The BT DF200 fax - Computer Science
Contents
1. does nothing on my DF200 which is connected to a public exchange transmission speed The key has no legend to suggest this alternate purpose and the fax s display does not show even the possibility of the option Given that protecting the fax against echoes is likely to be an infrequent action but one that is occasionally necessary we can assume the user does not remember how to do it without prompting the DF200 can only be said then to conceal this feature lt 18 gt 54 55 56 57 58 59 60 The volume can be increased beyond the loudspeaker s power rating as can be verified by increasing so that bleep tone distorts When paper is placed in the FAX slot the LCD shows NORM This does not mean ready for normal FAX transmit but normal resolution as opposed to GRAY or FINE When the FAX is being transmitted the DF200 says NORM NORM NORM or TRANSMIS including the dot alternately Is there any difference between NORM and NORM NORM After one occasion my DF200 was repaired these LCD displays changed toA4 NORM etc even when the paper being transmitted was not A4 Of course it is possible that my repaired PCB was a recycled older one and later design revision removed the superfluous and often incorrect A4 part of the message Whatever the reason there are still peculiar design choices that seem inexplicable the more so given that the differences between my original and repa2. after just 26 presses After 27 presses the user of the DF200 has a slim chance of accessing the alarm command after just 2 presses on the function number interface the user has a certainty of accessing some command and therefore of obtaining an example of how all commands are accessed A user can play randomly with the interface and learn how all of it works this is not possible on the DF200 and the user would have to be quite persistent to get any success from which to generalise somehow its mode of operation So there are some user interfaces that are better than the DF200 even if the user does not know how they work A city is not a tree Christopher Alexander s classic paper A city is not a tree Alexander 1965 makes an eloquent argument for not designing cities as trees but rather as semi lattices A tree like organisation isolates activities industry education health housing into separate areas without overlap This means people have to travel between areas and their lives become compartmentalised with increasing problems as they become older and isolated into regions specialised to old peoples needs However the tree structure suits designers because of its conceptual simplicity We see a similar effect in the design of the DF200 Its functionality is organised as a tree with each function isolated into its area Unfortunately the area it is isolated in is the specific and unique area the desi
3. back when the button is pressed When all messages have been played there is a final tone and the LED stops flashing and pressing PLAY has no further effect unless SAVE is pressed There is a problem if PLAY is pressed before the tone finishes the LED continues flashing indefinitely and pressing PLAY shows MESSAGE No 1 in the LCD so there is apparently a message but no message is played back This gives the impression that the user s actions have erased the message there is a message but it is cleared or if some other user was to use the fax it gives the impression that someone rang but left no message Neither of these interpretations are correct It is simply a DF200 bug The manual describes what the LCD displays It is rarely accurate perhaps partly because the LCD itself is short and often shows abbreviated words 41 The inconsistency between the DF200 and its manual is mentioned again for completeness General inadequacies 42 43 44 45 46 47 48 49 50 51 52 53 The DF200 and its extemal power supply overheats at standby it consumes 6 25W it gets rather hot for a user to place it on their lap The DF200 has a Ringer Equivalence Number REN of 1 5 when an ordinary telephone without benefit of a separate power supply has a REN of 1 A BT line is limited to a total REN load of 4 The paper support is flimsy and doesn t fold flat so it and its flimsy hinges can break
4. can transmit but cannot receive at 7200 baud When the handset is down it is possible to key a number which is displayed in the LCD panel but is discarded when the handset is lifted However in the same situation the one key abbreviations beep i e it is an error to press them when they could more consistently have shown the number they would have dialled out had the handset been lifted Without performing the print command which wastes paper and prints all one key abbreviations it is not possible to determine an abbreviation s full number without actually dialling out After the answerphone records a message either from a caller or from the user the LED on the button flashes When is pressed any recorded messages are heard and the LED stops flashing when all have been listened to or skipped If a new message now arrives earlier recordings will be erased unless SAVE is pressed When SAVE has been pressed subsequent messages are appended to the list of recordings however the LED does not flash in this case Thus there may be recordings whether the LED flashes or not Imagine recording a message and checking it sounds OK Doing so stops the LED flashing and thus whoever is supposed to take the message will not know one has arrived When the answerphone has received a message and recorded it the LED in the button flashes Flashing indicates that there are one or more recorded messages which will be played
5. inconsistency and general inadequacies we do not repeat structural design issues that were covered in the body of the paper Within these categories expanded below the lists are in no particular order In principle a cost benefit analysis can be made comparing the cost to users against the cost to the manufacturer The list of flaws might then be ordered in decreasing impact However in the flaws we list below the cost to the manufacturer is only in the effort of designers thought not in materials thus the cost would have been amortised very quickly and might well be considered trivial To facilitate reference the points are numbered Apparent lack of imagination 1 There is no UNDO button If the user notices a mistake they cannot do anything about it other than press STOP and start over again Some examples are given below where the consequences of a mistake are considerable such as having to re enter the entire fax set up and telephone numbers 2 Unlike many other faxes and despite being able to print all sorts of technical information including some related to non functional features the DF200 cannot print a command help summary One would have been very useful Indeed the manual p 20 suggests making a copy of part of the manual Ironically although the DF200 does have a photocopy feature the manual is bound in such a way that the DF200 cannot be used to photocopy its own manual 3 When the fax is being used
6. it might be conveniently implemented in this fashion if so this would explain the incoherence of the design Better programming practice suggests instead factoring out abstract operations e g what each user action does and implementing them top down This approach might have led to a collection of functions implementing buttons and a data structure representing the state space Done like this it would have been quite hard to make the user s actions do different things in different places in the state space When people design programs standard practice is to define abstract types operations modules the details vary but generally the idea is to design top down and to rigorously specify what is intended so that the outcome may be proved and or debugged against the initial ideas The reasons for doing so are to better manage complexity The design of the DF200 suggests that the user operations were not considered as part of the program design moreover the program design was probably such as to conceal any relation between the program and the user s language for operating the device There is a vast literature on algorithms The user s task so far as analysed in this paper is searching 2 The DF200 has a database of commands and to perform any task the user must locate the corresponding command The designers of the DF200 chose to use a peculiarly structured tree with no obvious advantages we showed that its maximum depth av
7. position to advance Digit is not marked as generating any alternate characters but in fact it generates 11 symbols space and brackets the manual is not specific Why space when this might have been inserted with in fact now moves right which only sometimes has the effect of introducing a space Why not put some of the characters in 0 which does nothing other than insert 0 Star moves the cursor left without deleting and moves right The symbols don t have any natural association with left or right motion though the is to the left of the key To delete a symbol from the LCD when permitted the decrease volume button which is not labelled as supporting this purpose is used Decrease volume VOL V deletes the last symbol and moves left but VOL A does nothing other than beep Except that if there are characters Pressing 2 2 2 2 2 2 2 8 8 entersa BAT lt 16 gt 34 35 36 37 38 39 40 to the right of the entry position the cursor then VOL V deletes all characters to the right and does not move left sequences in fax speed setting mode however it is used to protect in an unspecified way the fax from echoes whether a user would be likely to remember this hidden solution to an infrequent problem when it was needed is another matter The DF200 can transmit faxes at four baud rates but can only receive at three it
8. the LCD panel often shows the text REPLACE HANDSET Under these circumstances the user is holding the handset likely because they want to make a call and evidently the fax knows the handset needs replacing e g because the last called number has disconnected so why doesn t it simply perform whatever electrical operation is required itself and save the user the inconvenience It seems to me as a user that the fax is being terribly condescending as if it is saying Your last call has finished and you now want to start a new call In gt 10 Put the handset down and pick it up I could do it for you but you ought to know that s the proper way it should be done When my DF200 broke and had a PCB replaced I acquired someone s a certain A H Bell s phone number the phone numbers of their contacts and their Mercury number which is a charge code giving access to an alternative network The DF200 was able to print a list of the last 17 faxes transmitted and most of the numbers of the 13 faxes received I expect A H Bell would be interested to know all this When I rang the BT service to report the DF200 s fault the first thing I was told was to write down some instructions put the phone down carry out the instructions and in a few moments they would ring back Why can t the DF200 print out or fax its system parameters while holding a phone conversation Why are the instructions for printing out these impo
9. undefined the manual does not mention the third button and it appears to do nothing The handset itself also provides a button in that the functionality of the fax changes when the handset is lifted There is a loudspeaker which is used for monitoring the handset ear piece listening to recorded messages and for making a small selection of beeping noises capable of displaying 12 upper case letters and some punctuation such as as well as space 3 For example I omitted all details of the DF200 debug and trace commands which in any case are not documented lt 93 gt To access commands the user should proceed as follows which quotes verbatim including original emphasis from the user manual s explanatory chart Press OPTIONS and the press the number on the telephone key pad relating to the programming trail you want to follow At each successive stage of your trail press OPTIONS to scroll through the list in the box and press STARTICOPY to verify a function choice or continue to another box Where a box not J An example changing the dialling type from PULSE to TONE has been highlighted for you to follow START COPY again and then OPTIONS to change to tone dialling Press and the red STOP button to end the programming sequence Note Your DF200 can remind you which trail to follow so that you not need sic continually refer to this chart At the start of the programming sequence addi
10. we do not change the number of functions nor what they do we only modify by and we show what considerable variations are possible In the absence of empirical evidence the analysis below assumes every function has uniform weight This shows how such analysis might be of benefit even before a novel system has been built and hence one where empirical work would have been difficult lt 3 gt Analyses We consider three simple tasks how long it takes the user to perform a command knowing how to do it how long it takes to find a command and the worst case time to perform a command By time we are taking T as giving one second per key press Design Average time Maximum time Complete search Actual DF200 5 86 13 117 Linear search 24 5 49 49 Direct function 1 9 2 1 9 Binary tree 5 69 6 53 Basic trie 9 29 17 49 Hybrid trie 3 18 5 49 These figures should be read in conjunction with the explanatory text For example the rather high average cost of the linear search design is the consequence of having a very simple user interface which might be a worthwhile trade off Function access on the actual DF200 The average cost of a function access is 5 86 presses error free provided the user knows how to access the function the worst case is 12 presses from standby or no more than 13 presses from anywhere These figures assume the user knows how to get to the function perhaps by following exact inst
11. The BT DF200 fax Harold Thimbleby This is a very rough example of what CSM19 coursework might look like except I ve done far far too much work to show how many opportunities there are for analysis and criticism even for a simple device And I haven t covered everything that might be said at this level You should choose what you want to say and what topics you want to be interested in for your coursework If I was marking this example coursework I d criticize it for not referring to much literature or stuff on the web about relevant devices or designs It rants too much but it s hard to be the author and judge of that This report has no empirical evaluation in it based on anything to do with users there is no think aloud no cooperative evaluation no cognitive walkthrough no task analysis you have to judge whether to get carried away with analysis programming or user studies depending on your chosen device task Also I haven t given any detailed examples of the analysis or adequate data I could have written a program and used the program to print out stuff or I could have included a listing of the program or its data or I could have included a user protocol or questionatrre Abstract We argue that button controlled devices can be made better to improve their usability and that there are routine ways to do so that can be effectively employed in the early stages of the design process We examine the design of a d
12. a is of course that T b xc is a measure of the cost of accessing feature f From these considerations the average badness of a design measured as a time is T b e gt where the sums are over all functions f provided by the device A designer should seek to minimise this measure to make the design more efficient However the average cost is not the only measure of design quality For example max T b xc is the worst case Average and maximum are just two of many statistics we might consider which statistic measure the designer should optimise depends very much on the users tasks and environment More buttons mean a more complex physical space for the user and fewer buttons would mean a more complex logical space for the user properly weighing such trade offs rapidly gets into deep psychological waters and generally will raise design questions that cannot be answered a priori which is one of our aims All such considerations introduce uncertainty and make it harder to compare designs Clearly a different set of functions influences the relative cost of the functions to the user Rather than get bogged down it details it is easier to take T and c to be constant We then obtain the simple average Ds n where n is the number of functions The point is you should not simply increase the number of functions but should also consider the structure of the design and hence modify bp In our examples below
13. ce easy enough to manually explore systematically with any hope of certainty The analysis nevertheless showed quantitative inefficiencies and indeed confirmed the difficulty of a systematic exploration of the DF200 by hand Besides critiquing specific details of the DF200 design we have shown that alternative designs would have offered quantifiable usability trade offs We have shown that basic computer science namely designing algorithms for the user can be used to improve user interfaces using routine computer science we showed that useful usability analysis for a real device is feasible and results in productive design insights Acknowledgements Matthew Jones and Gary Marsden provided many useful insights References amp further reading C ALEXANDER A city is not a tree DESIGN 206 pp46 55 1965 BT undated DF200 User guide Apparently coded UM DF200 23175349 2 British Telecommunications plc London S J GIBBS amp D C TSICHRITZIS Multimedia Programming Addison Wesley 1994 D A NORMAN The Design of Everyday Things Doubleday New York 1990 H THIMBLEBY Press On Appendix 1 Basic features of the DF200 The DF200 is a small desktop fax unit with an external mains power supply The fax has 39 buttons 12 for dialling out 10 for abbreviated numbers i e for quick dialling 7 for controlling the answerphone recorder 3 for works one only works with a local switchboard but its functionality is
14. choosing voice or fax numbers appropriately Why not permit the programming to extend to other keys than just dialling If this was done the user could define one key to activate the alarm one to change the number of rings before the answerphone activates one to print out the programmed keys meaning which currently requires several commands and so on Thus whatever the user wanted to do frequently could be personalised and made very easy There are 16 error codes described in the manual p 25 These codes are numbers and or letters and seem completely arbitrary Why doesn t the DF200 use words for them when the errors occur It manages to show words like FINE at other times The DF200 can define one touch keys so that the user need only press a single button to dial a number It is not possible to dial a number and then have a button defined to dial it even though there is a that can redial the number immediately In other words the user cannot check that a number is correct before it is defined The REDIAL button does not behave the same way as dialling a phone number it only works when the dialling is direct to an exchange In summary it is possible to store frequently used phone numbers so that they can be recalled by a single button press at a later time It is not possible to store the last number dialled even though it is displayed on the LCD and the DF200 clearly knows it If you wish to store a numb
15. ction meaning they are then able to enter what specific data they wish as opposed to activating a function when the fax acts on the user s already entered data In all cases on the DF200 activating a function takes one more press than accessing it and entering the data or selecting from a menu Numerical key counts unless stated otherwise refer to numbers of key presses required to access a function starting from standby mode To illustrate the DF200 has an average function access of 5 9 key presses but a complete search cost of 117 key presses In other words the DF200 favours familiar users for they can achieve an average performance ten times better than users who have to search for commands The complete search cost of 117 is an under estimate of what a user would be expected to achieve on the DF200 for the search task is so complex that users would be expected to make many errors and so take more key presses than the minimum An alternative design proposed below has cost of 24 5 and 49 respectively This alternative on average is harder to use for any task by about 19 presses so it is over three times slower for familiar users On the other hand if the user is unfamiliar with the device on average they will only spend 24 5 presses searching for what they want which is a significant improvement over the DF200 which requires about 140 greater effort Moreover this alternative design is extremely simple and therefore has me
16. ding fax machine Worse after manual intervention to stop reception when a new roll of paper is inserted the DF200 takes the opportunity to report the reception and transmission logs You can explicitly request these at any time when they are wanted but to take the time to print them just when you are trying to continue receiving a fax is irritating The reception log reports the manually terminated reception as Code 06 Printer default which means printer incident during transmission according to the manual p25 My emphasis When paper is inserted to prepare for transmitting a fax the LCD changes to A4 NORMand it is possible to enter the number to be dialled which is then displayed in the LCD panel Yet is it not possible to send the fax lt 12 gt 11 12 13 14 15 start over I make this mistake being led down a garden path frequently When a double sided sheet is being transmitted obviously two steps are required First one side is transmitted then the other However the DF200 terminates the fax connection before the sheet can be taken out and put back into the intray There are ten buttons that the user can program to generate telephone numbers The keys also have a paper legend that the user can write on as a reminder of the keys meanings A nice feature is that the number transmitted changes if there is a piece of paper ready to fax this enables the same button to be used to call someone
17. easily The DF200 clicks despite digital recording technology The handset microphone distractingly picks up the sound of breathing We often find the DF200 shows that it has recorded a message but which in fact is null The FAX transmits images on paper placed face down in the slot So if the sheet to be sent has the recipient s FAX number on it it cannot be read A fax should transmit images face up and the DF200 should have some reminder on it that the image should be sent face down The DF200 picks up radio interference All the button labels are in capitals apart from fast forward and rewind which use abstract symbols volume increase and decrease are called VOL V and VOL A This may give the fax a technical appearance but most evidence suggests that lower case lettering is easier to read The button RECALL a name confusingly like the button REDIAL has no meaning except when connected to a private switchboard The DF200 provides three commands that have no functionality at all PRINT and PERFORM the manual does not define these commands BT s phone helpline said they had never before been asked by a user what they do BT said they are software built in the machine which it doesn t provide sic I think they may be a relic from the related but more sophisticated VF800 fax specify RECALL beyond saying it activates unspecified functions from a local switchboard thus the button RECALL
18. end on an exact 1 BT is a name of British Telecommunications plc J1b gt understanding of the actual DF200 at least any errors in my understanding will be consistent and will affect all examples equivalently The DF200 functionality and its analysis The DF200 has 49 functions of interest to us organised as a tree i e a hierarchy of choices giving access to options and data entry points The DF200 has several other functions such as standard telephone functions which we will not analyse The basic structure is easy to analyse for our purposes a small Finite State Machine is adequate Our analysis suggests that alternative organisations may have been preferable Our method is very simple we calculate the cost as a number of the user doing tasks with the DF200 and compare the numbers with the corresponding costs of a alternative designs that support the same functionality as the DF200 Though we will not modify the original functionality for our analysis we should point out that the functionality is questionable For example there are three types of print summaries for phone number abbreviations These and other functions might have been combined providing a simpler and more rational design Definitions Various functions on a device take differing numbers of keys to perform Entering an identifier takes two or more key presses depending on how long an identifier the user wants to enter We say that a user accesses a fun
19. er you have to enter it save it then check it it is not possible to save a number which following a successful phone call you know is correct I answer the DF200 and it is my wife asking if there are any messages recorded for her The answer is yes for the PLAY LED is flashing If I press the PLAY button we can both hear the error beep of the DF200 but it is not possible to play the message Instead I have to find the DF200 manual read the programming the DF200 for remote use section to her and leave her to ring again using the command sequences dial number press repeatedly until DF200 beeps Unfortunately my DF200 is set up with an access code the manual does not say how to enter the code It says After you have sent your access code and pressed the DF200 will immediately begin to play any new lt 13 gt 16 17 18 19 20 21 22 23 messages This seems to imply that the access code is entered before the The DF200 has two STOP keys Apart from one having a LED indicator I cannot see what their difference if any is The potential advantage of this design is that each STOP key is in a group of keys that activate functions that might require stopping one STOP is close to the answerphone playback keys the other is next to the START key which starts almost everything The DF200 has a paper tray so that several sheets of an out going fax can be held Unfortu
20. erage depth and total path length were all sub optimal It seems the operations available to the user to support any search were simply not considered Every book on AI or algorithms of which I am aware has a section on searching and all of them provide better ways of doing it than the DF200 employs The designers of devices like the DF200 seem to be unaware of standard algorithms They are certainly unaware that they ought to provide users algorithms or allow users to use effective heuristics at least as carefully chosen as they would chose when designing the insides of the device It seems the DF200 user is not treated as well as a computer would have been to do the same tasks Surely the user deserves to be treated at least as well as a computer Limitations to the approach There are several limitations to the methods presented including The model of the DF200 was approximate The analysis has no empirical foundation 2 It is still called searching even if the user knows where to look lt 38 gt The analysis assumes error free behaviour The analysis ignored time The analysis was trivial The analysis ignores interpersonal differences psychology social context Conclusions The DF200 shows a degree of arbitrariness that makes its analysis awkward In fact we performed an analysis of a simplified machine because the actual machine was not documented clearly enough to understand its structure nor was the actual devi
21. gner thought appropriate Unfortunately the user may not see it like that In the case of the DF200 the designer decided that printing was one area and defining one touch fax numbers another separate area How then does a user print one touch fax numbers In the print area The designer decided that setting the machine to behave as an answerphone would reside in the initialisation parameters answer param answer area but to record the lt 7 gt answerphone s message which a user might well wish to do at the same time resides in answering announcement recording somewhere completely different Similar points could be made about access codes abbreviated key names baud rates and so on If the fax was a city Alexander would recommend it was a semi lattice This would be appropriate for a fax a function such as print one touch fax numbers can be in both print and one touch fax number categories and the user could find it easily however they classified it The drawback of this organisation is that it increases the number of key presses required for a systematic search when the user has no idea how a function is classified though the trie approach demonstrated that the user interface can still appear very simple even if trie is not a familiar name to the user The computer science view If a program is implemented as a transition network then transitions can be added in ad hoc ways The DF200 is a simple FSM and
22. ired DF200 prove that the manufacturers can modify the design yet whichever is the more recent design can t be said to be better Why doesn t it say TRANSMITTING as the LCD is long enough An option allows the user to switch off FAX transmission status reports However despite switching it off the DF200 will still provide summaries are a curious mixture of ambiguity jargon and unnecessary abbreviations It also provided the report after 7 transmissions which seems an odd number to me Although the DF200 can clearly print Correct and Manual Call and so on the status column occasionally says Code 03 or Code 07 Why can t it put these codes into words too When the report says subscriber has stopped does that mean us or them What does Disconnected 210E000000 mean a code that has occurred during normal use of the DF200 The sequence OPTIONS accesses a list of editable configuration numbers Neither the command sequence nor the meaning of the configuration numbers is defined
23. just the user s direct actions A different sort of example is represented by the DF200 s REDIAL The designer has made the assumption that redialling the last number is an activity the user will often wish to do because the number might have been engaged in fact because this prediction is so specific it can be achieved with a single button press It is interesting then to observe that the same reasoning did not lead to a REDO command that not only could redial but could take the user back to the last command they attempted for instance just in case the user wanted to check their changes or to correct them Error prone use The previous analyses assumed error free behaviour Of course for key press sequences of 12 even if the user knows what is going on there will be times when the wrong keys are pressed Suppose the user does not know where a command is and is prepared to search randomly on the wise assumption that an intended systematic search might forever miss a command if the user did not understand the system s actual structure Take the alarm setting command on the DF200 which is the easiest command to access 3 key presses pressing keys at random requires at least 27 presses to have a better than even chance 50 08 in fact for exactly 27 presses of finding the alarms function Whereas if the direct function number interface was used see above instead any command could be found with better than even chance
24. ker replaces the earphone The feature supports hands free dialling MONITOR may be pressed when the handset is actually lifted but rather than amplifying the earphone to the loudspeaker one merely achieves clicks and beeps One of the potential monitoring functions enabling more than one person to listen to a phone conversation is therefore not possible monitor or not monitor all incoming calls answered by the answerphone are monitored It is not possible to set the DF200 so that it silently answers calls For example it is not possible to stop anyone in the same room hearing the phone message being recorded If paper is placed in the in tray the LCD displays A4 NORM If MONITOR then REDIAL are pressed the DF200 dials the last number used and sends the fax However if REDIAL is pressed before MONITOR the DF200 enters an undocumented mode called TRANS PARAM and then asks for the number of pages correction mode whatever that is transmit speed and then does nothing The DEF200 has eight digit keys with three to four letters 2 ABC 3 DEF 4 GHI JKL 6 MNO 7 PORS 8 TUV 9 WXYZ Digit 0 generates no special symbols The user enters an ID by pressing digits and to move to the next column though this extra meaning of is not shown on the button In other modes behaves like a digit it is a code that is transmitted to exchanges and pressing a number automatically causes the
25. nately if there is any paper in the tray no in coming fax will be received The intray is not as wide as standard US paper The paper guide only permits sheets of width 20 7 to 21 6cm With little effort the fax could have handled much narrower paper such as A5 The DF200 will transmit faxes with headers Whether a header is sent is an option classified as transmission param header transmission with without but the header itself is classified as user param id not as a header or header text The possibility of receiving faxes with and without headers is summarised in the manual but is nowhere explained There are several such discrepancies between the summary and the body of the manual loudspeaker without picking up the handset Either dialling or pressing REDIAL and the dialled number is displayed in the LCD panel This is useful as it gives feedback to the user what number is being dialled However when the handset is lifted the LCD displays ON LINE as if the user didn t know and ceases displaying the number dialled The DF200 can print status and other information on its paper However it leaves large gaps around 5 5cm between separate items This wastes the expensive fax paper Although the standby mode continuously displays the date and time in the LCD the time display is not available when the DF200 is doing anything else Thus when using the phone to hold an expensive international conversation it is not
26. nd realising prefix These modifications would bring the average access down to 1 9 It takes a maximum of 2 key presses to access a command this implies sometimes the user will have pressed only one key and not completed the command access What should the fax do in these circumstances The naive solution is that after a time out the DF200 takes 30 seconds the fax would revert to standby But it is also possible that the first key press was accidental should the user wait 30s for this error to be rectified What happens if the first press was an accident and went unnoticed by the user A better solution is to arrange that the command accessed is the result of the last two key presses as opposed to the precisely two key presses counting from standby This ensures that the user need never worry what an earlier key press was Regardless of whether the fax is in standby the user can enter two keys for the required command and it is accessed even when there were prior key presses partially accessing a command For example suppose command 43 is NUMBER OF RINGS and the user has already pressed 5 Pressing 4 accesses but does not activate command 54 the user would have to press STARTICOPY to activate it and next pressing 8 accesses NUMBER OF RINGS The DF200 provides a delete key curiously a double and unlabelled meaning for the decrease volume button this key can also delete function key digits It may not seem very
27. o on playing messages 6 Does nothing Answer on complex process Record a new announcement Note complex process that there is no way to confirm the recorded announcement 9 Answer off complex process Note Pressing STOP allows Press to stop any operation When setting up the DF200 the name and number of the FAX can be defined The number of the FAX can include spaces by pressing and by pressing as well as the usual ten digits It isn t possible to have a standard international number such as 44 0 181 363 6411 because there are no brackets available In contrast the name called the ID by the DF200 can have a range of letters and punctuation In the setting ID mode i now provides punctuation including a space as previously provided by whereas now beeps and does not produce even a sign displayed in the LCD panel Yet OPTIONS 5 prints a trace of the DF200 which I cannot interpret it being mostly hexadecimal numbers This trace feature has nothing to do with erase mode lt 15 gt 27 28 29 30 31 32 33 It is possible to enter an invalid hour such as 29 00 which the DF200 will accept and treat as the current time but it is not possible to set an invalid minute such as 12 69 Hours and minutes are obviously validated in different ways pressed its LED lights The effect is as if the handset is raised but the loudspea
28. omestic phone fax answerphone machine and show that alternative designs have quantitative advantages over the original Introduction The BT DF200 is a domestic telephone with fax and answerphone features Given the DF200 as a specific device here analysed in some detail we shall show that its design is inefficient given the choice of functionality its designers have chosen for it There are many design considerations that could have led to alternative approaches e g Norman 1990 For example the volume control could have had a better affordance than the current DF200 approach of using two side by side buttons If the volume had been controlled by a slider or a knob no doubt the designer would not have been able to even consider using the volume decrease button as a sort of delete key We may assume that market expectations manufacturing costs and other conventional reasons are sufficient to justify the style chosen The FAX ASSISTANT Gibbs amp Tsichritzis 1994 a video multimedia system using an animated human assistant to help people learn how to use a fax is interesting I find the DF200 very difficult to understand and remember how to use without referring to the manual and I had to expend considerable effort in determining how it actually worked for the purposes of this paper Though my limited ability to understand the fax during use may affect detailed numerical results below the general issues raised do not dep
29. possible to either see the duration of the call or even to see the current time The LCD simply shows the number dialled or if the handset is picked up after using the hands free dialling the essentially useless text ON LINE Inconsistencies 24 The DF200 has remote facilities so that a user may phone it and listen to recorded messages and perform a few other functions The user interface for remote facilities has no relationship so far as I can see to the normal user interface to the same facilities Thus the remote user not only has none of the appropriate button labels to help no LCD display for confirmatory feedback but an arbitrary key code to remember lt 14 gt For example although the DF200 has a button which behaves as a command to a remote exchange or as a space in a identifier in a DF200 key definition the effect of pressing it on a remote phone is the same as on the DF200 The following table compares all remote commands with their equivalent DF200 direct commands where there are equivalents Remote Equivalent operation ke Meanin on DF200 Listen to a message again Press lt 4 lt Delete a message Listening to message deletes it unless SAVE is pressed Go on to next message Press gt gt Listen to previous sic message Press lt lt 1 twice again 25 26 Listen to all messages new and old PLAY pressed again defers the playing and pressed again allows to g
30. rit because it is easier to learn how to use We are interesting in measuring how easy or difficult a device is to use Define the cost to the user of activating a function f as c A function with low cost is better than a function with high cost We can measure cost in many ways cost is the converse of utility The more often a user requires a function the greater its utility The more a function contributes to the user s business the greater its utility The utility could also be related to marketing considerations there would be little point having a selling feature that salesmen found difficult to use Using the notation cy is reminds us that the 12 gt cost of an operation to the user is not known unless the subjective factors of cp are agreed and established From a design specification regardless of what a user does or prefers we can easily and objectively measure the cost of accessing a function be measured in button presses This cost clearly influences the time it takes a user to do something since the user is physiologically limited in their button pressing rate also the more buttons that need pressing in sequence the greater the probability of error We can account for these effects by a function T that maps the button press count to a time for the user Note that T depends on f it may be that the user is more anxious about the use of certain functions than others which would affect Ty differentially The ide
31. rtant diagnostic information so complex they need writing down The DF200 broke again so I had a chance to ask about the stored data from the previous customer see above I was told that I should have pressed OPTIONS 9 to reset the fax I asked why this feature was not in the manual and I was told so that users did not activate the function deliberately my emphasis The undocumented erase feature does not have a confirmation It could have said in the LCD panel If you want to loose everything press START or some 16 character equivalent that would fit in the LCD panel In fact the DF200 gives you no choice After the second repair the DF200 was left in a permanent mode where pressing printed a test pattern and pressing printed an apparently unending trace in hexadecimal and code words These keys had this effect immediately when the DF200 was in standby so the features were all too easy to invoke by mistake BT s Helpline told me to reconfigure the DF200 to avoid this problem I should press OPTIONS A fa it then shows CONFIGURATION 1 then press i to change the first 0 to a 1 In fact as I found out you should press 0 The other digits all zeroes and ones are flags for automatic tracing log printing and so forth overall there are 72 configurations 9 groups of 8 bits When the DF200 receives a fax and runs out of paper it does not terminate the reception or report an error to the sen
32. ructions in the user manual an assumption that is questionable If the user has to search for the function then a systematic search of the functions amounts to a travelling salesman problem and the user is unlikely to do this efficiently If the user knows that the functions are organised in a tree and the STOP button can be used to return from any point to standby then they might embark on a depth first or breadth first strategy to search Either approach takes a maximum of 117 key presses and on average the desired function would be found half way at 58 5 presses plus one more press to activate it If despite being so systematic the user missed seeing the required function the first time they located it they would have to go round again taking them another 117 presses 175 5 presses on average if they spotted the desired command the very next time it was accessed If they don t recognise when they visit a level for the second time they could spend forever going around in circles Alternative linear search The DF200 clearly poses serious problems for users who know there is a particular command they want but do not know where it is to be found An obvious design approach to overcome this is to organise the commands in a simple linear sequence searching for a command now becomes trivial especially if the fax is given a button called say FIND COMMAND decrease button because it looks more like scrolling than
33. the OPTIONS key but this would make no difference to the analysis On average the function the user searches for will be half way through the list So a user has to press 1 49 2 25 5 keys In the worst case they have a lt 4 very simple and reliable procedure for accessing any function in a maximum of 49 presses If the functions are sorted most frequently used first then the average can be weighted and much better results than 24 5 would be expected in practice On the other hand if the list was sorted like this then a user might despair before finding an infrequently used command that in any case they were not too sure about using It might be better to order the commands alphabetically so that the user always has a sense of progress through the list Doing so would then raise questions of judicious command names related commands could have similar names so that they are close to each other in the sorted list Alternative direct function number Suppose a list of all functions was shown on the front of the DF200 together with a number 01 to 49 Every function could therefore be accessed with exactly three key presses say OPTIONS digit digit moreover the user would not need to remember what the functions were called or how they were organised as the summary would be an adequate reminder We can do better by using a more compact numbering scheme numbering the 48 functions as 0 1 2 3 4 50 51 and all numbers to 93 a
34. tional presses of the OPTIONS button will reveal the number on the telephone key pad you need to press for each sequence We now quote from the manual page 26 to see how to perform polling If the machine you are polling from has a password see page 19 you will need to match the passwords of the two machines Polling can only occur when both machines have the same password Page 19 refers only to access code which may be the same as password here Page 19 is called Using the answering machine from another location and does not appear to refer faxes polled or otherwise If polling can only occur when both machines have the same password this appears to mean that your own fax s password has to be changed to be the same as the remote fax This would seem to compromise your own password security Press OPTIONS and then 4 The display will show POLLING BASE and SELECT alternately I cannot find an explanation of the POLLING BASE feature anywhere in the manual Selecting it results in a request for a three digit base number then a four digit password then a number then a time During this process it is easy to get the LCD to display AN OT UNDEFINED which the manual does not define in its list of error codes In contrast if OPTIONS then 4 is pressed when there is a document in the fax the LCD shows DOC TO BE POLLED a feature explained on page 27 The manual continues No
35. useful or important to correct a two digit sequence but the main advantage of this facility is to increase the frequency with which the user can use the delete key and hence to increase their familiarity and skill with it Specifically it generalises the meaning of the key so that it depends less on modes Alternative search by binary tree The DF200 uses two keys OPTIONS and STARTICOPY to access each function Two keys are sufficient to perform a binary tree search say with OPTIONS going left and ISTARTICGPY going right If we do this with a minimal height tree we find the average function access is 5 69 key presses and the maximum is 6 where over half of the commands are to be found By the simple expedient of linking every command together at the leaves of the tree every command can be accessed by getting to the say left most lt 5 command 5 presses and then continuing left from there with a maximum of 48 more presses Like the actual DF200 this scheme relies heavily on the user understanding the organisation of the tree We suggest the tree might be organised alphabetically Even an alphabetic organisation groups commands helpfully for the user thus ACCESS set free or protected is adjacent to ACCESS CODE rather than in unrelated places as on the DF200 Alternative search by alphabetic trie Using more than two choices speeds searching a tree provided the increase in choice is systematic If we use lettered ke
36. w press OPTIONS again the display will show POLLING and SELECT alternately Then press START COPY The DF200 will then ask you to enter the number of the remote unit from which the document is to be polled To say it asks you is an exaggeration as the next sentence makes clear y y lt 10 gt Display shows ENTER NUMBER and goes blank while you key in the number on the telephone key pad The text ENTER NUMBER is displayed very briefly for about 1 second Unusually the LCD does not alternate between what you are entering and the ENTER NUMBER information If left to its own devices at this point the DF200 will shortly reset to standby In other words in the time you have available a blank LCD panel is asking At this stage you can also enter a memory key see page 14 for information on memory keys In fact page 14 immediately refers to page 23 for fax memory keys which discusses them solely in terms of transmitting faxes Here we are trying to receive one Moreover you cannot enter a memory key at this stage whether before or after the number What the manual means is that a predefined memory key can be used rather than be defined We allow that this is an ambiguity in the use of the word enter which could have been resolved by better wording Appendix 2 Specific design flaws of the DF200 Specific design flaws can be categorised under three headings apparent lack of imagination
37. will wish to access cp after accessing c It is possible for a sophisticated gadget to try to predict cp given the past choices of the current user but most systems make static assumptions about the typical user s behaviour An example from the DF200 is as follows if the user sets the network type public or private they can then very easily specify whether it uses pulse or tone dialling In fact activating the network command itself accesses the pulse tone command Here the designer s prediction has a dramatic improving effect on the efficiency of a certain sequence of actions On the other hand if the user changes the answerphone mode from fax answer to answer only then the manual recommends recording a new announcement so it no longer says sending a fax is possible Yet recording a new announcement is not easier to reach after the change in fact needing to record a new announcement is presumed to follow listening to the current announcement and it has to be reached from standby Another example arises after a power failure The fax prints a Check date and time warning when power returns yet setting the time is still as difficult lt 6 gt to do as ever it has to be done from standby as usual It could have been made easier the printout could have said Check date and time Press STARTICOPY to do this easily Here the prediction is based on the world s actions on the fax prediction can be based on more than
38. ys to choose the next letter of the function name that is using a trie we achieve an average function access of about 13 However most choices in a trie are unique and taking advantage of this reduces the numbers of button presses to 9 29 max 17 So far as the user is concerned the fax s commands are accessed by using a simple dictionary lookup A trie necessarily achieves an organisation at least as sensible as the DF200 s choice of command names The full trie requires 27 symbols using the DF200 manual s exact names for each command but by mapping the commands onto the letters that each of the DF200 s keys already uses and ignoring punctuation only 8 keys are required With this transformation the average and worst cases are unchanged So we do not even need to change the keyboard desired command appears in the LCD The LCD would appear to work like an old style command completion user interface If the user wanted a direct search e g not knowing how to spell a command this can be achieved by pressing OPTIONS which would take the user onto the alphabetically next command in sequence and wrap around at the end so no command is ever missed Variations Search by prediction To some extent but not consistently the DF200 supports search by prediction That is if some command c has been accessed some command cy is immediately easier to get to than by returning to standby The fax has predicted that the user
Download Pdf Manuals
Related Search