The BT DF200 fax - Computer Science
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
questionairre…
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 domestic 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. 1
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 sideby-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 & 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 depend on an exact
1
BT is a name of British Telecommunications plc.
w 1 s
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 function,
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 merit
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 f 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 c f is reminds us that the
w 2 s
cost of an operation to the user is not known unless the subjective factors of c f
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, b f
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 buttonpressing 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
Tf that maps the button press count to a time for the user. Note that Tf
depends on f —it may be that the user is more anxious about the use of
certain functions than others, which would affect Tf differentially.
( )
The idea is, of course, that T b f × c f 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 )c
∑c
f
f
f
f
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 bf × cf 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 Tf and c f to be constant. We then obtain the simple average:
( )
∑b
f
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 b f . In our
examples below, we do not change the number of functions nor what they do;
we only modify b f , 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.
w 3 s
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
Actual DF200
Linear search
Direct function
Binary tree
Basic trie
Hybrid trie
Average time
5.86
24.5
1.9
5.69
9.29
3.18
Maximum time
13
49
2
6
17
5
Complete search
117
49
(1.9)
53
49
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
instructions 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].
There are 49 functions in the DF200, and they can be searched linearly
using two existing keys: for example, [OPTIONS] starts the search, and
subsequent presses of it progress through the list of functions. Pressing
[START;COPY] activates the chosen function. (We could use the volume
decrease button because it looks more like ‘scrolling’ than 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
w 4 s
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, and realising
that as the digit keys currently do nothing when the handset is down they
could be used for accessing commands, without requiring the [OPTIONS]
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 naïve
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 [START;COPY] to activate it—and, next, pressing [3] 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 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 [START;COPY] to access each
function. Two keys are sufficient to perform a binary tree search, say
with[OPTIONS] going left and [START;COPY] 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
w 5 s
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 keys 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.
So far as the user is concerned, they now type the command (by pressing
the digit keys with the right letters) and pressing [START;COPY] when the
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 ca has been accessed, some command cb is
immediately easier to get to than by returning to standby. The fax has
‘predicted’ that the user will wish to access cb after accessing ca . It is possible
for a sophisticated gadget to try to predict cb 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
w 6 s
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
[START;COPY] to do this easily.” Here the prediction is based on the world’s
actions on the fax; prediction can be based on more than 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 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 designer 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
w 7 s
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 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, average 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.
w 8 s
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 machine3 because the actual
machine was not documented clearly enough to understand its structure, nor
was the actual device 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 & 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 & 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
monitoring and volume control, 3 for the main functionality user interface, 1
for fax detail control, and 3 apparently for phone control (one, [REDIAL]
works; one only works with a local switchboard but its functionality is
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.
There are 6 LED indicators, inset in buttons [ANSWER], [PLAY], [FINE],
[STOP], [START;COPY] and [MONITOR]. Finally there is an LCD display,
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.
w 9 s
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 [START;COPY] to verify a
function choice, or continue to another box. (Where a box
contains functions which require your choice—choices are in
brackets—you will need to press [START;COPY] to scroll and
not [OPTIONS].)
An example (changing the dialling type from PULSE to
TONE) has been highlighted for you to follow:
Press [OPTIONS], then press [1]. Press [START;COPY]. Press
[ O P T I O N S ] three times and then [ S T A R T ; C O P Y ]. Press
[START;COPY again and then [OPTIONS] to change to tone
dialling. Press [START;COPY] 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, additional 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 m a y 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:
“Now 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.
w 10 s
“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; 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, 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.
w 11 s
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.”
4
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!
5
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
important diagnostic information so complex they need writing down?
6
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).
7
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.
8
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] [*]q [#] (it then shows CONFIGURATION 1),
then press [1] 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).
9
When the DF200 receives a fax and runs out of paper, it does not
terminate the reception or report an error to the sending 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.
10 When paper is inserted to prepare for transmitting a fax, the LCD changes
to A4 NORM and 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,
w 12 s
though, because the user should have either lifted the handset or pressed
[MONITOR] first. Moreover, on pressing [MONITOR] and pressing
[REDIAL], the fax will not redial the number just attempted. One has to
start over. I make this mistake, being led down a garden path, frequently.
11 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.
12 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, 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.
13 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.)
14 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 [REDIAL] 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 number,
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.
15 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
w 13 s
messages.” (This seems to imply that the access code is entered before the
[#].)
16 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).
17 The DF200 has a paper ‘tray’ so that several sheets of an out-going fax can
be held. Unfortunately, if there is any paper in the tray, no in-coming fax
will be received.
18 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).
19 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.
20 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.)
21 After pressing the [ M O N I T O R ] key, the user can hear through a
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.
22 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.
23 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 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.
w 14 s
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
[STOP] on the DF200.
The following table compares all remote commands with their equivalent
DF200 direct commands, where there are equivalents.
Remote
key
Equivalent operation
on DF200
Meaning
[1]
Listen to a message again.
Press ww
[2]
Delete a message.
Listening to message
deletes it unless [SAVE] is
pressed.
[3]
Go on to next message
Press ss
[4]
Listen to previous (sic) message
again.
Press ww twice.
[5]
Listen to all messages, new and old.
[PLAY]
[5]pressed again, “defers the
playing”; and pressed again,
“allows to go on playing messages”
[6]
Does nothing.
[7]
Answer on.
(complex process)
[8]
Record a new announcement. (Note
that there is no way to confirm the
recorded announcement.)
(complex process)
[9]
Answer off.
(complex process)
[#]
“Note: Pressing [#] (STOP) allows
to stop any operation.”
Press [STOP]
25 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, [1] now provides punctuation—including a space as previously
provided by [#]—whereas [*] now beeps and does not produce even a
+ sign!
26 The undocumented [OPTIONS] [#] feature causes ERASE MODE to be
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.
w 15 s
27 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.
28 There is a button [MONITOR]; if the handset is down and this button is
pressed its LED lights. The effect is as if the handset is raised, but the
loudspeaker 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.
29 Despite the button [MONITOR], which appears to give the user a choice
(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.
30 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.
31 The DF200 has eight digit keys with three to four letters ([2]: ABC; [3]:
DEF; [4]: GHI; [5]: JKL; [6]: MNO; [7]: PQRS; [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. 4 In
other modes, [#] behaves like a digit (it is a code that is transmitted to
exchanges) and pressing a number automatically causes the position to
advance.
Digit [1] 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).
32 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).
33 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]u, deletes the last symbol and moves left, but
[VOL]q) does nothing other than beep. Except that if there are characters
4
Pressing [2][2][#][#][2][2][2][#][2][2][#][8][8] enters A BAT.
w 16 s
to the right of the entry position (the cursor), then [VOL]u deletes all
characters to the right and does not move left!
34 The button [PAUSE] normally introduces a pause in recorded dialling
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.
35 The DF200 can transmit faxes at four baud rates but can only receive at
three (it can transmit but cannot receive at 7200 baud).
36 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.
37 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.
38 After the answerphone records a message (either from a caller, or from
the user) the LED on the [PLAY] button flashes. When [PLAY] 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.
39 When the answerphone has received a message and recorded it, the LED
in the [PLAY] button flashes. Flashing indicates that there are one or more
recorded messages, which will be played back when the [PLAY] 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.
40 The manual describes what the LCD displays. It is rarely accurate,
(perhaps) partly because the LCD itself is short, and often shows
abbreviated words.
w 17 s
41 (The inconsistency between the DF200 and its manual is mentioned again
for completeness.)
General inadequacies
42 The DF200 and its external power supply overheats; at standby it
consumes 6.25W—it gets rather hot for a user to place it on their lap.
43 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.
44 The paper support is flimsy and doesn’t fold flat, so it and its flimsy
hinges can break easily.
45 The DF200 clicks despite digital recording technology. The handset
microphone distractingly picks up the sound of breathing.
46 We often find the DF200 shows that it has recorded a message, but which
in fact is null.
47 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.
48 The DF200 picks up radio interference.
49 All the button labels are in capitals (apart from fast forward and rewind,
which use abstract symbols; volume increase and decrease are called
[VOL]u and [VOL]q). This may give the fax a ‘technical’ appearance,
but most evidence suggests that lower case lettering is easier to read.
50 The button [RECALL] (a name confusingly like the button [REDIAL]) has
no meaning except when connected to a private switchboard.
51 The DF200 provides three commands that have no functionality at all
([PRINT], [DELETE] 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.
52 There are two buttons, [REDIAL] and [RECALL]. The manual does not
specify [RECALL] beyond saying it activates unspecified functions from a
local switchboard; thus the button [RECALL] does nothing on my DF200,
which is connected to a public exchange.
53 To protect the fax against echoes, press [PAUSE] when setting the
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.
w 18 s
54 The volume can be increased beyond the loudspeaker’s power rating, as
can be verified by increasing so that ‘bleep’ tone distorts.
55 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
to A4 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 repaired 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?
56 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!)
57 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?
58 When the report says “subscriber has stopped”—does that mean us or
them?
59 What does ‘Disconnected 210E000000’ mean—a code that has occurred
during normal use of the DF200?
60 The sequence [OPTIONS] [*] [#] accesses a list of editable configuration
numbers. Neither the command sequence nor the meaning of the
configuration numbers is defined.
w 19 s
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement