
  • Time pressure sometimes crowds out iteration and exploration.
    • Often things are independently discovered or developed at the same time (Zeitgeist: spirit of time)
    • Concerns that companies have:
      • Speed of development
      • Cutting costs
      • The competition
      • The need to satisfy multiple customers
    • The question becomes: How do you best focus and deploy your resources?
On Product Feature Creep
  • Product teams have a tendency to add more features than they remove. Users and stakeholders rarely ask for features to be removed
  • Pressure to add features comes from
    • Requests from existing customers
    • Competing products adding them
    • To acquire new customers (satisfied customers, rate of acquisition slows as you approach saturation)
    • Quest of product teams to expand and provide more value
    • To acquire new customers (satisfied customers, rate of acquisition slows as you approach saturation)

Continually adding features makes it hard to:

  • Understand & communicate
  • Use
  • Maintain
  • Improve
Competition-driven design

Is when you're matching the features of your competitor, leaving the customer no reason to choose either product. A better strategy might be to

  • Make improvements where you're already strong
  • Keep the product consistent, coherent and understandable
  • Meet the needs of the people who use the product

Quality comes from continued focus on the customer.

On the Impact of Technology on Industries
  • New technology unlocks new possibilities, entire industries can be remade following a new application of technology.
  • People and culture change at a slower rate to technology. It can take decades for a technology to be adopted. However, the rate at which technologies are adopted is increasing.
  • Most radical ideas fail. Most large companies don't tolerate failure well. They tend to be more conservative in adopting new technologies. Startups are more able to take risk. Most fail, but risk is asymmetrical as the upside is much larger. Startups are even more risky that we perceive, due to survivorship bias, Timing must be right, most companies that are early fail. Heavy momentum of legacy inhibits change.
QWERTY vs DVORAK
  • Originally chosen to benefit the mechanics of a typewriter. QWERTY layout helped the mechanical type bars approach one another at large angles, minimising the chance of collision. Otherwise it was designed for speed, attempting to make common letter pairs be split across the right and left hand.
  • The DVORAK layout is slightly faster than QWERTY. However, the switching costs are high, and the difference isn't large enough to justify the time investment for most people
  • Chord keyboards are faster still but aren't suitable for beginners and they require more training. They're only used in rare circumstances like transcribing court documents.
Incremental vs Radical Innovation.
  • Most innovation comes from incremental improvements to existing products, continual testing and refinement. The gains per cycle are often small, but the cumulative effect is dramatic (e.g iPhone, cars).
  • Radical innovations that result in step changes or paradigm shifts are much less frequent. It may still take decades for a radical innovation to be adopted and for its transformational effect on industry to be realised.
The Design of Everyday things
  • Human evolution is measured on a scale of thousands of years. Human culture takes decades or centuries to change.
  • The world has become reliant on technology. Socrates thought books would mean people would stop thinking for themselves and debating.
  • Humans and machines are both more powerful when augmenting each other. The power of an unaided mind is highly overrated, it is things that make us smart.
  • Technology has freed us from the trivial. We don't have to find food, we don't have to remember numbers or directions, we're free to focus on other things.
Think about the environmental impact of what you produce.
  • Production
  • Emissions while in use
  • Extending the duration of use
  • Repairability and right to repair
  • End of life

Do incremental generations encourage over consumption? Maybe a subscription model would incentivise more durable goods

Design is successful only if the product is successful (purchased, used and enjoyed). Design should pay attention to the total experience and the total lifecycle. Design should be concerned with function, usability and understandability.

Individuals and small teams have better access to information, tools and platforms than ever before, maybe they'll support a renaissance of talent.