- 1 Slides
- 2 Readings
- 3 Additional Resources
- 4 Reading Critiques
- 4.1 Krithika Ganesh 13:35:48 9/6/2017
- 4.2 Mingzhi Yu 14:30:12 9/6/2017
- 4.3 Jonathan Albert 17:36:07 9/6/2017
- 4.4 Muneeb Alvi 19:18:56 9/6/2017
- 4.5 Spencer Gray 20:52:06 9/6/2017
- 4.6 Tahereh Arabghalizi 21:08:32 9/6/2017
- 4.7 Amanda Crawford 21:26:52 9/6/2017
- 4.8 Xiaoting Li 21:53:05 9/6/2017
- 4.9 Ahmed Magooda 22:35:32 9/6/2017
- 4.10 Sanchayan Sarkar 23:42:47 9/6/2017
- 4.11 Charles Smith 0:34:59 9/7/2017
- 4.12 Xingtian Dong 1:20:57 9/7/2017
- 4.13 Akhil Yendluri 1:37:46 9/7/2017
- 4.14 Yuhuan Jiang 1:38:45 9/7/2017
- 4.15 injungkim 6:22:49 9/7/2017
- 4.16 Kadie Clancy 8:11:42 9/7/2017
- 4.17 Ronian Zhang 6:23:49 9/12/2017
- Direct Manipulation Interfaces Hutchins, E., Hollan, J., and Norman, D., Human-Computer Interaction, 1(4), 1985
- User Technology: From Pointing to Pondering, Card, S., Moran, T., ACM Conference on the History of Personal Workstations, 1986
An application of GOMS in a modern context (eye tracking and text input)
- Chinese Input with Keyboard and Eye Tracking - An Anatomical Study Wang, J., Zhai, S. and Su, H., In Proc of CHI 2001, pp349-356
- The Mouse, the Demo, and the Big Idea, Wendy Ju, HCI Remixed: Reflections on Works That Have Influenced the HCI Community, The MIT Press (January 31, 2008)
Krithika Ganesh 13:35:48 9/6/2017
DIRECT MANIPULATION INTERFACES: This paper highlights the importance of "feeling of directness" when designing an interface by defining 2 aspects to directness: distance and engagement, where distance can be further classified into semantic and articulatory distance and engagement is classified into conversation and model-world metaphor. The paper is important as it helps us to understand what went through the minds of the pioneers who designed the early interfaces such as Sketchpad and how in the long run we have overcome hardware limitations to create highly interactive devices. This paper stresses on how high level languages can bridge the gulf between the user's goals and the system requirements. Also, tools like Android studio which provides us a platform to develop apps follows the drag and drop paradigm mentioned in the paper. In today's world almost everyone owns a smart phone which is designed using android OS and high level languages like Java. I did not find any new results, in fact most of the examples explained in the paper which required input were just mouse clicks or keyboard input. If the paper was written in recent times, it may have taken into consideration inputs like hand gesture detection and finger print sensor. The paper in context of conversation metaphor mentions that 'the implicit role of interface as an intermediary to a hidden world denies the user direct engagement' which I disagree. Taking an example of the Kahoot app (coming under the conversation metaphor) that we used in class, we had to select shapes which represent the choices on the board. These shapes were a representative of the answer choices and were not representing the real-world objects (as in the model world metaphor which may have represented a thumbs up for yes), irrespective of that, students in the class were engaged and felt involved in the class while using the app which is a live example that not all interfaces need to follow the model world metaphor. One point which I did really like in this paper is that although the author throughout the paper is in favor of direct manipulation, in the end he agrees that too much of it will curb innovation leading to missing out on development of new technology.
USER TECHNOLOGY FROM POINTING TO PONDERING. The paper talks about the interaction of computers and humans what might be called a 'user technology'. Going from pointing to pondering, it explains about physical interface, cognitive interface, conceptual interface and task interface. While explaining the physical interface, they explain that it’s hard to design a device faster than a mouse (Fitt's law), limiting factor being eye-hand coordination. In recent times, touch screen devices are used, hence it would be interesting to benchmark it with the mouse and other devices and check how optimal it performs. Also, what is interesting is that it’s been around 20 years since the paper has been published and there is no way humans have found any method to improve eye-hand coordination system. Can we use a machine to control the coordination in the future? For cognitive interfaces, models including GOMS (Goal, Operation, Method and Selection), Key stroke level mode and Model Human Processor are proposed which gives us a fresh perspective to design interfaces but while explaining the keystroke model, the author explains only method R and benchmarks it method W and method S without explaining what these methods are. For conceptual interface, he uses empirical studies, by partitioning the users’ behavior (no model user and model users) based on skilled execution and problem solving modes but he fails to give details about what these modes - model space, methods space and tasks space actually are. For task interfaces, the author proposes notecards which helps one to formulate ideas. Recently there are idea management software like bright idea and zoho projects which have adapted few principles from Notepad. The paper focusses mainly on Xerox's PARC's research findings which reduces the generality of the research findings. In 1986, Apple and Microsoft were competitors and it would have been interesting if Xerox benchmarked their methods with the methods used by Apple and Microsoft.
Mingzhi Yu 14:30:12 9/6/2017
The paper Direct Manipulation Interfaces talked about the cognitive account of the advantages and disadvantages of direct manipulation interfaces. It also introduces the concept of direct manipulation interface and its history. In the paper, the author explored 2 characteristics: distance and engagement. It points out that the direct manipulation is essential to make the system more interactive and less difficult. From the cognition aspect, a system engages the users by providing the feeling of controlling the object directly in a domain. Also, the author discussed the shortcomings and strengths of the direct manipulation interfaces. The paper is old but many ideas in it have been implemented on mobile devices nowadays. The second paper has the similar ideas. It suggests that the interface should directly allow the user to interact with the domain objects. It also talked about the human capabilities as a limiting factor. By knowing these limitations, we can expect a huge space to improve. It also clearly points out that the interface language should be clear for users to easily understand. Although this paper is more like a narrative summary that summarizes the previous findings instead of analysis, this paper is still valuable because it proposed theoretical guidance and models. It is also interesting to me because it discussed the human limitation in understanding the complex intellectual task and the potential to overcome it with computer-based tools.
Jonathan Albert 17:36:07 9/6/2017
Direct Manipulation: This paper spends a great deal of time cataloguing what direct manipulation interfaces are in addition to what they entail. The authors conclude by labeling directness as a path to be considered by a system designer, reminding readers that this is not the only path and that it is attendant with costs as well as benefits. The authors' comment that they "know of no really useful direct manipulation programming environments" was somewhat humorous to me, since I have found that this statement is still true.Beginner-friendly game design platforms are especially plagued by this limitation. Often their flow/event/object editors provide a simple interface for basic tasks, such as drawing a certain sprite at certain coordinates or moving according to an arrow key's input. Those editors give the facade of directness by adding functions to objects as soon as an icon representing that function is dragged onto the appropriate space. However, a gulf of execution is encountered when one object must reference or control something other than itself, or if it must perform many actions in a loop. To circumvent this, many engines just "punt" and expose a scripting backend which requires a quantum leap in user expertise--often via an interface disconnected from the friendlier, direct one. I have never seen an engine that meshes ease-of-use and utility without requiring cognitive context-switching. Considering the tradeoffs of directness could aid in the creation of an engine usable by "professional hobbyists"--those with enough knowledge to be frustrated by the limitations of the direct model and yet are intimidated by the knowledge required by the "conversational" programming interface. Such a tool could aid in the education of programming by providing an entertaining frontend to a problem space. At the same time, it should seek to blur but not ambiguate the path to virtuosity in the programming dimension, such that programming becomes a natural progression into productivity and specificity, instead of an intimidating wall of magic symbols haplessly used as a copy-paste destination. Today's engines would benefit from such considerations. Pointing to Pondering: This paper catalogues its authors' attempts in developing models to measure and enhance human-computer interaction. Particular attention was given to a user's mental model of a system, its degree of use in routine and complex tasks, and its relation to the intended, abstract model of the system. When the authors mentioned, in the context of developing a new program, encountering "idiosyncratic, exploratory users," I recalled a project at my workplace. We wanted to convert information from a more or less free-form employee assessment document to a structured, input-fields-based web form. The difficulty lay in how differently users would structure their responses--some would type a wall of text, some would manually double-space and align their text in lieu of using the word processor's functionality. While the conversion task was likely orthogonal to HCI (since I could manually account for special cases in this one-time operation), the patterns in the data confirm an idea laid out in this paper: users operate on their limited knowledge and seek to minimize the load on their own minds. Thus, an HCI system would benefit by considering user laziness--barring any obsessive-compulsive predilections. Several methods were discussed in various levels of detail (i.e., Model Human Processor, GOMS, Keystroke-Level, etc.). I think that the descriptions of these models and related examples and diagrams could have been moved to an appendix in the paper. Salient points in the introductory portions of the paper are somewhat buried in these technicalities. What is important to note is that humans tend to operate based on a limited "working set" of memory, that these operations can be effectively characterized by coarse models, and so on. How these models work is beneficial information for follow-up research, but its inclusion at the outset may cause readers to lose interest and fail to see whether these important ideas were viable or successfully expanded upon. Though it may seem self referential, I think the paper would have benefitted if the authors applied their own discoveries to the structure of their paper.
Muneeb Alvi 19:18:56 9/6/2017
Critique for Direct Manipulation Interfaces Summary: There are two fundamental representations of human-computer interaction that are discussed: a conversation metaphor and a model-world metaphor. These two representations are compared and contrasted through various measures including the gulf of execution, the gulf of evaluation, and directness. I was very intrigued by this paper. Once I read the paper, I found myself agreeing with most of the arguments in terms of modern human-computer interaction and interface design. However, when I went back and checked the publication date and saw the year 1985, I was surprised that the information in the paper still holds up over 30 years later. The paper starts by describing the conversation and model-world metaphors. It then goes into details about the metaphors using various factors such as the gulf of execution, which is how well an interface allows a user to perform an action the user is thinking of doing in a particular domain, and the gulf of evaluation, which is how well a system displays the current state of all objects that the user is working with. As I was reading, I was thinking how this relates to programs that I have used. The two metaphors reminded me of a CAD program that I used during my undergraduate career. The program allowed users to combine different 3d shapes (such as spheres, cones, etc…) to model real world objects. The properties of those objects such as color or number of edges could also be modified. The program provided many different ways for modifying the object. One way was to click on an object and modify the properties window. This is like the conversation metaphor. I had to request to the interface which property I wanted to change and how to change it rather than directly acting upon the object itself as I would have done in the model-world metaphor. Entering property values required me to check the object to make sure it was modified correctly. I much rather preferred clicking on the object and modifying it through direct interaction such as by clicking and dragging to make the object bigger or smaller. In this way, I did not have to check that the object was changed to my preference, but instead, it was as though I was interacting with the object itself. The paper then goes on to describe that although the model-world metaphor is very appealing, it is not always better than a conversation model. For example, the paper mentions that a model-world metaphor can require a great amount of abstraction from the low-level implementation in order to meet the needs of the users in particular domains. A great deal of abstraction limits the operations that can be done at the low level. In addition to this, the user continues to think of the domain as he/she did before. This means that a high-level abstraction can limit new or creative ways of thinking about issues. In the case of the previously mentioned CAD program, I also found cases where direct interaction with the object was not actually needed nor wanted. For example, during fine grain tuning of an object’s properties, it is much easier to enter a number in a text box and tell the interface through a conversation of the desired size than it is to click and drag on an object very carefully. In conclusion, these two models for interface design have many differences in aspects such as user friendliness, learning curve, and translating user thought into an action on the screen, and each metaphor has varying degrees of applicability. Critique for User Technology: From Pointing to Pondering Summary: This paper highlights the human side of human-computer interaction. It essentially treats the human as a computer and tries to analyze various tools or methods that can be used to quantify various tasks a human performs as he interacts with a computer. The paper starts out with acknowledging that human behavior must be a factor when designing personal systems. At the time of writing, the paper evaluates the impact personal workstations will have and that they must be designed with human behaviors in mind. The first human task the paper analyzes is the task of moving a mouse. A study is done comparing the speed of using a mouse vs using other methods such as a joystick. This study reveals two key findings. The mouse is the clear winner with the least amount of time needed and the mouse data points can be mapped using Fitts’s Law. The mapping to Fitts’s law shows that mice are already optimal devices and that the limiting factor in mouse speed is the human’s physical abilities. After analyzing this issue, the paper then describes humans using the Model Human Processor. This model essentially treats the human as a computer with its own processors, long-term memory, working memory, and output in the form of motor functions. This model creates a point of reference from which to analyze the human in the human-computer relationship. I was surprised by the findings on the first few pages. I did not think as much of the human side of the human computer relationship. I never thought about trying to model human movement using formulas so we can determine good hardware design for computers. However, I have thought of human’s as computers before. In my personal model, I feel as though humans are extremely complicated CPUs and that if enough complexity was programmed onto computers, a computer could mimic human behavior someday and maybe even surpass us. The paper then continues to describe various models used to predict behavior such as the GOMS model and the Keystroke-Level model. I was impressed with the usefulness of these models even considering their simplicity. For example, the paper discussed that Xerox used the Keystroke-Level model to predict expert behavior. I always appreciate when research and theoretical work can have far reaching practical values. Next, Unit Tasks are described and the authors continue to use the Model Human Processor. They say that humans’ limited memory is why we have to divide tasks into unit tasks or sub-tasks. They also mention that if we were able to process both input and output at the same time (like a computer), we would not need unit tasks. I reflected on this and I agree. When working on a computer, I am either trying to do/output something, or I am trying to understand/input something. I don’t remember a time when I was trying to do both concurrently. The paper goes on discussing other topics such as the need for designers of systems to have a good user and system model before releasing the system because it is important for the user to understand the model in order to best use the system and come up with creative never-before-seen solutions. Lastly, the paper mentions that the interface should be made in such as way that it should allow the user to perform a task naturally which is essentially the argument of the other reading we were assigned for today.
Spencer Gray 20:52:06 9/6/2017
In the paper Direct Manipulation Interfaces, the authors define and recommend the practices of creating a direct manipulation interface. Direct manipulation interfaces are a graphical approach where the user believes that he or she is interacting with the objects themselves, instead of the interfaces representation of the objects. This paper does not create any new results, techniques, or methodologies, however that does not mean it lacks significance. Before this paper, there was no concise summary of direct manipulation interfaces. There did not exist a formal definition or explanation of the intricate parts of direct manipulation design such as the gulfs of execution and evaluation. In this sense, the paper is a success. It is a great paper that can be cited by future research into the topic of direct manipulation interfaces. Since the paper was written in 1985, not everything described in the paper applies to interfaces today. The paper speaks of the hardware limitations that have prevented interface designers from developing more graphical interfaces. In addition, it states that the conversation metaphor interfaces are more prevalent than the model-world. We no longer have the hardware limitations that existed in the 1980s, and most interfaces today use the model-world approach. Interfaces such as desktop and laptop computers, smartphones, and tablets all fit the description of the model-world metaphor. Since this was simply an issue with when the paper was written, it would not qualify as a blind spot. However, one improvement that I would make if rewritten today, would be to use more examples of real interfaces. Most of the author's examples were real world examples that were outside the realm of computing, such as the example of filling the container of water. These examples were excellent in conveying the author's ideas. However, it also would be beneficial to cite commonly used interface in addition to the non-technical examples. This would provide a deeper understanding for users by showing how the interfaces they are familar with fit into the context of direct manipulation interfaces. In the paper User Technology: From Pointing to Pondering, the authors trace the history from their perspective of the analysis of user psychology and its effect of interface design decisions. They focus on studying how humans interact with machines, instead of focusing on the machines themselves. This paper is important because at this time, 1986, very little empirical research had been conducted on the optimal ways to design an interface for the benefit of the user. Even though computers are designed for human use, most studies focus on the machine. Studying user psychology can lead to an increase in productivity for the user. The part I found most interesting was the discussion of the mouse. The mouse is a piece of hardware that all users take for granted in the modern day. I found it interesting that the mouse will never be replaced by a different tool for pointing on a computer because it has been shown that the limiting factor in our productivity is human hand-eye coordination. I also found it interesting when the authors needed to differentiate they experiments between expert and novice users of a system. The novice users used more problem solving strategies of searching for a solution based on the problem space, but expert users developed cognitive skills. One thing that I would have liked to see would have been an analysis on how to develop an interface that leads novice users to developing cognitive skills. Different stratigies in interface development could lead to cognitive skills being developed quicker. An interface that is easy to become an expert in will be more successful small learning curves are likely to entice new users.
Tahereh Arabghalizi 21:08:32 9/6/2017
1- Direct Manipulation Interfaces In this paper, the authors address both the advantages and disadvantages of “Direct Manipulation Interfaces” that were introduced as good forms of interface design at the time. This work is important from one point of view because it was published in 1985 when interface design was at the beginning of its era and elaborates some fundamental concepts about direct manipulation interface design using good examples. But it seems week from another perspective because it does not offer new methodologies. It states that one of the advantages of direct manipulation systems is that they feel so natural because of the increased feeling of directness (impression or feeling about an interface) that results from using fewer cognitive resources and also the increased feeling of engagement (feeling that user is directly manipulating the objects of interest) that results from a good direct manipulation output language. On the other hand, regarding the disadvantages of direct manipulation interfaces, this work specifies that firstly, they have issues with accuracy because they put the responsibility on the user to control actions and secondly, they prevent users from getting excited about new technologies because they provide them with a wide domain of knowledge. With regard to the relation between this work and new technologies, as said in the previous paragraph, this article expresses that direct manipulation systems trade off one set of virtues and vices against another so it is important that we understand these trade-offs when we want to design interface for our systems. These hints can help interface designers to think about both users and system requirements and consider all the merits and demerits of different types of designs before they choose a specific interface design for a system. I also think that direct manipulation interfaces are still being used these days because the need for computers that make interactions with human more natural, is getting more and more. ------------------------------------------------------------------------------------------ 2- User Technology: From Pointing to Pondering In this paper, the authors trace some of the history of their understanding of users and their interaction with workstations. This huge work is a set of theoretical and empirical studies that built a vision of an applied science about computer users. They introduced four levels in this vision: 1- At the physical interface level which was a typical human factors experiment that compared a set of systems to determine which one performed best by testing proposed mathematical models. 2- At the cognitive level, they built a theory of the behavioral continuum between problem solving and cognitive skill and showed how practice on a task would gradually change problem solving behavior into skilled behavior. 3- At the conceptual level, they stated that users often have mental models of the systems they use, and that such models enable performance of novel tasks. 4- The authors believe that the most interesting problems are at the task level: understanding the nature of complex intellectual tasks and finding ways to build idea-structuring tools, both representation tools for structuring ideas and display tools for browsing ideas. At each level, the applied science was shown to be practical, in particular by being influential in the design of the Xerox Star. Since this paper is a story of tremendous amount of efforts that were done in the early years of the birth of computer, it can be considered as one of the most significant works in human computer interaction field and their theories are still applicable on new technologies. However, it would have be better if the experimental works and theories were explained in more details or in separate articles.
Amanda Crawford 21:26:52 9/6/2017
Direct Manipulation Interfaces Hutchins, E., Hollan, J., and Norman, D., Human-Computer Interaction, 1(4), 1985 Direct Manipulations investigates the cognitive impact of user interfaces based on the 'feeling of directness' concept. According to the paper, the feeling of directness is minimizing the gulf in which a user interface may span between the user and program through semantic distance, articulatory distance, and direct engagement. The regarding of anticipating and measuring a user's behavior was noted as impossible at the time. This was a stark contrast in belief when compared to Card's and Moran's paper called User Technology: From Pointing to Pondering . Hutchins, Hollan, and Norman urged developers to design programs and systems that moved from the strict and comfortable conversation metaphors (text based interactions) and to delve into using the model world metaphors using the up and coming graphical design objects. In addition, the authors pushes programmers to develop for non-programmers so that they can develop tools that can be used to solve problems in many different domains. User Technology: From Pointing to Pondering, Card, S., Moran, T., ACM Conference on the History of Personal Workstations, 1986 User Technology: From Pointing to Pondering discusses tools to characterize and measure user interaction behaviors through the strands of the physical, cognitive, conceptual, and task interfaces. At the time of their study, their claim was that the current set of Human Computer Interaction research was largely empirical and evaluative. The motivation of this paper was to address the shortfall of mathematical methods that accounted for the user's behavior through their AIP and SSD projects at Xerox PARC. Through their investigation, the were able to create the Model Human Process model. The Model Human Process is an engineering model based on a set of principles of operations. One of their notable contributions used the Model Human Process Model (MHP), in specific, Fitt's law to develop a mathematically sound decision on deciding whether to replace the mouse pointing device with other alternatives. In addition, the MHP describes how to develop research methods using the GOM's operating principle, which stands for Goal, Operators, and Methods. Through a variation of the GOM's principle, also known as the Keystroke Level Model, it was found that a user's behavior is based on unit task chunking and that users operates in three types of problems spaces. Their vision for their research was to learn how to develop interfaces that create a symbiotic relationships with their user's to solve complex and intelligent problems.
Xiaoting Li 21:53:05 9/6/2017
1. Direct Manipulation Interfaces: In this paper, authors explore the advantages and the disadvantages of direct manipulation interfaces, identifying that distance and engagement are two main factors that can improve directness. Semantic distance and articulatory distance are the two forms of distance that we need to take consideration. Semantic distance concerns meaning of an expression while articulatory distance concerns form of expression. How a system bridges the gulf of execution and the gulf of evaluation plays an important role in deciding the distance between the user and the system. To reduce semantic distance, either higher-level languages should be developed or user should develop the ability to build new mental structures to understand the system. To reduce articulatory distance, a richer set of I/O technology should be provided. The paper points that to increase direct engagement, interface should provide users with the illusion that they are interacting with objects in the real world when they are using the system. In the end, the paper points that while direct manipulation can be beneficial in some way, lacking of accuracy, difficulty with handling variables and amplifying user’s knowledge of a domain can be the problems. When we design a system, we should keep it in mind that whether to use features of direct manipulation and how we use these features depend on the tasks that we need to accomplish. This is an old paper, but it gives detailed description on the factors that decide the directness, which helps system design in today’s work. For example, augmented reality glasses increase direct engagement, giving users the feeling that they’re interacting with objects in the real world. 2. User Technology: From Pointing to Pondering By studying users and their interactions with personal workstations, authors show their findings in Human-Computer Interaction and their efforts at building related applied science in this paper. Authors mainly introduce their efforts at physical interface, cognitive interface and conceptual interface and introduce how these three interfaces were applied in the design of the Xerox Star product. Authors also introduce their current research on looking for solutions to help users better represent and manipulate their ideas on a system. As for the physical interface, authors develop mathematical models to compare user performance on different devices to learn that mouse is the fastest device. User’s limited information-processing capacity is the factor that constrain user’s performance. At the cognitive level, authors simplify the GOMS (Goals, Operators, Methods, Selection) model to get the Keystroke-Level Model. They discover that the use of the Keystroke-Level Model enables designers to calculate user’s performance with various interactive systems. Again, user’s limited information-processing capacity is the factor that constrain user’s performance. At the conceptual level, authors carry out an experiment by dividing users into two groups, one with users who were taught conceptual model while the other group with users who knew nothing about the model. Authors discover that when dealing with novel task situations, user model can be useful. It is impressive authors combine mathematical methods with experimental methods and show that each of the three levels is practical in real world by giving examples of how they were used in the Xerox Star product. And in addition to calculation models, authors also mentions the importance of taking psychology into consideration when designing a system, which brings inspiring guideline for today’s work in system design.
Ahmed Magooda 22:35:32 9/6/2017
User Technology:From Pointing to Pondering From what I understood the paper tries to model in a scientific way how users interact with machines and model what skills people usually use or need to deal with a specific task or device. The paper also claims that people have mental models and it tries to introduce these models and how they are generated and what are they used for. In my opinion I think the paper have some pros and some cons. From a pros prospective, I think the paper have a different point of view as it tries to analyse not the ease of use of but the human psychology developed while dealing with things and also the needed skills to deal with things. I think analyzing the psychological dimension can be beneficial, for example providing a theory for how the mouse can be considered the upper bound for human machine interaction and other similar examples are really interesting about this paper. The paper actually provides detailed analysis of many behavioural and psychological aspects which deems the paper good source. On the other hand, I think the paper somehow can be summarized in somehow a shorter version, also the paper is not easy to grasp I think that it needs to be somehow reordered or simplified a little bit. I also think some of the claims are subjective with no proof to support it, which in turn weakens the paper for a degree. Some of the points that I suppose could have been avoided or can be redone: - In all the experiments, no mentioning of how large the groups of people are. This needs to be mentioned as it can actually indicate if the findings can really generalize or not. - The paper can be simplified. ---------------------------------------------------------------------------------------------------- Direct Manipulation Interfaces This paper discuss the direct interaction between users and machine and how to give the user the feeling of direct manipulation of objects and a fast feedback of how objects react to help doing the task in concern. It also discusses how to reduce the gap between the user requirements and the provided means, either from the designers part or the user part, either by losing generality to gain specificity or by get some generality on the cost of changing the user mindset. The paper illustrates the two different types of directness (semantic and articulatory). The paper also relates the input languages to the output languages and how both can be related to users. I think the paper represents a good base for analyzing human-machine interaction and the different degrees of directness of interfaces alongside solutions to reduce the gap between the human and the machine. I think the paper is well organized except that the paper has some repetitive parts that can be removed to summarize it a little bit. While I can consider the paper as a good paper on the theoretical dimension, I still consider it weak on the evaluation dimension. The paper didn't actually provide any experimental proofs of the claims and arguments provided, maybe it is hard to get an experiment in such context but I think that it is still doable. On of the comments I have on this paper is the lack of referencing to existing applications and analyzing them, the paper introduced the ideas by real life examples (Piano, water tanks, so on ..) however I think a real life applications that can be analyzed and rebutted with or against would have been better.
Sanchayan Sarkar 23:42:47 9/6/2017
CRITIQUE 1 (Direct Manipulation Interfaces) In this paper, the author explores the two fundamental cognitive dimensions of directness: Distance and Engagement; on whose foundation Direct Manipulation Interface (DMI) is evaluated. The author also states the pros and the cons of such interfaces and their importance in Human Computer Interaction. The paper is extremely important in providing the foundation of what ‘direct’ means in the context of interface design. The author emphasises the notion that the ‘directness’ of any interface design is qualitative in nature. Therefore, there cannot be hard lines drawn on quantifying the criteria in measuring directness. Hence, the onus of evaluation of directness lies in the cognitive perception of the interface. The author further mentions that this cognitive aspect has two dimensions: Distance and Engagement. The author does a wonderful job in delineating the distinctiveness of the two metrics which can, to the new reader, appear to have a lot of overlaps. The author clarifies that Distance is essentially the gap between the way in which an user conceives a task and the way in which he perceives the results from the system. Gulf of Execution is the gap from what user perceives to input representation of the interface. Gulf of evaluation is the gap from the output side of the interface to what the user perceives. ‘Directness’, in context of distance, would mean minimising these gaps. The author further mentions that Distance itself has got two attributes: Semantic and Articulatory. Semantic Distance deals with relationship between the meaning of what the user wants to say and that of what the interface has to offer. Whether the interface language represents things in a way that the user conceives in his mind is a fundamental question on which Semantic Directness lies upon. It also means that a semantically direct interface would be one where the user can express his task concisely with what the interface presents. In order to reduce this distance either the interface has to be taken closer to the user or closer to the system. The former would require the use of high level Languages. The author is in favour of this method as it would essentially reduce the cognitive load on the user by using high level abstractions for the user’s tasks. However, the author cautions as too much specialisation of such interfaces would lose the generality and thereby inducing more job on the designer’s part to know much of the system. The author brilliantly cites the example of Extensions of LISP and UNIX as such interfaces, to support his point. Similarly, if the output of the system is presented in such a way where the user requires minimum understanding, the interface would undergo a specialisation which might make the system either too large or too specific. Therefore the author mentions that there is a trade-off involved between specialisation of interfaces and retaining the general primitive operations of the interfaces. He also cautions against quasi-directness like automated behaviour and virtuosity as both seemingly make the interface feel very semantically-direct but in real it’s only an illusion due to the user’s adaption to the system or user changing the task in hand due to much familiarity with the interface. Aside from semantic directness, the author briefs over articulatory directness. Here, not just the meaning of expression needs to be aligned but also the physicality of the expression. This is quite relevant in today’s time where touch surfaces of smartphones have come into being. For example, in a racing video game, one needs to swipe the screen faster for quick turning and slower for gentle turning. Even though swiping is the meaning of expression as far as turning the steering is concerned , it’s articulation in the speed of swiping is also an attribute for the interface to feel more direct. The author also delineates the second dimension: Engagement. He mentions that directness is often categorised as problems with two metaphors: Conversation and Model-World. The author clearly favours the later as it is one whether the interface acts less like an intermediator and more like a world in which actions on the objects are performed directly on the objects. Here, the representations of the objects behave like objects themselves. This is extremely similar to modern day text editors like MS-Word where whatever we type we can see it immediately on the screen and at the same time do real time editing operations on the texts themselves. The author is clearly indicative of the fact that such direct engaging interfaces are the course for the future. Even though the author does a great job is clearly explaining the distinctiveness of the two dimensions of directness, he also reconciles the relationship between the two by stating a 2-dimensional space of interfaces where by altering either of the dimensions of directness, one can get a more direct versus a less direct interface. An example of an interface having less distance but is more of conversation-metaphor oriented is a High Level language, where even though the user can translate his ideas to tasks efficiently, he still would not be working on the objects themselves. The author doesn’t hesitate in reinstating that the ‘Directness’ is qualitative and always dependent upon the task or domain it works upon. To conclude, the author has done well in illustrating two different fundamentals of directness. Although, he questions the application of Direct Manipulation Interfaces, he is quite optimistic. Even though this paper is quite comprehensive on DMIs , it is repetitive in nature. For example I would have preferred the author to categorise Distance Directness together rather than mentioning recurrences of it in multiple occasions. However, such are minor caveats. The paper, as a whole, is a foundation for Human Computer Interaction especially in today’s world, where the emergence of smarter gadgets require the use of DMIs in a much larger scale than ever before. CRITIQUE 2 (User Technology: From Pointing to Pondering) In this paper, the authors assert an understanding of the user rooted in the applied cognitive theory. It also presents a narrative of experimental evaluations done of rooting the four aspects of interfaces: physical, cognitive, conceptual and task based upon a cognitive understanding of the user. The paper is one of the most important work for novice and expert interface designers who want to understand the design principles of a comprehensive interacting interface founded upon the four pillars : physical, cognitive, conceptual and task. It does an excellent job in experimentally illustrating that the mouse is the best pointing input interface and in the process framing a model based on Fitt’s Law. Not only does the authors generate a model for mathematically evaluating any input pointing device, they also understand the human factors responsible for the efficacy of pointing. This gives a solid foundation on which physical interfaces can be built. Similarly when it comes to cognitive interface, the authors state that given a task and information, users do not opt for the rational thought line of problem solving but instead resort to breaking down the task into smaller sub-tasks and using familiarity of sequences of such sub-tasks, solve the problem. The authors therefore presents a goal-centric based Keystroke-Level Model in which given a task, method and motor skill parameters of the user, one can predict how long will a trained user take to complete the task. Both the models are based upon the limitation of a human in processing streams of data at a stretch and both of them establishes a solid cognitive understanding of the user. When it comes to conceptual interface, the authors demonstrate that those users who have a mental model of the system actually performs better, when it comes to perform intellectually challenging problems, compared to those that doesn’t have any familiarity with the system. The author proposes a novel approach of designing the conceptual interface as one based around “problem-spaces” where users can search for solutions rather than one where users does conventional problem solving methods. This understanding of the human behaviour presents an opportunity for designers to mould the user’s mental model of the system by designing an intended user model. This aspect of the research has largely being unexplored and therefore this paper comes as a major insight. However, the real benefit of this paper in the final aspect: the task interface. Here, the author presents a novel aspect of interaction where the user can interact with the system on ideas. This would essentially make the interface help the user in solving tasks by presenting structured ideas thereby making the users tackle on more “inventive” problems which are otherwise are intellectually extremely difficult. The author states of a novel system called “Notecards” which is digital equivalent of collecting ideas from sources and then forming chains of ideas to solve tasks. The author posits that the experiments designed on this interface are a contradiction to that of earlier aspects. In the task interface, the experiments are not in controlled environments with trained experts but with an open source of users. Task interface has enormous relevance in ‘Autocorrect’ feature of most current keypads where the interface suggests a variety of probable options which moulds the user’s mind in thinking of the word or a the chain of words. It is also present in modern day ‘Graph’ feature of ‘MS Excel’ where on presenting the data, the interface automatically comes up with plenty of pre-made graphs which invites the user into selecting the appropriate design of graph. The only caveat of the paper would be in the lack of illustrated examples that completely meets the stated aspects of the interface. Even though the paper tries to communicate that through carefully designed experiments, illustrated examples would have brought more clarity. Despite this, I would recommend that this paper is a landmark paper in presenting a comprehensive account on all aspects of an interactive interface and can serve as a solid foundation for novices and experts alike.
Charles Smith 0:34:59 9/7/2017
Direct Manipulation Interfaces: This paper explored ideas about distances between execution and evaluation in interfaces, and the trade-offs between different designs. While focused mainly on Direct Manipulation, it also presented several shortcomings, including use with repeated tasks, and the use of variables. This paper was important to help understand trade-offs faced in designing an interface, and how to help users accomplish their goals. It gave a great description of distances between the user and the execution of the goals, and the evaluation of the result, and explained how these trade-offs apply to many different situations. In today’s technology, there are plenty examples of interfaces like the ones described, with easy to point out advantages and drawbacks. The author also mentions other interface ideas, such a sound, which wasn’t as feasible at the time, but is commonplace today. It would be very interesting to see a paper updated with the more advanced technology we have today. From Pointing to Pondering: This paper looks closely at user inputs, such as the mouse, and several proposed alternatives, none of which found to be better. This paper also looks at how users can use programs to come up with innovative solutions to problems, based upon their knowledge of the program. Right off the bat, this paper looks at finding an alternative for a mouse, and fails to find one. In today’s technology mouse alternatives are everywhere. Most popularly, the touch screen. Seeing this study redone using modern technology may produce different results. The authors also look at how users can solve innovative problems on a postfix calculator, with and without knowledge of a stack. The experiment shows that both groups can use the program for normal case problems, but differ on more complex issues. It would be interesting to see how this extends to other problems, relying on different structures and algorithms.
Xingtian Dong 1:20:57 9/7/2017
1. Critique for “Direct Manipulation Interfaces” After reading the thesis, I think most part of the thesis is not related to the title “Direct Manipulation Interfaces”. Please tell me if I’m wrong. First, I want to talk about direct manipulation interface style. This interface style is becoming more important in today’s world. Most operating systems can’t exist without windows, icons, mouse and so on. This interface style is also extremely important in video games, CAD and virtual reality. ‘What you see is what you get’ as the author said, makes this interface style extremely user-friendly. This is also a way to make a complex process easy. Like someone build units to represent some functions, so users can combine different units to realize the program they want. This makes programming easier for novices. Then, I think the most important thing in this thesis is that the author defines two characters to evaluate if an interface is user friendly: distance and engagement. And for distance, the author defines two dimensions of distance: semantic distance and articulatory distance. I this ‘distance’ is not related to direct manipulation interface. It is related to another interface style—Command Language. Now it is a trend to use high level language. People are thinking about how to make it easier to program and they do what the author said ‘Higher-level functions written in terms of lower-level ones make the system easier to use when the functions match intentions’ . Their goal is to cut down the semantic distance. High level programming language like JAVA, Python and R are popular programming languages nowadays. C and assembly language are seldom used. But for articulatory distance, it is still a big problem even if we can do speech processing which largely reduces the semantic distance. But it gives us a direction about what the future interface should do, it should understand the articulatory meaning of what the user says to reduce the articulatory distance. 2. Critique for “User Technology: From Pointing to Pondering” I think this paper is still very important today. Computers have evolved really fast because of the development of all kinds of techniques, but human do not evolve that fast. Their physical limitation, how to cognize stay the same. We can still use the models in the thesis to analyze human beings. The most important methodology it provides is different kinds of interfaces and how to evaluate them. The four kinds of interfaces are physical interface, cognitive interface, conceptual interface and task interface. I think this is extremely important. It tells us we can analyze customers in several levels. From the basic physical level which can help us design physical input devices to task level which we need to build completed organization charts to analyze it. So before designing an interface, we need to analyze which level does it need and then use suitable model to analyze it.
Akhil Yendluri 1:37:46 9/7/2017
Direct Manipulation Interfaces : This paper discusses on how to improve the interaction between the user and the computer. The aim here is to make the user interaction with the computer more natural and intuitive. The author wants to make the user feel as if they are manipulating the object itself. The author explains this by introducing the concepts of distance and engagement. The paper proves that a directly manipulative interface improves the user's understanding of the subject while reducing the errors. It investigates the various aspects of distance and engagement which affect direct manipulation of interfaces. The paper has some shortfalls considering the fact that it was published in 1985 and the current advances in technology have tried to achieve the problems mentioned in the paper. Application of direct manipulation interfaces on programming can help in improving directness to a certain extent. But applying cognitive skills is part of programming thus limiting the directness. Direct manipulation has had a very successful impact on user based applications like sketchpad. It has made the interface more intuitive and easy to use. The future of direct manipulation technology is now moving towards Virtual Reality and smart devices which gives the user a more real-world feel and experience. User Technology: From Pointing to Pondering This paper analyses the interaction between the users and their workstations(user technology). The author tries to understand the behavior of the user while interacting with an interface. He then uses this information to create a design that works best for the user. They consider the physical, cognitive, conceptual and task interface while analyzing the behavior of the user. The use of scientific experiments to prove that mouse was faster and accurate than most other peripherals, model human processor to establish the maximum performance speed of a user and social experimentation on user's to understand the impact of cognitive skill is a perfect example of scientific analysis on user technology. Today more and more devices are based on touch interface rather than mouse. Touch screen displays and wearable technology have revolutionized the industry and have made user interaction more easier. The author also analyzes the cognitive structure of the user and uses it to create a model called GOMS. But one shortfall of this model is that it assumes that each keystroke and mouse movement takes exactly the same amount of time. This model does not consider the fact that each person has a different rate at which the think and work thus taking different amount of time to do the same work. Moreover the author tries to predict the various ways in which the user interacts in a particular situation. But it is not always possible to predict all the variations involved thus making this mode of evaluation susceptible to errors. The User Interface Design Methodology perfectly analyzes the various concerns involved in developing an interface. This methodology holds true for present day interface development too. As the paper was published in 1986 some of the challenges then have become reality now. Modern day computers are capable of simulating entire flight journey to computing complex scientific equations.
Yuhuan Jiang 1:38:45 9/7/2017
== Direct Manipulation Interfaces == This paper concretizes the vague notion of direct manipulation, by explaining what the feeling of directness is using terms from the field of cognitive science. The significance of the paper lies in that it was a novel effort (back in 1985) to account for the feeling of directness, and argue the pros and cons of direct manipulation. There are three main theoretical contribution in this paper: (1) The concept of distance. The paper defines a distance between the user’s intentions and the facilities a system provides. The argument that the directness the user perceives is resulted by reducing such distance provides a good starting point for explanation. (2) Two aspects of directness. The paper identifies distance and engagement as two aspects of directness. The distance, as viewed from the user’s point of view, consists of the gulf of execution and the golf of evaluation. The direct engagement is defined as the qualitative feeling that the user is directly controlling the actual object (not the program). (3) Two forms of distance. The paper states that two forms of distance exist: semantic distance and articulatory distance. The semantic distance relates user intention with the meaning of the interface language, while the articulacy distance relates the physical form of an expression in the interaction language and its meaning. Nowadays, direct manipulation is key to user experience. One of the most prominent pursuer of directness is Apple. When the original iPhone was introduced, people find hand gestures such as pinch-to-zoom, swipe-to-scroll engaging, and they increases the usability of the smart phone considerably. Why were iPhones successful? Using this paper’s terms, we know that the iPhones do not only minimize the distance, but also increases engagement. The distance is minimized because the user find it natural to use a finger. The engagement is increased because iOS uses life-like on-screen objects (rounded buttons, draggable sliders, etc.) to provide output. I find two very obvious blind spots of this paper: (1) The paper merely presents a cognitive account, but does not clearly state why such an account for direct manipulation is useful. For example, it does not answer the question of how can future user interfaces benefit from this paper. (2) One point is missing in the discussion of problems of direct manipulation (the last section). The implementation of a user interface with direct manipulation is oftentimes much more complex and time consuming than ordinary interfaces. In order to mimic life-like objects, at least two additional prerequisites must be satisfied: (1) fast enough CPU to render animations, (2) screens with fast refresh rates. Back in the time when this paper was authored, neither was easily accessible. == User Technology: From Pointing to Pondering == The paper is a summary of the history of the development of the applied science of users, by discussing various levels of user technology, namely the physical interface level, the cognitive level, and the conceptual level. The four types of interfaces used as representatives of each development phase were physical interfaces, cognitive interfaces, conceptual interfaces, and task interfaces. The paper is not a complete history of user science, as it only focuses on user’s interactions with workstations (and works at Xerox only). The paper is important because it provides a chronological view of the history of user technology. It consolidates what people understood about users at the time (the year of 1986), and can inspire the creation of new systems. The paper proposes quantitative laws to provide mathematical insights of how effective the different types of interfaces are. The findings, as found by empirical studies of this paper, were: (1) For physical interfaces such as pointing devices, user performance is limited by how much the user can process information (2) For the cognitive interfaces such as text processing, user performance is limited by the user’s cognitive skills. The findings can help engineers to, for example, determine the number of buttons on the mouse. (3) For the conceptual interfaces, the authors found that mental models can only work well in specific subtasks, because mental-based problem solving seems to be mentally demanding, which is oftentimes avoided by users. The authors proposed the task interfaces as the future. They used their at-the-time on-going project NoteCards as an example. It demonstrates how a user interface can help users compose their idea structures in a visualized way. And since the nodes in the NoteCards system is editable, it also demonstrates how the idea structure can be manipulated and evaluated. The authors also demonstrated a 3D organization chart browser, which shows to user the most details of only one node, and hides away details of other nodes. The user navigates through objects by “flying around” the chart. The blind spots of this paper: (1) Focusing only on Xerox products seem too biased. How generally applicable are the conclusions drawn from Xerox-only products? Could the inclusion of interfaces implemented by other companies somehow change the conclusions drawn by this paper? (2) It is missing the discussion of the segue between each type of interfaces. How did we transform from pointing devices to more higher level interfaces? Why was the lower-level insufficient?
injungkim 6:22:49 9/7/2017
1. Direct Manipulation Interfaces 1.1. Summary: This article explains the underlying basis for direct manipulation interfaces, and identifies two aspects of directness which are the distance between user’s intentions and machine features and the engagement between input and output. The distance can be reduced semantically and articulatory, and the output language that presents representations of object can have a better feeling of direct engagement. 1.2. Interpretation/Critiques: Direct manipulation interfaces were very groundbreaking in 1980s. At that time, Microsoft Windows released after MS-DOS, and it eventually let the people use the personal computers easily. As the article mentioned, by reducing the distance semantically and articulatory between user’s intentions and facilities provided by machine, the direct manipulation interfaces can be a good interface design. However, in recent computing technology, there are diverse types of I/O devices and the machine can analyze and train tons of input data. Also, I think that automated behavior can reduce semantic distance. For example, if a machine has trained the user’s behavior, it can predict user’s intention and execute automatically even though the users are not familiar with the system. Therefore, direct manipulation interfaces need to cover broader interface features with recent technology. If we would work on this work again, we had better do several types of experiments or survey to strengthen ideas. For example, to check the feeling of direct engagement, we may give an interface to control a certain system to many people who do not have knowledge about the system, then we can check how many people agree whether the output shows the representations of objects or not. 2. User Technology: From Pointing to Pondering 2.1. Summary: This paper shows the history of human-computer interaction by organizing with physical, cognitive, conceptual, and task interfaces. The article explain these 4 interfaces with some examples such as a mouse as a physical interface, processing devices as a cognitive interface, skill-level of user as a conceptual interface, and structuring and manipulating users’ ideas as a task interface, and these examples help to understand how each concept can help the interaction between human and machine. 2.2. Interpretation/Critiques: This paper is good to know the previous human-computer interaction and the way to divide the interface types in 1980s. We can still divide current interface with 4 types from the device to the human with recent technology. When we consider the recent technology, the example of each interface can be substitute with touch screen (instead of mouse) for the physical interface, cloud computing/parallel computing/distributed computing (instead of memory, processor, and switch) for the cognitive interface, and so on. Also, it is better to be divided by considering new technology such as Microsoft Cognitive Service, Amazon Alexa, or Uber automatic driving technology. With diverse recent technology, performances such as the unit task, challenges or problems of task interface are resolved. The conceptual interface can be experimented among trained users and train-needed users with machine learning techniques so that explicit conceptual model can measure if the system is built easy enough to the users. As a minor writing critics, there are vague expressions that have to be more specific. For example, sentences like “because a long time would have to be spent training the users”, or “we were busy pursuing the performance issues we have discussed” are vague and make the paper have weak assertion.
Kadie Clancy 8:11:42 9/7/2017
User Technology: From Pointing to Pondering In this paper, the authors provide a history of the understanding of users and their interactions with workstations based on their own personal efforts at creating the new field of Human-Computer Interaction. The authors simplify this history by focusing on four levels: the physical interface (pointing), the cognitive interface, the conceptual interface, and the task interface (pondering). At each level, the authors develop facets of the applied science according to theories of cognitive mechanisms underlying user behavior. These theories provide designers methods to better understand users in various ways, including obtaining practical time estimates for actions and understanding how users engage in skilled behaviors versus problem solving behaviors. This paper is fundamental to the field of HCI as it depicts the evolution of the science from the testing of theories on existing technologies at its beginnings, to the current state of HCI with the creation of new technologies based on the understanding of users. It offers both theoretical evidence and also supplements the practicality of the theory through the design of the Xerox Star. The paper outlines fundamental models at each level of the narrated history. The Model Human Processor is a simplified architecture of the user that may be used to compute predictions about human performance to guide the design of systems. The GOMS model establishes a user’s cognitive skills for performing tasks to allow designers to predict what a user will do next. The conceptual model, or simple non-technical abstraction of a system, provides users with an effective problem space in which to work in certain situations. This paper relates to today’s technologies by demonstrating basic principles, models and frameworks for how technology should be tailored to users’ cognitive abilities in order to design a successful product. This paper narrates the authors’ personal perceptions of the history of HCI, beginning with physical interaction of the user with devices, and cumulating in the interaction of the user with a system to perform complex and intellectually difficult tasks. Direct Manipulation Interfaces Direct manipulation interfaces can feel either natural or tedious to use, but have been commended as a good form of interface design. This paper addresses the pros and cons of direct manipulation, and the tradeoffs that must be considered when applying a direct manipulation interface to a specific task. While the concept of direct manipulation can not be well quantified, this paper attributes the “feeling of directness” to two phenomena: distance and engagement. Distance refers to the distance between one’s thoughts and the physical requirements of the system being used. A short distance means that the translation is straightforward and requires little cognitive effort. Engagement refers to the feeling that the user is directly manipulating the object of interest. More direct interfaces allow users to act on objects as if the representation is the object itself. Amid claims that direct manipulation interfaces will make standard programming obsolete, this article provides a realistic and balanced study of direct manipulation interfaces that includes the issues as well as the benefits. For example, direct manipulation interfaces have difficulty handling variables and are difficult for users to control actions with precision. This paper examines the underlying basis for direct manipulation, which had not been done until this point. The authors recognize the potential for direct manipulation interfaces to be powerful, but note that the tradeoffs may be too costly for some tasks. There are many examples of direct manipulation interfaces utilized in today’s technologies: word-processing like Microsoft Word, spreadsheets like Excel, and many video games. But as the author predicts, direct manipulation interfaces have not proved conversation metaphor interfaces obsolete. Conventional programming languages are still utilized for many tasks.
Ronian Zhang 6:23:49 9/12/2017
User Technology: 1. Summary: By pointing from a idea of improve mouse, the author tried to catch the nature of human-computer interaction by classifying the applied science of the user into physical interface level, cognitive level, conceptual level and task level and trying to build models for each of them. 2. Critiques: (1)interpretation: In physical level, the author modeled the mouse movement events by Fitt’s law and pointed out that the limitation of mouse performance is actually in the eye-hand coordinate system. In cognitive level, he applied human information processing into text-editing task, and decomposed user’s cognitive structures into GOMS(goals, operators, methods and selections) model. Then he simplified the GOMS model into engineering model (keystroke-level model) which could predict performance. He divided cognitive skills into unit tasks and argued that the limitations of user’s working memory limit the performance which lead to the conclusion that user’s capabilities are of vital importance when designing interfaces. In the conceptual level, the author argued that the intended user’s model and the building of a proper conceptual model are important in interface design by doing experiments of partitioning the user’s behavior. In the task interface level, the author focused on how to help users develop explicit mental models of idea structure and used their NoteCards to help user in tracking, browsing large information structure. (2) critiques: We now still use the mouse, but we somehow change the “pointing” part where the paper starts: IBM red dot is best solution for non-mouse laptops(10 years ago), then the touchpad improves the speed by using gestures, and the physical input device now includes touchable screens. But the performance still somehow limited by the eye-hand coordination system. Being a totally freshman in this field, some parts of the paper also rises my curiosities: Why not make distant buttons larger a solution for speeding up the process? Could different intelligence-level users get different interface? Could learning user’s specific capabilities lead to a customized interface? I believe some of the assumptions made by author are too unrealistic which makes me doubt the models and experiment results. The 3rd part of the paper is too long, a flow chart of the research progress would help reader better understand the paper. Also, it’s important to keep the figure and paragraph on the same page! ————————————————————————————— Direct manipulate interface: 1. Summary: this paper explored 2 distinct aspects of feeling of directness: distance (the generation of the feeling of directness) and engagement (the feeling of how directly one is manipulating the objects) and discuss the ways to improve the directness of user interfaces. 2. Critiques: Distance could be classified into semantic distance (the distance between user’s intention and meaning of an expression) and articulatory distance (the distant between meanings of expressions and the physical form). To reduce the former one: construct higher-level and specialized languages or the user could build new mental structures. To reduce the latter one: offer richer set of I/O devices. Direct engagement (direct interaction with the objects in a domain) feeling could be improved by providing user with a interactive world. However, when all things are done directly, problems seem to rise in handling variables, distinguishment, accuracy. Giving the certain historical background, the paper could be seen as extremely far-sighted and imaginary. Especially in the direct engagement part, many softwares are successful only because they give the user an immediate feedback and feeling of interacting with real world object. It’s hard to image how people at that time could think outside the box. The paper was well-organized, but compared to the former one, it’s difficult to see the progress of researching in this paper and how the ideas were added up.