Human Performance Modeling
- 1 Slides
- 2 Readings
- 3 Reading Critiques
- 3.1 Jonathan Albert 17:52:04 9/12/2017
- 3.2 MuneebAlvi 19:08:43 9/12/2017
- 3.3 Ahmed Magooda 16:35:37 9/13/2017
- 3.4 Tahereh Arabghalizi 18:27:30 9/13/2017
- 3.5 Xingtian Dong 20:36:20 9/13/2017
- 3.6 Mingzhi Yu 20:43:45 9/13/2017
- 3.7 Xiaoting Li 21:51:16 9/13/2017
- 3.8 Spencer Gray 22:15:37 9/13/2017
- 3.9 Charles Smith 22:32:11 9/13/2017
- 3.10 Krithika Ganesh 23:13:34 9/13/2017
- 3.11 Sanchayan Sarkar 23:26:50 9/13/2017
- 3.12 Kadie Clancy 23:45:06 9/13/2017
- 3.13 Ronian Zhang 1:28:01 9/14/2017
- 3.14 Yuhuan Jiang 1:40:54 9/14/2017
- 3.15 Akhil Yendluri 2:17:57 9/14/2017
- 3.16 Ruochen Liu 3:31:21 9/14/2017
- 3.17 Merhrnoosh Raoufi 8:59:58 9/14/2017
- Beating Fitts' law: Virtual enhancements for pointing facilitation, Ravin Balakrishnan, International Journal of Human-Computer Studies, 61(6). p. 857-874.
- Beyond Fitts' Law: Models for Trajectory-Based HCI Tasks, Johnny Accot, Shumin Zhai Proceedings of CHI 1997, p295-302
- An error model for pointing based on Fitts' law, Wobbrock, J.O., Cutrell, E., Harada, S. and MacKenzie, I.S. Proceedings of CHI 2008, pp. 1613-1622.
Jonathan Albert 17:52:04 9/12/2017
Pointing Facilitation: This paper focuses on variables in Fitt's Law, how they can be tweaked in a virtual environment, and whether such modifications improve user performance. The authors come to the conclusion that none of these modifications "work uniformly well" for typical GUIs. The results discussed in the paper were likely fundamental in the design of certain, modern operating systems' interfaces. Taskbars in Apple's OS expand to meet a user's cursor, and the icons thereon do the same. Microsoft Windows has settings to control pointer movement in proportion to mouse speed. And thankfully, neither possess some of the more invasive techniques that tend to raise difficulties in a cluttered-desktop environment, like "drag-and-pop." Nevertheless, I find that today's laptop trackpads are behind in terms of ease-of-use compared to mice or stylus/touchscreen inputs. While more recent advancements such as multi-touch pads allow for scrolling or other invaluable shortcuts, standard pointer movement is still clunky. This may be an artifact of the limited space allotted to the physical trackpad, but I sense that improvements can be made. Considering Fitt's Law in relation to this may lead to such advancements. Trajectory: This paper attempts to, in the spirit of Fitt's Law, quantify human reaction times in actions other than pointing. It surveys a number of experiments and the equations resulting from them, attesting their accuracy. Given the usefulness of Fitt's Law, this approach of explaining fuzzy concepts in mathematic language could be fundamental to revolutionizing the way certain design tasks are approached. The paper lists menu design as an example, and that has many facets to it. Button layouts, whitespace placement, and how and when menus expand could be more objectively tuned by these results. I do not foresee the entire process being objectified any time soon, but proven efficiencies are beneficial, even in small quantities. I can imagine these techniques being employed in GUI design programs and other related WYSIWYG editors. Currently, MS Forms and Google's Material Design guidelines offer rough suggestions to improve a UI's look and feel. Further, these guides are usually buried in a help manual somewhere, or available on a webpage completely external to the editing application; this can disincentivize new users of such systems from designing according the given principles. If more mathematically based rules were incorporated into the editors themselves, they could offer designers hints as they created their interface, perhaps reducing the chance of "that one menu" from accumulating poor characteristics. If these hints follow proper UI principles (e.g., providing visibility by highlighting or animating the change that could be made), they could teach new designers and reinforce existing ones, allowing more user interfaces to generally improve their usability.
MuneebAlvi 19:08:43 9/12/2017
Critique of Beating Fitts' Law Summary: This paper attempts to analyze ways of potentially allowing humans to interact with a mouse in ways that optimizes outcomes for Fitts' law. Some of the methods discussed are minimizing distance, drag-and-pop, and object pointing. The first method discussed to minimize Fitts' law is to reduce the distance between the cursor on the screen and the objects that the user wants to get to. There are two primary ways of achieving this. The first is to do something that most interfaces do today and that is to have menus located near the mouse. This happens already in most interfaces today when we right click on an interface and a vertical menu is brought up. However, in the paper, there is an alternative pie menu. I think that the vertical list can be implemented similarly to the pie menu if the cursor was in the middle of the list when the user right clicks instead of at the top left. This would reduce the average distance to all items in the list but maybe the user always wants to click on the first item so in that case, it would actually be worse than the cursor being in the upper left when the list appears. Perhaps in this case, the designers of the interface can predict which items in the list will be selected the most and place them near the cursor when the list first appears and place the other features further away in the list. The Drag and Pop method is also very interesting but I agree with the authors that a major downfall is not being able to predict which icons should be brought to the user. Maybe in the future, the interface can memorize which programs the user uses the most and bring them closer to reduce distance. The last method discussed is removing empty space using object pointing. Even thought the author says that it is not the best solution for modern desktops, I believe it is a good solution for mobile devices where the amount of space in between icons is already small. This fact is already utilized in many modern mobile operating systems such as iOS and Android. Critique of Beyond Fitts' Law: Models for Trajectory-Based HCI Tasks Summary: This reading focuses on aspects of interacting with a computer outside of just pointing and using Fitts' Law. These aspects are tested through a tunnel steering test. While the first reading focused on improving the speed of selecting items, this reading focuses on alternate interactions with the interface such as drawing in 3d or even drawing 2d curves. The way it goes about it is very clever. I found the approach that the authors used very methodical yet very applicable. They start simple with a goal passing test which they eventually discover is the tunnel traveling task. I appreciated how they use the example of tunnel traveling and then jump to another related task (goal passing) and show how it relates back to the introduced topic. They then try various configurations of the test such as a narrowing tunnel and a spiraling tunnel. The results of the difficulty of the tests were not that surprising to me based on personal experiences of playing games when i was younger where I had to keep drawing lines avoiding lines placed by other players. However, it was surprising that the authors were able to relate all the different tests through equations which would let them predict the completion time in other similar tests. In the end, i also was surprised how this relates to motions we perform with a mouse every day such as finding and clicking on an item in a list of items. Perhaps, if we used the pie-menu as described in the other paper, we might see different results in terms of difficulties since the pie is not a narrowing tunnel, but rather an expanding tunnel.
Ahmed Magooda 16:35:37 9/13/2017
‘‘Beating’’ Fitts’ law: virtual enhancements for pointing facilitation In this paper the authors summarized the approaches adopted by researchers to achieve the goal of enhancing pointing devices accuracy with Fitts' law as the base of the work. They found that the work done can be clustered into one of three categories. - Work that targets minimizing D: in this work researchers, aims to minimize the distance between objects in concern and the desired target, sometimes by removing empty distances and others by moving the targets to be as close as possible to the object. It was found that these approaches can't scale well to multi object selection and at the same time can be annoying to users. - Another direction of research was based on increasing W, by trying to either predict the target of the user or by enlarging objects that are close the pointer. This track was also found to suffer from bad generalization to multi selection, and also to configurations where a big number of objects exist, which can render enlarging objects hard. - The last track adopted by researchers, is trying to decreasing D and increasing W at the same time. In this line of work researchers targeted using variable C-D gain by moving pointer faster when it is still far from the target and move it slower when it is approaching the target. The issue with this approach is the lack of accurate cursor prediction to achieve satisfactory results. I think the paper did a good job summarizing the different lines of work that target the enhancement of pointing devices, while providing some inter-analysis based on fitts' law. ----------------------------------------------------------------------------------- Beyond Fitts’ Law: Models for Trajectory-Based HCI Tasks: In this paper authors investigated ways to incorporate fitts' law in other configurations, which lead to trajectory-based interactions. They investigated the validity of other models in case of configuration change like (moving in a direct line with infinite points, moving in a narrowing tunnel and so on). They proposed what they called the steering law, They followed that with multiple experiments to see how the steering law can be employed. I guess the paper is good, the experiments are organized and detailed alongside provided results. I guess one of the shortcomings of this paper is the experiment statement, there is no clarification of the number of participants who undergo the experiment, which can be an indication of how likely these results are going to generalize.
Tahereh Arabghalizi 18:27:30 9/13/2017
‘‘Beating’’ Fitts’ law: virtual enhancements for pointing facilitation by Ravin Balakrishnan : This article addresses the speed of user action regarding pointing and clicking in graphical interface based on Fitts’ law. So the main point is to increase target width and decrease moving distance to improve this speed. The authors’ state there might be methods to improve mouse performance because virtual pointing does not have the limitations of the physic al world. So they propose a model called “optimized initial impulse” in which the key parameter is an equation S = k (D/T) where T is time or duration of movement and D is distance. One weak point in the paper might be that it does not say how changing these parameters affects clicking a wrong item. The paper also mentions that resizing buttons can make using the interface more difficult. However, it does not mention what the level of difficulty is. Does it need more clicks on incorrect item or does it take more time to select the correct item? On the other hand, this paper was quite important in its time and some of its ideas are still used in today technologies like designing cell phones. ---------------------------------------------------------------------------------------------- Beyond Fitts’ Law: Models for Trajectory-Based HCI Tasks by Johnny Accot and Shumin Zhai : This paper reports the outcomes of some experiments that users had to move a cursor to a particular target while keeping the cursor inside a path with fixed width. Based on the results, they introduce another law including the time needed for all the mentioned tasks which is similar to Fitts’ Law but it is also beyond that. The authors used mathematical induction to generalize drawing of a path. They used more experiments to justify their models for navigating via straight, curved and spiral paths. In my opinion, the authors could have added more experiments. For instance, an experiment in which the cursor remains outside the path for a specific amount of time.
Xingtian Dong 20:36:20 9/13/2017
1. Reading critique for ‘”Beating” Fitt’s law: virtual enhancements for pointing facilitation’ I think this thesis is somehow useful to today’s world. The thesis naming “Beating” Fitt’s law, but actually, the author listed several experiments which make use of Fitt’s law. The experiments tried either to reduce D, increase W, or change both of them. Though most of them are not successful and applicable on computer. But they gave people some failed example and some ideas that might be applicable on other interface. I think the most important thing that I learned from this paper is that scientific definition, modeling and exploration of merging concepts is really important. Without Fitt’s law, the scientists will be really hard to estimate the result of their experiments. 2. Reading critique for ‘Beyond Fitt’s Law: Models for Trajectory-Based HCI Tasks’ I doubt the result of this thesis from several points: a. Only around ten subjects participated in each experiment. I think ten subjects are too less for an experiment. What were their race? How old were they? What ware the gender of them? I think only ten subjects could cause great randomness. So that the result is not reliable. b. Each subject repeated the experiment for about a hundred of times. Although each condition may have different difficulties, the subjects might gradually got used to that pattern of tunnel even if the condition changed somehow. c. In experiment 3 they only checked the condition when W1 > W2, they didn’t checked the condition when W1 < W2 People have bias about their visual input, a narrowing tunnel should be different from a broaden tunnel. They only checked one the conditions and the draw a conclusion. It’s not reliable. d. The author generate a global law after experiment 3 The author only examined several simple pattern of tunnels and draw a global law on a complex pattern without examining it. I think the global law is not reliable at all.
Mingzhi Yu 20:43:45 9/13/2017
The first paper "‘Beating’’ Fitts’ law: virtual enhancements for pointing facilitation" is a summarization and discussion of how the Fitt's law is applied in enhancing the pointing tasks. 4 major approaches are taken according to the 4 variables in the Fitt's laws. This paper is not tedious and it gives many interesting previous implementations of methods. I agree that for the mouse job, yes, Fitts' laws is an appropriate measurement. However, since this paper is published early in 2004, the PC technology has changed in many ways. The mobile devices, which used a touch pad, play an important role in people's daily life. Either the way of pointing and size of devices has dramatically changed. I will be wondering if the Fitt's laws still apply. Like what mentioned in the last part of this paper, the mouse sensing is proved to be another one important factor, which apparently not a variable in the Fitt's law. The technology has changed rapidly but the same evaluation might not be the absolutely valid. The second paper "Beyond Fitts’ Law: Models for Trajectory-Based HCI Tasks", which is one of the primary citations in the first paper. It shows some new interpretation of the Fitt's laws. They kept its spirit but explored some other possible existence of other regularities. This paper gives some answer to what I was wondering in the last paper, which is very interesting because this paper is published even earlier than the first one. However, it does show some innovated understanding of the existed laws. The same to steering task, the motion of dragging in nowadays touch control is another type of motion rather than simply pointing. We may ask if the Fitt's law is still valid in this case ?
Xiaoting Li 21:51:16 9/13/2017
1. “Beating” Fitts’ Law: virtual enhancements for pointing facilitation The author carries out a survey on a variety of research on making virtual pointing easier than its physical counterpart by changing the distance between users and targets and/or changing the virtual objects sizes. In addition to reviewing current research, the author points out further research directions including directly comparing various techniques mentioned in the review, exploring combinations of various techniques and getting feedback from end-users via indirect observations as well as quantitative measures. This paper reviews a large number of techniques on how to improve virtual pointing to let readers get a great opportunity to get to know the state-of-the-art techniques in this field. However, in my opinion, instead of simply comparing the techniques by reviewing previous works, the author should carry out some experimental research such as observation experiments. Besides, in addition to improving virtual pointing based on the quantitative methods, we should take many other factors into consideration, such as types of input devices, user’s everyday habits and whether the techniques can be applied in general or they can only be effective in one specific area. If we can take these aspects into consideration in future works, the result might be more persuasive. 2. Beyond Fitts’ Law: Models for Trajectory-Based HCI Tasks Due to the limitation of Fitts’ Law (can only be applied to pointing tasks), it cannot be used as a quantitative model to evaluate trajectory-based tasks completed by computer input devices. In this paper, the authors carry out four experiments to explore the existence of quantitative models for trajectory-based tasks. Based on the results of these four experiments, the authors confirm linear relationship in three types of trajectory tunnels, including straight tunnels, narrowing tunnels and spiral tunnels. The authors’ research work has enriched quantitative tools in the area of HCI. It is impressive that the authors make contribution to enrich quantitative tools to evaluate user interface, but the more important thing that I learn from this paper is the method they use when they carry out the research. In this paper, the authors carry out four experiments. They first start with a very simple and basic experiment and then add constraints or difficulties based on the result they get from the previous experiment. This kind of step-by-step experiment methods lead them to several linear model. And based on these models, they define a global law and confirm the linear relationship in trajectory-based tasks. In our future research work, we could use this method and it may give us impressive result.
Spencer Gray 22:15:37 9/13/2017
In the first paper, "Beating" Fitts' Law, the researchers survey the research in pointing at targets in computer interfaces and the attempts to improve upon the constraints of Fitts' law. The researchers describe the two main approaches as reducing the distance to the object that the user wants to point to or to increase the width of the object. Both optimiziations would decrease the movement time from the users current location to the target. Both of these optimizations are very intuitive, but the strategies to optimize them are not. What I found interesting was the design of widgets such as the pie-menu to make all targets equidistant. While the drag and pop technique seemed more harmful than beneficial, techniques such as the pie-menu present interesting solutions that cannot impair the user experience. In the second paper, Beyond Fitts' Law, the researchers exteneded Fitts law and applied it to more complicated tasks. They wanted to see how and if Fitts' Law applied to steering tasks. By varying the length of the tunnel, the width of the tunnel, and the shape of the tunnel, they confirmed that an extended version of Fitts' law does apply. I found it interesting that they started with a goal passing task, then extended that to a tunnel by creating an infinite number of goals to pass through. To me, that was a very insightful application of the results they discovered in the goal passing task. However, if I were to redo their approach, I would use more subjects in the experiments. While the subjects did complete enough trials for the results to be accurate, the researchers only had 10-13 subjects per experiment. The results would be much more concrete with a larger and more diverse test group. That being said, this is still a significant paper in the HCI field because it adds more quantitative tools to a field that is mostly qualitative.
Charles Smith 22:32:11 9/13/2017
On: Beyond Fitts’ Law The authors of this paper took a quantitative approach to look at how Fitts’ law can be related to other ideas, or more specifically, relating to drawing and motion instead of pointing. While the primary idea of Fitts’ law is still in play (time increases with larger distances, and smaller icons) the equations do need to be changed to better reflect these more difficult actions. On: “Beating” Fitts’ law The author of this paper takes looks at how to ‘cheat’ Fitts’ law in interfaces. The author looks at changing the size of objects (w), the distances to objects (d), and both. A big concern when changing pointing would be the user’s reaction to such a change. The author talks about having the cursor ‘jump’ across empty spaces, menus showing different sized objects, and other. For a user’s point of view, these changes would greatly disrupt their workflow. Changing the sizes of objects or decreasing the distances when the user is pointing seems like a pretty novel idea. These increased sizes and decreased distances still seem like they are bound by Fitts’ law. The author talks about how the speed is likely due to the user having to make corrections when moving to the target, and needing feedback to make those corrections. Perhaps an interface that automatically corrects for a user’s errors, such as by predicting the icon the user was targeting, and pre-select that item.
Krithika Ganesh 23:13:34 9/13/2017
Beating Fitt's law: This paper presents different pros and cons of the methodologies proposed to beat the Fitt's law by reducing the target distance or by increasing the target width or both. The author discusses the following methodologies for reducing distance: design widgets that minimize D (popup menus), bringing potential targets to the cursor (drag and pop) and object pointing, the following methodologies for increasing width: area cursors and expanding targets, the following methodologies for increasing width and reducing distance: changing the C-D gain. The author mentions "Of all techniques object pointing is arguably the one with the most significant performance gains", but fails to mention how. In fact, techniques like object pointing and expanding target are not applicable when there are many icons together. What interested me the most is that one could increase the overall efficiency by dynamically varying the CD gain based on input device speed. The author titles his paper as "beating Fitt's law", but all the techniques mentioned are based on Fitt's law and are not proving the law wrong, hence its more appropriate to state it as improving the Fitt's law or consequences of Fitt's law. The first finger-driven touchscreen was invented by E.A. Johnson in 1965, but this paper which was published in 2004, only discusses about mouse and stylus as input devices leaving out touch screens which are used widely in the 21st century era. Also in today's world input is not just by pointing, one can use voice(audio), an image of one's self (Iphone X's image ID), figure prints etc as inputs, and it would be interesting to benchmark the performance of these inputs and find out if there is mathematical law like Fitt's explaining the constraints faced while using these input devices. Beyond fitt’s law: Models for trajectory-based HCI tasks: This quantitative paper explains why Fitt’s law is not an adequate model for trajectory based tasks and takes several experimental steps to derive and validate quantitative relationships between completion time and movement constraints in trajectory based tasks. The author of the paper knows how to break down the problem and goes on to explain from the lowest complexity experiment like goal passing task, experiment 2 increasing constraints, experiment 3 narrowing tunnel and to the highest complexity spiral tunnel experiment and derives a local law. What is interesting is that the author takes a pragmatic approach in which he verifies each of the hypothesis he proposes with experiments, hence confirming the prediction. The author in detail explains about the experimental setup used for each experiment but fails to explain the rationality behind why he made the choice (number of subjects, number of trials). Also, I felt as the complexity of the experiment increases from experiment to experiment, the number of practice sessions should have increased for increased for the subjects, which may have reduced the error in the more complex experiments.
Sanchayan Sarkar 23:26:50 9/13/2017
CRITIQUE 1 (BEATING FITTS’ LAW: Virtual Enhancements for Pointing Facilitation)----> This paper by Ravin Balakrishnan is a survey paper done on various pointing techniques developed that breaks the physically bounded principles out of the picture. It also compares the pointing techniques against the general pointing modelled by Fitts’ law. One of the strengths of the paper is that it is very well categorized into three categories: A) Methods that attempt to decrease distance, D of Fitts’ Law, B) Methods that attempt to increase the size, W of fitt’s law and C) methods attempting to both. The categorization brings out an important characteristic that changing one of the variables breaks the physical boundaries as these variables are unchangeable within the physical space. The paper also asserts the main explanation for the human motor action is an optimized initial impulse model where the user initially makes a rapid movement towards the target but resorts to small corrective movements based on closed-loop feedback when it comes in close proximity of the target. The paper cites an important paper “(Graham and Mckenzie,1996)” which introduces vocabulary of motor space, virtual space and a linking transfer function between the two spaces. The paper then enlists major contributions in reducing distance like Calahan et al. on PIE-menus that represents multiple targets at equal distances from the cursor or drag and drop of Baudisch et al. which brings proxies of targets near the cursor or the Guiard et al. where the cursor skips empty spaces between one target and another. An interesting commonality is all of these reduce the D in the motor space but not in virtual space. However, all of them faces a problem when the virtual space is cluttered or when there are multiple targets. When it comes to increasing the size of target, I really liked the idea of dynamic expansion of targets by McGuffin and Balakrishnan (2002) which states the target expands once we move closer to it. This is quite common in a video game called FIFA, where highlighting a selected country expands it. From a user’s perspective, this is a very good method. However, Gutwin et al. showed that expansion of targets when viewed from fish eye, also results in spatial distortion of target. It also suggests to factor in the dynamic characteristic of the pointer into Fitts’ Law. As far as techniques in category C are concerned, none of them finds relevance in today’s world with the increasing application of touch or stylus. Another merit of this paper is presenting future research scopes such as research in predicting the cursor movement or factoring in user’s experience in creation of a mathematical model. However, had the paper presented a comparative quantitative analysis of the various methods against the Fitts’ Law, it would have far easier to evaluate the techniques. Also, none of the techniques are evaluated against each other. A shortcoming therefore would be the lack of comparative graphs, diagrams or charts that shows such an evaluation. Despite this, the paper gives a broad landscape of the current research and therefore is useful for readers to get a quick overview, in the domain of pointing facilitation techniques, simultaneously. <------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------> "CRITIQUE 2- Beyond Fitts’ Law: Models for Trajectory Based HCI Tasks"----- In this paper, the authors present a new law that can form as a theoretical quantitative model upon which input devices can be evaluated. The paper asserts itself building upon the limitations of Fitts’ Law and and presents a better model for a wide array of motion based interactions. The paper is highly appreciative of Fitts’ Law and it’s enormous relevance as being one of the rarest quantifiable model that can predict outcomes without the need of empirical evidences. However, I agree when the authors state that the Fitts’ Law was meant for only one movement, i.e pointing. As good as it is, HCI has evolved a lot and with that comes a lot more trajectory actions than just pointing. Therefore the authors’ claim of presenting an equally impactful quantitative model for a robust range of movements only supports the importance of the law. Therefore the authors claim what is called ‘Steering law’ which can give a mathematical estimate of time-accuracy relation or a speed-accuracy relation. The reader should be highly appreciative of the four stage well illustrated experiments the authors designed to prove their claim: Goal Passing, Tunney Steering, Narrowing Tunnel and Spiral Tunnel. In the first experiment, there is a direct correlation between the proposed model and the Fitts’ Law. It’s only in the next experiments where the proposed methodology shows a linear relationship rather than a logarithmic one. Second point of interest feature that came out of these experiments is that the error rate kept on increasing. This was never assumed in Fitts’ Law. This has the potential of opening a new direction of research for modelling error rates. The third benefit of this paper that it is able to postulate a generalized law for time estimation for tracking movements in whichever shape. Along with that it also postulates a local law modelling the speed and the accuracy. One of the biggest negatives of this paper is that it occasionally claims of establishing mathematical relationships and yet there is no proof to back it up. For example, the author claims that the “Narrowing Tunnel” apparatus can be abstracted as both set of elementary “steering” tasks and a set of elementary “goal passing” tasks with same index of difficulty. Yet it offers no proof. I have observed similar recurring patterns in the paper. Despite this, the paper has enormous relevance in both input-device-operated and touch-operated devices of this age. For example, in ‘Microsoft OneNote’ one can see experience a trajectory based motions in drawings or taking handwritten notes. Designers must keep “Steering Law” as the basis on which they can evaluate their input devices. This has even bigger implications in Animation tools that deals with complex polygon structures. Therefore, despite the minor flaws of this paper, its’ scope has tremendous application in today’s age.
Kadie Clancy 23:45:06 9/13/2017
“Beating” Fitts’ Law: Virtual Enhancements for Pointing Facilitation: Pointing in the physical world, and classically in the virtual world, is well described by Fitts’ Law. Recently, HCI researchers have recognized that pointing in the virtual world should not have to be bound by the same constraints as the physical world. Thus, we may actually be able to “beat” Fitts’ Law to make virtual pointing faster and easier than physical pointing. The authors provide a survey of the recent research on obtaining artificial pointing performance gains in the virtual world. The survey is grouped into three categories: attempting to decrease D (the distance from target), attempt to increase W (the width of the target), and attempting to both decrease D and increase W. A few interesting ideas concerning decreasing D include pie-menus, the drag-and-pop technique, and the object pointing technique. Research on increasing W includes area cursors, and dynamically expanding targets. Finally, research ideas that involve both decreasing D and increasing W revolve around the idea of dynamic C-D gain based on input device speed with “sticky targets.” This paper is important as it provides a comprehensive background and overview of the important ideas along with future research directions of the field, which provides a starting point for anyone interested in beating Fitts’ Law. Since pointing is such a fundamental task due to GUI prevalence, even modest improvements in pointing performance can have serious impact on user productivity. This paper offers both the performance benefits of the selected techniques, as well as the shortcomings. Specifically, many techniques may produce performance gains when considering an isolated target, but difficulties present themselves when the goal is selecting a single target out of several closely situated targets, as interfaces often are in reality. This emphasises the importance of not relying solely on the results of a task-specific controlled experiment. Beyond Fitts’ Law: Models for Trajectory-Based HCI Tasks: Fitts’ Law addresses only one type of movement: pointing to targets. With computer systems being increasingly used for other tasks, like navigating through nested menus and moving in 3D worlds, the authors realized the need for a more accurate model to describe these trajectory-based movements. In this paper, the authors added to the limited set of quantitative laws used for HCI research and design by presenting experiments, analysis and applications of the steering law. Using a series of four increasingly constrained goal-based task experiments, the authors were able to formulate a global law that predicts the total time to perform a steering task. The authors were then able to derive local law variations from that global law, like one that models instantaneous local speed. The global law was first derived using Fitts’ Law and then fit to actual user experiments; each experiment allowed the model to be further generalized. The results of the study can be used to model interactions with GUIs, or as a means to compare designs that use trajectory-based movements. This paper is important as it adds to the limited set of quantitative tools for motor control research and user interface evaluation, beyond the classic Fitts’ Law. This is a step towards modeling other types of regularities in human movement, beyond simply pointing to targets, with mathematical equations. The authors provide a formal research method consisting of user experiments and the fitting of quantitative models. They ultimately demonstrate that the logarithmic relationship between movement time and width of the target in a tapping task also exists for a goal passing task. The authors also recognize the limitations to these simple laws, and points of further research in this area.
Ronian Zhang 1:28:01 9/14/2017
Beyond Fitt’s Law: 1. Summary: This paper expands the Fitts’ law from pointing-target events to trajectory-based tasks. By giving experiments, the author shows that his evaluation could apply in straight, narrowing and spiral tunnels: steering time is linearly related to inverse of the width and speed of the movement is linearly related to the width. 2. Critiques: The paper starts by pointing out the limitation of Fitts’ law and argues that it is not adequate for trajectory-based tasks. By conducting experiment, the author proves that a steering task with constrains on both sides follow the same logarithmic law as Fitts’. He forwards the problem into tunnel traveling tasks (though I doubt about why the difficulty evaluation should simply be the add up if there are more intermediate goals), and concludes that steering time is linearly related to inverse of the width and gives the model. Then, he transfer the problem into narrowing tunnel and use the same integral way and experiments to build the same model. Then, he tests the method for complex spiral tunnels and draw the same conclusion (even though I highly doubt the experimental results of “eleven subjects” could represent the behavior of humankind). After that, he conducts the law for instantaneous speed and tries to apply the solution into menu selection problem (even though I doubt whether it could be simply described as 2 steering tasks, since the end position of the vertical move could highly influence the horizontal move). The paper shows a common way of doing research: summarize the problem -> find differences between current one and former-well-known problem -> build a model -> test it -> the iteration of applying the solution of more complex problem and test. ————————————————————————————— Beating Fitts’ Law: 1. Summary: This paper argues that pointing in visual world does not necessarily abide the Fitts’ Law. It covers methods of facilitating pointing by reducing D, increasing W and do both.It points out that all methods fail to perform well when dealing with multiple targets and combining various techniques could be promising. 2. Critiques: The covers a huge amount of topics: the possible solution for each method, the drawbacks of them and potential solutions to them. It starts with the review of Fitt’s law, followed by the deviation( the possibility of hitting the target). The reducing D methods contains: moving targets close to cursor (limited types of target) -> (use hotkeys), bringing potential target sets towards cursor (overly clutter region near cursor) -> explicit trigger || widget with parameters that users could specify, removing unused space (ruin the layout and usability) -> jumper when sparse && regular when close -> problems: when selecting individual pixel && objects tiled together && annoying in real usage. The increasing W methods contains: area cursor (obscure underlying data, difficult to select closely grouped data) -> use regular when close && use area when far apart, expanding target (affect neighbors when in close proximity) -> predicts the trajectory of cursor (difficult to achieve). Doing both methods contains: dynamically change the c-d gain (increase when cursor outside the target, decrease when inside) -> trap when multiple targets appear along the way. The article is very thorough and after 13 years, we could see some of them are already embedded in the system (Mac dock as presented in the article, mouse acceleration, pie menus are common in games). But it really confuses me why the other seemingly good idea never come into my sight. The paper was written when Fitts’ Law was widely accepted and acted as the fundamental knowledge of human interface design, it tried to beat it even though very few of the points truly get out of the range of Fitts’ Law.
Yuhuan Jiang 1:40:54 9/14/2017
== ‘‘Beating’’ Fitts’ law: virtual enhancements for pointing facilitation == This is a survey that investigates virtual pointing devices, especially how these devices “beat” the Fitts’ law and become easier than physical pointing. The paper further argues that even the state-of-the-art pointing devices do not work well when objects are placed together closely. According to the Fitts’ law (which is proposed for the physical world), to make grabbing an object easier, one could either (1) decrease the distance, or (2) increase the width. Following this logic, the paper first reviews device designs that minimizes the distance, then reviews designs that increases the width, and finally designs that does both at the same time. The following are the ways this paper reviewed for reducing the distance: 1. Move the targets closer to the cursor where feasible; 2. Temporarily bring the potential targets to the cursor; 3. Remove the empty space between the cursor and the targets. The following are the ways this paper reviewed for increasing the width: 1. Use area cursor (instead of the point cursor) 2. Expand the target directly. The examples this paper reviewed for doing both is to dynamically change the C-D gain. == Beyond Fitts’ Law: Models for Trajectory-Based HCI Tasks == This paper proposes an alternative for Fitts’ law to model trajectory-based interactions. Traditional Fitts’ law no longer applies to nested-menu navigations, curve drawings, and 3D world navigating. The way the authors develop the new model is very interesting. It consists of four experiments. The first experiment is based on a goal passing task with constraints on both ends (i.e., two goals), which suggests that a steering task also follow the logarithm of Fitts’ tapping task. The second experiment increases the constrains by adding more goals (>2). A recursive model is proposed for multiple goal movements. The third experiment narrows the width of the tunnel. The proposed model finds that the index of difficulty is proportional to the log of the ratio of the widths, and disproportional to the difference of the widths. The fourth experiment incorporates a spiral tunnel. The paper does not claim these models also have limitations, due to (1) human body limitations and (2) the local law can take path curvature into account. Nevertheless, the models proposed in this paper can serve as quantitative tools for evaluating and guiding future designs.
Akhil Yendluri 2:17:57 9/14/2017
Beating Fitts' Law: Virtual Enhancements for pointing facilitation, Ravin BalaKrishnan
The author here tries to improve the performance of pointing devices in computer by enhancing the Fitts' law. The author suggests various methods reducing D(minimizing widget size, bringing potential targets to the cursor, removing empty spaces), increasing W, reducing D(area cursors, expanding targets) and increasing W together(dynamically changing C-D gain). In the present day UI cursors change shape and size dynamically depending upon the canvas/target. The paper doesn't take external factors such as small size of trackpad, uneven surface for mouse, greasy fingers on trackpad, etc into consideration. Although in today's technology most of the pointing and selecting work is done by touch interface. Therefore the original Fitts' law would stand as accurate mode for measurement of time. But new research for finding the time for pointing and selecting in Virtual Reality Interface is required as the mechanisms used there are quite different.
Beyond Fitts' Law: Models for Trajectory-Based HCI Tasks, Johnny Accot and Shumin Zhai
Fitts' Law helps in finding the time required for pointing/selecting when objects move over a 2D plane. But to calculate the Index of Difficulty for actions such as drawing curves, moving in 3D worlds requires us to find a new way of computation which is precisely what this paper throws light on. The author begins by explaining the concept of Fitts' Law and why it can't be applied in today's technology. He then performs four experiments which are Goal Passing, Increasing Constraints, Narrowing Tunnel and Spiral Tunnel where the first experiment is used to find the Time for a person to pass a line from goal A to goal B. Then he introduces constraints in the second experiment by bringing in intermediate goals. In the third experiment he changes the width of the tunnel at one end and in the final experiment he distorts the tunnel into a spiral and finds the Index of difficulty, plots the relationship between time and Index of difficulty, Speed and Path Width. This paper gives us insight about the various types of interaction in a touch interface while finding the duration, speed and effort required to do such operations.
Ruochen Liu 3:31:21 9/14/2017
1. “Beating” Fitt’s law: virtual enhancements for pointing facilitation: This paper is a survey of recent research or new techniques that aim to facilitate pointing at targets in graphical user interfaces. According to the paper, essentially, there are three methods to reduce the movement time (MT): primarily reducing distance (D), primarily increasing width (W) and both decreasing distance (D) and increasing width (W). Generally, a shorter movement time means a better pointing experience and a better interface design. 1: Three methods are mentioned in the paper to primarily reduce the distance (D). Pop-up linear and pie menus can effectively reduce (D), though it is hard to apply on other interface elements. The method of temporarily bringing potential targets to the cursor works best on a sparse virtual desktop. But it may cause occlusion and false activation while working on dense desktops. Object pointing may be the technique with the best experimental results, but there are still many practical considerations. 2: Two techniques are presented in this paper to primarily increase the width (W). Compared with point cursor, area cursor performs better when targets are far apart. When the targets are close together, the performance of the two methods are identical. The dynamically sized widgets could be an effective strategy for maximizing the use of screen without sacrificing the target selection performance. 3: The basic adaptive C-D gain technique, which is based on device movement speed instead of the location and size of targets, can improve pointing performance for single isolated targets. However, this technique may face more problems when multiple targets are presented. On the whole, this paper illustrates recent research on new techniques for enhancing pointing performance in graphical user interfaces. 2. Beyond Fitt’s Law: Models for Trajectory-Based HCI Tasks: Basically, this paper presents the inspiration, analysis, four concerning experiments, and the design applications of the steering law. Compared with the Fitt’s law, which is robust and quantitative to be applied to human-machine interaction design and research, steering law has a bigger potential and can be used to enrich the research space. The main contribution of this paper is the presentation of steering law. Steering law is consist of two parts, which are integral form and local form. Integral form declares that the steering time is linearly related to the index of difficulty. And local form states that the speed of movement is linearly related to the normal constraint.  In this paper, the index of difficulty is defined as the integral of the inverse of the shortest distance between the two base lines. The invention of steering law is important because since Fitt’s law is not an adequate model for trajectory-based tasks, steering law provides a new way to get models for trajectory-based tasks like writing, drawing and steering.
Merhrnoosh Raoufi 8:59:58 9/14/2017
Beating Fitts' law: In this article, the author explained that virtual pointing is different from physical one because many of constraints of the real world do not exist in the virtual environment. Thus, the author believes some improvement can be done to virtual pointing so that it beats the Fitts' Law. The article introduced some enhancements in three categories; decreasing the distance parameter, increasing the width parameter and applying both of them. To, reduce the distance, the article suggested some alternative designs such as circular instead of linear one. Another idea it introduced was to bring virtual proxy of items that are likely to be clicked near to the cursor or removing empty space between the cursor and objects. To increase width, the author argued that instead of sharp pointers we can use area cursor. This will facilitate pointing by increasing the chance of pointing widgets. The author expressed that the size of widgets can be dynamically increased to gain faster pointing. Further, the control-display gain can be changed dynamically in a way that it facilitates pointing. ---------------------------------------------------------------------------------------------------------------------------------------- Beyond Fitt’s Law: In this paper, the author argued why the Fitts' law is not proper for modeling trajectory-based tasks. In this article, they went through some experiments to find regulations for modeling trajectory-based tasks. The result of the study revealed that for steering tasks there is still logarithmic relationship just as the Fitts' Law says. The experiments went beyond that to investigate a globalized regulation for trajectory-based tasks.