`David Rogers, Feimo Hou, Chandelle Vuolo, Benjamin B. Bederson
`Computer Science Department
`University of New Mexico
`Albuquerque, NM 87131
`(drogers, feimo, bederson@cs.unm.edu)
`
`ABSTRACT
`We have implemented an interface which supports tossing
`as an alternative to screen-wide dragging of icons, and have
`explored how this basic mechanism may be generalized. We
`discuss its advantages and limitations, and report the results
`of our user testing. Preliminary results show that tossing is
`intuitive and is perceived to be faster than drag and drop.
`
`KEYWORDS
`Tossing, animated interfaces, user interfaces, animation.
`
`INTRODUCTION
`Animated interfaces have been proposed as a way of giving
`the user a better understanding of actions taken by objects
`on the screen. They allow intuitive transfer of information
`from the system to the user [1][5], and help the user to navi-
`gate large information spaces [2][3][4]. The SELF system
`[1] incorporates animation techniques from cartoons in
`order to maintain the user’s understanding of the interface,
`and give entities a more consistent presence. Animated
`icons [5] utilize animation to give the user information
`about use and properties of a specific icon. Pad++, a multi-
`scale interface [2][3], incorporates animated zooming into
`data as a method of exploring large information spaces.
`A widespread interaction that users perform with current
`interfaces is dragging objects and dropping them onto other
`objects, expecting certain actions to take place as a result.
`For example, dropping an object onto a trash can icon
`causes that object to be “thrown away”. Note that the action
`of dragging the icon has no semantic meaning in the drag
`and drop action, it is merely a way of getting two objects
`into a physical relationship which itself denotes an action,
`and tells the system that an action is required. We feel there
`is a more natural method of achieving this same interaction,
`which allows the user to indicate which objects are to inter-
`act. In our system, the user makes a simple gesture to indi-
`cate a task, and the objects are animated to complete that
`task. In addition, tossing has the potential to be much faster
`to execute than drag and drop because the act of moving the
`mouse long distances is eliminated.
`
`TOSSING
`We have implemented tossing, a gesture which supplements
`
`drag and drop. A toss is made with a pointing device while
`holding a graphical object. It is analogous to throwing a
`piece of paper into a trash can or a file into an “in basket”.
`In our interface, this gesture allows the user to initiate an
`action which is completed by algorithms and information
`resident in the system. It is our hope that use of computa-
`tion by the interface will make the user’s task simpler, as it
`utilizes computational resources which are currently not
`exploited by other interfaces.
`
`IMPLEMENTATION
`A basket is any item on the screen which can receive a
`tossed object. Receiving an object involves performing
`some action on that object, i.e. erasing a file, copying a
`file, or writing a file to disk. Our interface is designed to
`allow users to create baskets of their own, and to create
`actions to be performed on received objects. A ball is any
`item on the screen which can be tossed. Note that an
`object can be both a basket and a ball.
`The main difficulty in tossing is deciding which screen
`object is the intended target of the toss. We have imple-
`mented a Basket Manager for this purpose, which tracks
`all potential targets, and chooses the most likely target for
`a toss. A ball, target pair is then passed to the Animator
`routine, which forces an animation path for the ball to end
`in the target. Thus, the user never misses completely, but
`he may throw something to the wrong target. Note that if
`one basket is active, the user will never miss, no matter
`where the object is tossed.
`The problem of tossing an object to an incorrect target
`
`Petitioner Samsung Ex-1025, 0001
`
`
`
`Accuracy Testing
`
`100.0
`
`95.6
`
`90.8
`
`100%
`
`50%
`
`0%
`
`Completion Rate
`
`3
`
`1
`
`2
`Number of Targets
`were in the general direction of a target, but were exagger-
`ated in order to create interesting animations, or to insure
`selection of a particular target. For example, a toss to the
`lower left of the screen selected the lower right corner tar-
`get in the two target demo, and created an interesting ani-
`mation in the process.
` Apparently, users got better at
`tossing as they used the system, for the three target test
`shows an increase in accuracy over the two target test.
`Physically manipulating the mouse to indicate a toss is an
`action that comes more easily to some than others. Some
`people found the system intuitive and easy to use and con-
`sidered it a good idea, and some had a difficult time with
`the mouse, or didn’t understand what the system was
`doing with their gestures. In general, however, the users
`felt tossing worked well for one target, but found that the
`system did not always detect a toss gesture.
`
`CONCLUSION
`We have implemented the general action of tossing, and
`have explained how it may be used in several specific and
`general interface interactions. It seems that the tossing
`mechanism could serve as a useful option for novice users
`in current interfaces, and could be used more extensively by
`expert users who are able to gesture more accurately.
`This interface was written entirely in tcl/tk and is available by
`anonymous ftp from ftp.cs.unm.edu.
`
`REFERENCES
`[1] Bay-Wei Change, David Ungar, Animation: from Cartoons
`to the User Interface. UIST ‘93.
`[2] Ken Perlin and David Fox. Pad: An Alternative Approach to
`the Computer Interface, Proceedings of 1993 ACM SIG-
`GRAPH Conference, 57-64.
`[3] Bederson, Hollan, Perlin, Meyer, Bacon, & Furnas. Pad++: A
`Zoomable Graphical Sketchpad for Exploring Alternate
`Interface Physics, Journal of Visual Languages and Comput-
`ing, in press.
`[4] Lamping, Rao, & Piroli A Focus+Context Technique Based
`on Hyperbolic Geometry for Visualizing Large Hierarchies,
`CHI’95, 401-408.
`[5] Baecker, Small, Mander Bringing Icons to Life, CHI’91, 1-6.
`
`would render tossing useless if it were used in a situation
`in which an incorrect target choice by the system produced
`undesired or dangerous results. The drag and drop mecha-
`nism is successful because it is explicitly directed, and
`there is little chance of a missed target. If we are to con-
`sider tossing as a useful alternative to drag and drop, we
`must demonstrate situations in which tossing is accurate
`enough to be acceptable, and where its ease of use allows
`it to be helpful.
`The default tossing environment, in which there is one
`possible target on the screen, has 100% accuracy, so the
`question is one of preference. If the user likes tossing,
`s/he could use it effectively with one target.
`A more general situation has several active targets, but
`allows the user to place them so as to maximize accuracy.
`Or, many baskets may occupy the screen, but only some of
`them may be active at any moment. During a session, a
`user may change which basket is active according to the
`specific task being perfomed. Additionally, the activation
`of targets could be made automatic, so that each applica-
`tion was able to activate targets with which it frequently
`interacted, or that the user found most useful to have active
`for that application. In a word processing application, per-
`haps the printer would be the active target, while in an
`archiving application, tossing would activate a tape drive.
`Due to both hardware and software limitations, tossing has
`limited accuracy. It works best in situations with a single
`target, or where each of several targets are all satisfactory.
`An example of this can be found in the solitaire game on
`the Macintosh Performa. Here, a card may be tossed, and
`the system will choose the best placement for the card
`from the possible legal moves. Users often don’t care
`which of the legal moves are executed, so tossing is an
`excellent interface for this game.
`
`USER TESTING
`We conducted testing in which people tried the system
`with various numbers of targets, and then were asked to
`perform two accuracy tests, one with two targets active,
`and one with three targets active. These targets were
`placed in the corners of the screen, as shown in the figure
`at the beginning of this paper. The two target test used the
`upper right and lower right targets, and the three target test
`used all three targets. Users were also asked for specific
`data on ease of use, and finally for general comments
`about the system. Of the users tested, two had used the
`system before.
`Some comments from users were -“I would like it for
`default type tasks.” “Easier than dragging it all the way
`across a screen, particularly for some of the larger screens”
`“Actually much more efficient than dragging - when it
`works” “Takes skill with the mouse ...”
`Users found that a toss for a specific target need not be
`directed perfectly towards the target. Once users under-
`stood how the system worked, they made tosses which
`
`Petitioner Samsung Ex-1025, 0002
`
`