# User-centered vs. human-centered > Jef Raskin distinguishes between user-centered design and human-centered design, and emphasizes the importance of human-centered design. In this article, I want to clarify the distinction between the two terms according to Jef Raskin's view, and explain the importance of human-centered design. distinguishes between [user-centered design](https://wiki.g15e.com/en/pages/User-centered%20design.txt) and [human-centered design](https://wiki.g15e.com/pages/Human-centered%20design.txt), and emphasizes the importance of human-centered design. In this article, I want to clarify the distinction between the two terms according to Jef Raskin's view, and explain the importance of human-centered design. ## Background: Donald Norman and Jef Raskin (1935) majored in at . Afterward, he completed a master's degree in electrical engineering and a PhD in psychology at , then served as a researcher and lecturer at Harvard University's The Center for Cognitive Studies. After that, from the late <1960s> to the early <1990s>, for about 20 years, he worked as a professor in the psychology department at . From <1993>, he worked at (then called "Apple Computer"), and his title was User Experience Architect. In <1998>, he founded the together with . He is especially well known for books such as and . studied mathematics, physics, philosophy, and music at Stony Brook University, then received a master's degree in computer science from . After that, he completed the PhD coursework in computer science at the same university, but reportedly did not receive the degree because of a conflict with his advisor. Supposedly, he directly implemented a method that his advisor had claimed was impossible to implement, and demonstrated it. In a later interview, he said that at the time he felt like Galileo (source: [A Conversation with Jef Raskin](https://ubiquity.acm.org/article.cfm?id=941396)). Until <1974>, he taught art, photography, computer science, visual arts, and more at UCSD. In <1978>, he joined , and he was the person who started the Macintosh project. However, in <1981>, he left the company after a clash with , and the design of the Macintosh that was actually released later was quite different from Raskin's ideas. In <1987>, he released the Canon Cat, which faithfully embodied his own design principles, but unfortunately it did not achieve commercial success. Both men were at in the early <1970s>. They first met there, and continued to interact actively and enjoy arguing about design (source: [In Appreciation of Jef Raskin](https://jnd.org/in_appreciation_of_jef_raskin/)). Unfortunately, Jef Raskin passed away from in <2005>. ## Donald Norman and user-centered design In the field of software design, what is called [user-centered design](https://wiki.g15e.com/en/pages/User-centered%20design.txt) is a concept contrasted with developer-centered or computer-centered design (or the absence of design). It is an approach that tries to design around the user rather than the developer (in the broad sense meaning everyone involved in development) or the computer. Similar ideas existed even earlier. For example, 's <1955> book is a widely read classic in industrial design. But until the early <1980s>, in the software field these ideas were hardly known or practiced. (This was the era when pioneering research was being done at , but it was not yet widely known to the public. would not be published until 1983.) In software, this movement began to emerge after the mid-1980s. The person who widely popularized the concept of user-centered design in this field was Donald Norman, thanks to his <1986> book and his <1988> book . Although these books were not the first use of the term user-centered design, their publication marked the point at which the term and concept were introduced in earnest into the software field. User-centered design emphasizes two big things. - **First**, one must understand universal human physical/cognitive characteristics. This requires an understanding of ergonomics, cognitive ergonomics, cognitive psychology, and so on. - **Second**, one must identify the characteristics and needs of the core group of users who will actually use the software. For this, various methods, participatory design, and the like are needed. The first edition of mostly deals with design principles related to universal human physical/cognitive characteristics, but in the section "Designers are not typical users" in Chapter 6, it mentions the importance of user research. > The only way to find out whether users can use a product well is to test it with users. By users here, I mean people similar to those who would actually buy this product. Moreover, interaction between designers and users should begin at a very early stage of the design process. Once the process has progressed even a little, it may become too late to apply fundamental changes. Designer [Dan Saffer](https://wiki.g15e.com/pages/Dan%20Saffer.txt) says this in his book [Designing for Interaction](https://wiki.g15e.com/pages/Designing%20for%20interaction.txt). > Designers who follow user-centered design chant, like a mantra, "I am not the user." It is a phrase that reveals the essence of user-centered design well. ## Jef Raskin and human-centered design published [The Humane Interface](https://wiki.g15e.com/pages/The%20humane%20interface.txt) in <2000>, released in Korean as "Human-Centered Interface." A literal translation of the original title would be "the humane interface." I find the translation practice of arbitrarily changing book titles regrettable. In this book, he proposes the expression 'human-centered design' as a concept contrasted with the existing 'user-centered design,' which emphasizes the importance of user participation. The relevant passage reads like this. > Although users' task-related needs may vary enormously, your user population shares many psychological traits. Before exploring design solutions or considering differences among individual users, you can greatly shorten the design process by making good use of the psychological commonalities possessed by all humans. Only after doing this preliminary work can the interface designer reflect the characteristics of individual users and user groups in the design and satisfy diverse needs. This crucial first step—ensuring that interface design follows universal psychological facts—is often habitually omitted from the design process. Back when I was doing a study group with this book (probably around <2009>), I discovered that this important part had been bizarrely mistranslated in the Korean translation, though I don't know what it's like now. If I rewrite the quotation according to how I understood it, it is this. It is a point that all three of the following are important: identifying the traits shared by all humans, identifying the traits shared by the target user group, and identifying the differing traits possessed by individual users within that target user group. But among these, the effort to identify and reflect the very first one—the traits shared by all humans—is often omitted in current user-centered design practice. Norman's writings from the mid-1980s emphasized two things: understand human universals, and evaluate with potential users. About 15 years later, Raskin, worried that the latter of these two was being overemphasized to the point that the former was often not sufficiently considered, emphasized the term 'human-centered.' ## What specifically is different? The concept that Jef Raskin calls human-centered design is, in principle, included within the existing concept of user-centered design, and in practice the two are usually used interchangeably. For example, regarding nearly the same concept, the [U.S. Department of Health and Human Services](https://www.usability.gov/what-and-why/user-centered-design.html) uses the expression 'user-centered design,' while [the international standard ISO 9241-210:2019](https://www.iso.org/standard/77520.html) uses the expression 'human-centered design.' And the U.S. Department of Health and Human Services document refers to the international standard document. In fact, even Norman, who in the first edition of used only the term 'user-centered design,' began using terms such as human-centered design or human-centered development procedures in the preface to the 2002 revised edition. In the <2013> revised and expanded edition, every instance of the term user-centered design was replaced by human-centered design. (For reference, the 2013 edition is close to an almost completely rewritten book, synthesizing the contents of his later works such as and .) In the 2013 edition, Norman defines human-centered design like this. In a way, it does not seem very different from Jef Raskin's claim. > (Human-centered design is) an approach to design that puts human needs, capabilities, and behavior first, then designs to accommodate them. Good design starts with an understanding of psychology and technology. (rest omitted) Even so, I still think Raskin's argument has meaning. First, there can always be a gap between the principled meaning of a term and the practice that is actually carried out. I think one reason Raskin deliberately used and emphasized the term human-centered design in contrast to user-centered design is precisely this gap. Many design methodologies and processes that claim to be user-centered seem to have a tendency to focus excessively on user research or usability testing, without making sufficient effort to faithfully apply an understanding of human universals. Second, there is also a difference in concrete orientation. For example, Raskin argues that before considering 'the characteristics of individual users and user groups,' we should first consider 'universal human characteristics,' and accordingly he is wary of designs that offer separate interfaces for novices and experts, or provide customization features for individual users. Norman, by contrast, points out the problems with the concept of the 'average user.' What would we need to do differently in order to practice human-centered design more faithfully? Three things come to mind. 1. Understand interface guidelines more deeply and follow them well 2. Reconsider what users or clients mean when they say something is 'intuitive and natural' or 'unintuitive and unnatural' 3. Understand humans more deeply Let's look at them one by one. ## First: interface guidelines 'Interface guidelines,' '[heuristic evaluation](https://wiki.g15e.com/pages/Heuristic%20evaluation.txt) items,' and so on were made to be broadly usable by taking universal human psychological characteristics into account. Therefore, simply following guidelines faithfully can prevent a great many problems even before usability testing. Put differently, guidelines can impose various constraints on the designer, and these constraints have positive aspects. - First, because they greatly reduce the range of possibilities the designer must consider (the search space of design solutions), they save time. - Second, creativity is better exercised under constraints than in a situation with no constraints at all. Let's take an example. He emphasizes **the fact that human attention is an extremely limited resource** (which is why [social media compete to monopolize users' attention](https://wiki.g15e.com/pages/Social%20Dilemma%20(film.txt))). In particular, the human is singular and does not change no matter how much training one undergoes. For example, speaking to a novice driver while driving can be especially dangerous because attention focused on 'driving' is shifted to 'conversation.' If repeated practice forms a habit for a task, that task becomes capable of so-called 'automatic processing,' and therefore no longer requires attention; it is not that the number of loci of attention increases. Based on the singularity of the locus of attention, one can propose guidelines like the following. - **Help habit formation**: From the standpoint of the singularity of the locus of attention, a good interface should be one that helps form natural [habits](https://wiki.g15e.com/pages/Habit%20formation.txt). That is because only then can attention be focused on the task itself rather than on the interface for performing the task. - **Eliminate modes**: One factor that interferes with habit formation is the '[mode](https://wiki.g15e.com/pages/Mode%20(interface.txt))' of an interface. That is because the system's response to the same gesture can differ depending on what the current mode is. To use such an interface, one must always check what the current mode is before performing a gesture (mouse click, touch, keyboard input, etc.). For this reason, he pursues a **modeless interface**. Donald Norman presents three criteria for reducing errors caused by modes, whereas Jef Raskin argues that since cannot be eliminated, the modes themselves must be removed. - **Satisfy novices and experts at the same time with a single interface**: Before being novices or experts, users are human. Before distinguishing between novices and experts and thinking about separate interfaces for them, we should consider whether there is a single interface that can satisfy both at once. ## Second: reconsider the expression 'intuitive and natural' Another memorable passage in Raskin's book is his view on the expression 'intuitive and natural' (6-1. Intuitive and natural interfaces). He notes that when users say 'the interface is intuitive,' what they mean in the vast majority of cases is 'familiarity,' 'becoming familiar because they learned it before,' and so on. The expression 'natural' is also often used similarly. In this context, he says the following. > The belief that interfaces can be intuitive and natural often hinders the improvement of interface design…. A client wants to make a product sufficiently superior to that of its competitors. But to be superior, it must be different from what already exists. (Usually, the greater the improvement, the greater the difference from the existing way.) Therefore, if 'intuitive' is used to mean familiar, a superior design cannot be intuitive. What the client wants is probably something that achieves major improvement while remaining almost identical to the existing way, but such a hope is possible only when the existing way has major defects that can be fixed with only small modifications. Designers need to distinguish rigorously among the following three concepts in relation to the usability of an interface, and when a user says something is 'intuitive' or 'natural,' carefully analyze which of these they mean. - **Intuitiveness**: You can tell how to use it the very first time you see it, without trial and error. - **Easy to learn**: With a little explanation or some trial and error, you can quickly figure out how to use it. - **Ease of use**: Once you become familiar with it, it is convenient to use repeatedly. According to this distinction, the mouse is not intuitive, but it is easy to learn and hard to use. - **Intuitiveness**: According to records observing people who saw a mouse for the first time, one can see various forms of trial and error: people trying to do something while holding the mouse in their hand (the shape of the mouse clearly expresses the [affordance](https://wiki.g15e.com/pages/Affordance.txt) of 'holding it in the hand'), people trying to turn it upside down and use it like a trackball (older mice almost always had a ball), people bringing the mouse to their mouth and trying to issue voice commands into it like a microphone (this was not an actual observed case, but a scene from an episode of Star Trek), and so on. Therefore, some aspects of the mouse are not intuitive. When you move an object placed on the desk (the mouse) horizontally, a symbol (the mouse cursor) moves on a vertically oriented digital screen; this is also not a very natural mapping. - **Easy to learn**: It is very easy to write down or verbally explain the basic way to use a mouse. After hearing a short explanation, you can use it immediately. In that sense, the mouse has high learnability. - **Ease of use**: No matter how long you use a mouse, it never becomes as efficient as a keyboard. For example, unlike a keyboard, which you can use with your eyes closed, it is extremely difficult to send the mouse cursor to a desired location with your eyes closed. To use a mouse with sufficient precision requires deliberate and prolonged effort. Like a pro gamer. That is why tools people use repeatedly ([Excel](https://wiki.g15e.com/pages/Microsoft%20Excel.txt), Photoshop, programming editors, and so on) have countless keyboard shortcuts, and the longer someone uses such a tool in everyday life, the more they tend to prefer shortcuts. Shortcuts have low learnability but high ease of use. The best interface is one that is intuitive and also high in both learnability and ease of use. But there are cases where, unavoidably, all three cannot be satisfied at once, and then one must think about which characteristics to sacrifice and by how much. No customer would want an interface for withdrawing cash that lets them do so in only 3 seconds if they first have to read and master a manual for several hours. Ordinary customers want an interface that requires no prior learning even if it takes 30 seconds. On the other hand, a bank clerk is highly likely to want software that lets them process customer work as fast as possible, even if it requires considerable advance learning. Because they have to use it constantly, countless times every day. From this passage alone, it may look as though he is arguing that intuitive and natural interfaces do not exist at all, but I do not necessarily think so. For example, if you move the mouse to the left, the cursor on the screen also moves to the left, and this can be said to be more 'natural' than another kind of mapping (for example, moving the mouse to the left causes the cursor on the screen to move upward). Donald Norman calls this [natural mapping](https://wiki.g15e.com/pages/Natural%20mapping.txt). Or, compared with entering commands through the keyboard (for example, typing the command "rm" to delete a file), issuing commands by directly manipulating objects on the screen with a mouse (for example, dragging a file object into the trash) is more natural. Meanwhile, compared with the conventional way of manipulating a cursor on the screen using a mouse placed on a desk, the current way of directly touching the screen can be said to be more natural. (Mouse-based GUI interfaces have been called '', but now that touch screens have become ordinary and devices are spreading, the term has become somewhat inappropriate.) To summarize, it is worth keeping in mind that when a user or client says "an intuitive and natural interface," we need to carefully analyze whether they specifically mean intuitiveness, learnability, or ease of use; that if we unavoidably have to choose among them, we need to think about which characteristics to keep and discard depending on the context of use; and that the more substantially improved an interface is compared with the existing method, the less familiar it will be, and therefore the more likely it is to be judged "not intuitive." At the same time, in communication with clients, forcing them by saying "this way is better, so accept it" is violent. I think a great deal of thought and strategy is also needed about how to propose things better, and how to persuade or negotiate. ## Third: understand humans more deeply We should know the guidelines well and follow them carefully, but that alone has many limitations. UI designer says this in . > Applying user interface guidelines is not nearly as straightforward as following a cookbook. Design rules often mention goals that should be achieved rather than telling you exactly what to do. Guidelines also tend to be fairly generalized in order to increase their range of applicability, and that in turn means they leave a good deal of room for interpretation when you try to apply them to a specific situation. To interpret guidelines appropriately for a given task and use them well, we as designers need to deeply understand why the guidelines were made that way in the first place. The field that studies human cognitive characteristics for the purpose of applying them to design is called cognitive ergonomics, [human factors](https://wiki.g15e.com/pages/Ergonomics.txt), cognitive engineering, and so on. (Jef Raskin proposed the term cognetics, but it does not seem to be widely used.) The more deeply we understand universal human cognitive characteristics, the better we can apply guidelines, and the more help we can receive even when thinking through specific situations not mentioned in the guidelines. Raskin actively recommends making use of various quantitative evaluation methods such as [GOMS](https://wiki.g15e.com/pages/GOMS.txt), , , and [Fitts's law](https://wiki.g15e.com/pages/Fitts%20law.txt). Quantitative evaluation itself has value, but also because to understand and apply quantitative evaluation methods, one must necessarily understand the universal human psychological characteristics that underlie them. > The advantage of (quantitative methodologies) is that, in the process of understanding why these methods work, they naturally help lead you to an understanding of how humans interact with machines. Can a deeper understanding of human cognitive characteristics really give concrete help in design? Isn't it merely an academic curiosity disconnected from practice? ## Example: eye movements Rather than broadly enumerating universal characteristics of human cognition, I think it would be good to consider a concrete example. Let's think about visual perception, and especially eye movements. Understanding visual perception can be valuable in various fields such as , , and . The time it takes to manipulate a mouse by hand and move it to a desired point can be roughly estimated according to Fitts's law. The closer the target is, and the larger the target is (taking movement direction into account), the faster it is. So in mouse-based interfaces, the elements that need to be clicked most frequently are placed in the corners of the screen (since the mouse cannot go beyond the screen, the target size is effectively infinite), or are designed to appear near the mouse (distance to the target is almost 0). What about the movement of gaze? Does it follow Fitts's law similarly to mouse movement? At first glance it seems like it would, and actual measurements do vaguely look that way, but in fact there is a big difference from mouse movement. That is because the movement of the eye and the movement of the hand are different from each other. [Saccadic eye movements](https://wiki.g15e.com/pages/Saccadic%20eye%20movements.txt), unlike hand movements, are 'launched' instantaneously from one point to another, and because they are unrelated to muscular hand movement and do not involve sensory-motor coordination, **if the distance of the gaze shift (the size of the rotation angle) is not extremely large, almost the same amount of time is required regardless of distance**. The following video shows well the difference between saccadic movement, in which the eye is instantaneously 'launched' in little bursts, and the smooth pursuit movement that follows the movement of a particular object. [![](https://blog.kakaocdn.net/dn/WMlqI/btqRAj9AQII/N7dwXyYgktupGiqaUrs0LK/img.gif)](https://hackaday.com/2018/04/11/redirected-walking-in-vr-done-via-exploit-of-eyeballs/) If the distance is large, a second saccadic movement is needed after the first movement to correct the position more precisely, so more time is required; because of this, in some situations results can be observed that appear similar to Fitts's law.[^1] If we take into account such well-known facts as that visual resolution drops sharply outside the of the retina, that the visual angle of the fovea is about 2 degrees, that the distance between a mobile device and the eye is about 30–50 cm, that a saccadic eye movement takes roughly 0.02 to 0.2 seconds, that visual input is almost blocked while a saccadic movement is occurring, that if the movement distance is not excessively large there is almost no difference in movement time, and that peripheral vision has low resolution and poor color discrimination but perceives motion well, then many 'constraints' arise in the design space. Constraint is often used with a negative meaning, but in situations where the designer's potential search space is excessively broad, constraints are a good thing because they can reduce unnecessary searching (that is, design trial and error). ## Application: interface design for a first-person air combat game Where specifically could such constraints be applied? Let us assume a situation where we are designing the interface for a first-person air combat game for mobile devices. Even before meeting potential users, we can know more facts in advance than one might think. - Since the player does not know where enemy aircraft will appear, they will mainly look at the center area of the screen, but they must also scan the screen overall. - Since the visual angle of the is about 2 degrees, assuming the distance between the mobile device screen and the eye is 50 cm, the area that can be seen clearly at a given moment is less than about 2 cm of the entire screen. Users feel as though they are always seeing the whole screen clearly, but in fact they are not. If the distance between the screen and the eye is not 50 cm but 30 cm, the area captured by the fovea narrows to about 1 cm. - Since the screen size of a mobile device is roughly 15 cm diagonally, the eye has no choice but to make extremely busy saccadic movements, whether the player is conscious of it or not. - Because peripheral vision perceives movement well, even while the player is staring at the center of the screen, if new information appears in the periphery, the player can recognize that information relatively easily and a saccadic movement will occur toward it. But considering that saccadic movements are already happening busily, and that visual input is almost blocked while a saccadic movement is occurring, there is a possibility that even if some important indicator appears in one corner of the screen (for example, a radar at the upper right), it may be missed ("Huh, when did that indicator appear?"). - Therefore, if important information is shown in the peripheral area of the screen, adding an animation effect of at least about 0.2 seconds when the information is updated can greatly reduce the possibility that the user will miss it. - Information that must be looked at together does not necessarily have to be extremely close together. Even if there is some distance between them, eye movement speed is almost the same (unless an additional corrective saccadic movement occurs). Of course, considering the context of game design, one must think about whether the process of training oneself not to miss suddenly appearing important indicators is itself part of the or not. ## The importance of user research and usability testing Because interface guidelines must be general-purpose, there is an aspect in which they are intentionally written loosely. Theories about human visual perception and cognition still have many shortcomings, and we also cannot completely predict how users will use a designed artifact. In addition, while guidelines and theory can help us solve a given problem well (verification), they cannot tell us whether the problem we are trying to solve is actually a problem users experience in the first place (validation). In other words, if we're not careful, we may end up making a strange design that **solves well a problem that did not need solving**. (See: [Verification and validation](https://wiki.g15e.com/pages/Verification%20and%20validation.txt)) Therefore, and usability testing are still necessary. However, a great deal of thought is also needed about how to design the tasks, how to ask interview questions, how to analyze and understand users' words and actions, and so on. See also: [Distorted user-centered design](https://wiki.g15e.com/en/pages/Distorted%20user-centered%20design.txt). ## Summary and conclusion User-centered design emphasizes two big things. - **First**, design should reflect universal human physical/cognitive characteristics. - **Second**, it should reflect the characteristics and needs of the target user group that will actually use the software. Jef Raskin uses the term human-centered design rather than user-centered design in the context of pointing out that, although the former of these is tremendously important, it is often not considered sufficiently compared with the latter. The better we understand universal human cognitive characteristics, the more we are able to discover and use good constraints that reduce the designer's trial and error and can even increase creativity. However, **he is not arguing that if one properly understands universal human characteristics, usability testing becomes unnecessary.** He means that, just as efforts to understand the target user group are important, efforts to understand and reflect universal human characteristics are important as well. ## Additional note: Which human is the 'universal human'? In ergonomics, the 'human body' often means 'the body of the average male,' and many of the standards used over the past several decades are [tailored to men](https://www.theguardian.com/lifeandstyle/2019/feb/23/truth-world-built-for-men-car-crashes). In the traditional sense, '' has been close not to the usability of all users but to the usability of the 'non-disabled.' Even if usability is high, if accessibility is low, some people cannot use it. The '[accessibility](https://wiki.g15e.com/pages/Accessibility.txt)' just mentioned means barely 'a level of usability sufficient to gain access,' and securing accessibility does not in itself mean good usability. (Fortunately, the concepts of usability and accessibility are gradually expanding as these limitations are improved.) Then what should 'universal human characteristics' mean in human-centered design? I think that when trying to practice human-centered design, reflection on what exactly 'the human universal' is, and what 'normal' or 'general' means, must accompany it as well. ## Footnotes [^1]: https://research.fb.com/publications/an-explanation-of-fitts-law-like-performance-in-gaze-based-selection-tasks-using-a-psychophysics-approach/