Memory Aids

Christian mogensen@cs.stanford.edu http://www-pcd.stanford.edu/mogens/377/mem.html

[Stanford] [PCD] [Christian]

User Studies

The target population consisted of Stanford Law students. A simple ten minute interview based around the following questions was conducted with four randomly chosen students.

The interview style was free-flow - the questions noted were backups in case the subject exhausted one avenue.

The interviews were all recorded on a microcassette tape recorder. The interviewees were all given a brief introduction to the survey before questioning started. Two interviews were with individuals, one was a dual interview.


Analysis

There were three broad types encountered in the survey of four Law school students:
The Organizer
Keeps track of everything in a notebook or organizer that is carried everywhere. New data are immediately noted in the book. Checks the book religiously for appointments and reminders. This sort of person is very structured.
I take a lot of notes... lots of notes. Write things down a couple of times. Write them down a couple of times, stick them in the computer after that...
[Computer?]
Just trying to organize them in a more logical format, than my notes, 'cause my notes tend to be...
[later]
I keep all my paper notes in binders, divided into classes.
Mr. Memory
Remembers everything in long term memory. Does not use external memory aids. Appointment are put in long term memory store by rehearsing or by writing the information down on paper. The paper notes were not used for later recall.

The Mainframe
The day's schedule is noted in a special place and checked every morning. New items are recorded on paper scraps and added to the central store at the end of the day. During the day the student relies on long term memory of the day's schedule. Semi-structured.
I just read them [my schedule notes] before I leave home.
The PDA should support remembering appointments and organizing schedules in order to be useful to law students. Given the limited writing space on both PDAs, they are not going to be used for writing essays or taking class notes - the keyboards are too small and slow to type on. The primary activities left are scheduling and random note taking.

Students took varied but simple approaches to the problem of remembering. No one used voicemail or answering machines actively for scheduling. E-mail was not a large factor - it was checked infrequently (average twice a day: morning and evening). Note: This may change as the law school is now promoting Macs and e-mail to the Law students.

Comparative Analysis

The Psion can be summed up in one word: simple. The MagicLink is rich.

The MagicLink is a varied and extensible environment which tries hard to convince the user that the objects on its screen are real. It almost works. Paper notes scatter about the desk, rabbits run loose, bells chime and windows float everywhere. The hand of Don Norman is clearly visible as much of the interface is clearly visible on the screen.

For example, help is always available from the "circle ?" button in the upper left, the current action is always undoable by tapping the go-back arrow in the upper right corner. Buttons click and drawers open. It is cute and compelling, and rewards exploration by revealing such nifty gadgets as a spreadsheet in a drawer and a card game hidden in a cupboard.

The primary functions of the MagicLink are found on the desktop: note taking (in the notebook), tracking addresses (rolodex) and appointments (calendar), and sending messages (the phone/message pad). Other functions are hidden away in drawers in the desktop or in the corridor. The filing cabinet makes it easy to file things away, but makes it hard to put things in specific folders. The drag and drop metaphor only takes you so far.

The lack of a keyboard is not felt until you attempt to make a note longer than a sentence. The Link is also pretty heavy and eats batteries fairly quickly.

The Psion 3 is flat. Compared to the MagicCap interface it is downright dull. No flashy animation, no spiffy desktop graphics, but a distinctly unfussy interface. The MagicCap interface is attention getting, constantly prompting and fussing over the user. The Psion just gets on with it.

The minute keyboard is a small improvement over the on-screen keyboard of the MagicLink. It's better because you can use two fingers to type with, it has a shift key and there are actual keys to press.

The Sony is very good at scheduling - unfortunately it is too big to carry in a pocket (a separate carrying case is available), and the screen will be scratched if it is flung into a carryall. The Sony is obviously pitched at someone who has more time and space than a hurried student. The message function in the Sony is easy to use, but not immediately applicable to law students. They are quite happy to check their e-mail once or twice a day, and have regular study group appointments (a scheduling problem again). Sketching, another MagicCap forté, is of limited applicability to law students - pen and paper do the job cheaper.

The Psion is not very creative at scheduling - it can fit a very short description in for each hour. The separate to do list is very useful though. On the other hand, it takes less time to set up an appointment on the Psion. The Psion is small enough to fit in a jacket pocket, and its flip-top casing will protect it from the rigors of traveling in a student's bag. The Psion has better note taking capabilities, in that the keyboard can be used for copying down small amounts of text without too much difficulty. Sketching is non-existent on the ver. 3 Psion.

GOMS Analysis

The user's goal is to add an appointment to the scheduler. The analysis includes the steps to activate the scheduler application and make a lunch appointment (Noon to 1 pm) with Bob.

Sony MagicLink

Targets on the MagicCap interface are approx. 2 cm across. The screen is about 15 cm across and 10 vertically. i.e. on average the stylus must move 7 cm from its current position to a target. The time to press a screen button (0.3 s) is more than the time to press a keyboard key (0.1 s from Card) because the whole hand has to move (Fitts law: 5 cm/1 cm= 0.3 s).

Also note that system response time can fluctuate a lot, depending on the amount of data in the system and the most recent tasks performed.

Total time: 2.5+2.4+9.0 = 13.9 s

Psion

Since the keyboard is fairly small, I take the keypress time somewhat higher than the standard value: 0.2 s per keypress rather than 0.1 s.

Pressing the hotkeys located under the screen requires a Fitts move (4 cm roughly to the key row, target size: 1 cm) and a press, giving a hunt time of 0.3 s, and a peck time of 0.2 s, in total 0.5 s per hotkey. The peck time is 0.2 s because the keys are harder to press than the normal keyboard keys.

Total time: 2.9 + 1.2 or 0.6 + 0.5 or 2.4 = 4.0 s (best) or 6.5 s (worst)

The Psion, while more complicated from a learning standpoint, is much faster than the MagicLink simply because it does not take so long to load its software components. Entering text is a Fitt's law task on the MagicLink, but only a slow keyboard task on the Psion, making the Psion much faster for text entry. The nifty interface of the Sony is simply too cumbersome for this sort of operational task analysis. i.e. GOMS does not tackle explorative interfaces well.

The MagicLink is much easier to figure out by poking at the screen compared to the Psion.

Activity Theory Analysis

The Psion aims at emulating the praxis of PC users. It follows the same Microsoft Windows-like GUI conventions with separate applications, dialog boxes, scrollable lists and file names. This set of conventions are adapted to the small environment in interesting ways: all applications are always loaded, no waiting for the hard disk to boot; you can jump between applications using separate push-buttons (similar to the meta-buttonbar in newer Microsoft Office products).

The applications mimic the large PC based applications to some extent, so the Psion includes word processing, scheduler, database, spreadsheets and a programming language. Operations on files are similar to using a DOS shell environment. Highlight the file you want on a command screen, and then press ENTER to load it.

Operations within applications mirror their PC counterparts: press the menu key to choose a special operation, enter text through the keyboard. Text editing (an activity) is similar to editing on a PC: cursor keys to navigate around text, shifted cursor keys to select text and so on (operations) - very similar to the Windows environment (another instrument).

By looking at the activities and operations the Psion facilitates, we discover its intended audience: people who are familiar with PCs, and who need to use similar functions while on the go. This market covers most of the knowledge workers today. PCs have become inescapable. People who are unfamiliar with PCs will have some trouble with the Psion, but anyone who has used Windows will feel right at home.

Normanic Analysis

The Sony MagicLink interface embodies a lot of Norman's ideas on iterface. This follows from the idea of the manipulable interface. If it isn't visible, you can't manipulate it with the pen. Hence things which are invisible on the Psion (like the Help menu) because they have a separate key on the keyboard, must be made visible on the MagicLink's slate (the circle question-mark opens help).

The conceptual model that appears to underlie the interface is one of making things physical: things can be dragged around and dropped into containers. Above all, things must be simple. These goals conflict, since physical things can become quite complex. Hence it must abstract away as much detail from these things as possible without reducing them to icons.

Creating a letter therefore involves

These operations are all straight forward and very visual. There is a gap between the concept of "Opening a drawer" and "tapping a pen on a picture of a drawer", but once the idea of "tapping" as an equivalent to handling (opening, pushing, pressing) something is learned, the gap is easily bridged.

Manipulations sometimes map successfully in the MagicCap world. Pointing at an object and dragging it around usually moves the object around with the pointer, as if it has been picked up. The object can then be dropped elsewhere by raising the stylus. The image of the object will follow the dragged pointer, providing useful feedback about what the pointer has picked up and where the system thinks you are pointing.

Error Analysis

MagicCap has a few problems - not all its objects can be dragged as you expect, and not everything can be dropped where you want to.

The classic HCI flaw "the keyhole property" strikes the tote-bag in the middle of the screen. The idea is that things can be stuffed into the tote bag for later retrieval and then later taken out somewhere else. The tote-bag indicates its state by becoming one pixel wider when not empty. This is not a useful idicator of what is inside. Since the tote bag contents cannot be seen as we navigate around MagicSpace, we must remember what we put in the tote bag and where we were going to put it.

You can therefore easily forget that something is in the tote bag and then later forget why you put it there. The tote bag should become more assertive in showing that it's carrying something, so you won't forget that you are carrying something. The tote bag should ideally give you a hint as to what you put in it last, so that the keyhole is made larger at least.

The problem is screen space again. Where can we fit a description of the last thing in, without losing the totebag? Also, how do we show the last thing inserted in a meaningful way? The things in the totebag are not named, and there isn't room to show a full sentence or much beyond a single word..

The whole MagicCap interface can be seen as a solution to the keyhole problem: by leveraging the understanding that people have of rooms, corridors, houses and streets, the interface make it obvious to everyone how different rooms link together, and how to move between them. The interface then extends this model to looking close up at things (notebooks, postcards, etc.) to give us the zoomed in views.

The PDA Market

The Sony is too far ahead of its time. As noted earlier, it has too rich an interface for use on the go. You need time to use the Sony. For a bit more money you can buy a Mac Duo that does much more and is just as portable. In reality is hobbled until the wireless e-mail and fax capabilities become available. Until then it can't substitute for the office Mac (not powerful enough), nor can it be used as an organizer: too slow.

The Psion is a very nifty design. It is a 'real computer' since you can write programs for it, it comes with all the essentials built in, and most of the office workers of the western world will recognize the interface. The Psion can actually be used for taking quick notes and can cope with skipping back and forth between the many tasks that crop up in the real world. Anything can be interrupted and resumed later. Finally it is rugged and cheap compared to the Sony. It is designed to be a portable computer.

The Sony is good for impressing friends with, while the Psion would be the one you actually used daily. Of course, all these pocket organizers have to compete with paper filofaxes and their ilk. The Psion has the edge over an organizer with its spreadsheet and wordprocessor. Of course, people who aren't likely to use organizers are not likely to use either of these PDAs unless they offer really big advantages.

(Ever notice how people with organizers also seem to own pagers and mobile phones? MagicLink is obiously waiting for the world to catch up.)