Content uploaded by Simon Nestler
Author content
All content in this area was uploaded by Simon Nestler on Sep 06, 2021
Content may be subject to copyright.
CHALLENGES OF DOWNSCALING
AND UPSCALING IN HUMAN-
CENTERED DESIGN
SIMON NESTLER
FAKULTÄT INFORMATIK
TECHNISCHE HOCHSCHULE
INGOLSTADT
ESPLANADE 10
D-85049 INGOLSTADT
SIMON.NESTLER@THI.DE
SVEN QUADFLIEG
HAMM-LIPPSTADT
UNIVERSITY OF APPLIED
SCIENCES
DR.-ARNOLD-HUECK-
STRASSE 3
D-59557 LIPPSTADT
KLAUS NEUBURG
OSTFALIA HOCHSCHULE
FÜR ANGEWANDTE
WISSENSCHAFTEN
KARL-SCHARFENBERG-
STRASSE 55-57
D-38229 SALZGITTER
ABSTRACT
Design has power – to improve lives, to empower
people and to break down barriers. Successful
design requires (besides many other tasks) a
comprehensive analysis and deep understanding of
the target audience. However, current design
approaches, for instance established in the field of
Human Centered Design, lead to multiple biases:
Design neglects a multitude of needs when it uses
downscaling to make complex target groups
manageable. Downscaling must therefore
disproportionately consider special needs within
the design process – and upscaling must be able to
compensate these biases again. The approach
presented in this paper delivers three benefits:
Conflicts between general and specific
requirements are resolved, efficiency and equity
are given equal consideration, and synergies
1
BITV is a German regulation on Barrier-free Information
Technology; in German: “Barrierefreie-Informationstechnik-
become possible even at the resource level. By
systematically analyzing and linking the issues of
downscaling and upscaling in the context of design
processes, the paper provides guardrails; these
guardrails guide the design process and support a
better focus to the general and specific needs of the
target group.
INTRODUCTION
Design is a powerful tool: it can improve lives,
empower people, and break down boundaries and
barriers. At the same time design can also discriminate
and exclude by simply not working for everyone or
even manifesting or reinforcing existing sexism, racism
or existing discrimination. A key to design that does not
discriminate or exclude anyone is strongly linked to the
idea of usability and accessibility. This concept is linked
to the idea of usability and accessibility: Good usability
and accessibility should guarantee that artifacts and
processes can be used equally well by all users.
In Germany, the topic of UUX (Usability and User
Experience) is currently receiving growing attention.
One of the triggers is the BITV
1
, which defines legal
Verordnung“ (https://www.gesetze-im-
internet.de/bitv_2_0/BJNR184300011.html)
No 9 (2021): NORDES 2021: MATTERS OF SCALE, ISSN 1604-9705. www.nordes.org
standards for the accessibility software in public
authorities (Algermissen et al. 2005). Especially the fact
that accessible software is the result of a complex
process leads to various challenges. When considering
the related ISO standards
2
on Human Centered Design
(HCD)
3
, it becomes clear that usability is the result of a
process. Combining this insight with the implications
from BITV, accessibility is nothing more than usability
for groups with specific needs. Thus, the HCD moves in
a field of tension between specific needs (accessibility)
and general needs (usability). As a consequence, design
teams are faced with the challenge of balancing these
needs on a day-to-day basis.
However, accessibility is only one example of such
specific requirements. From the perspective of HCD,
there are a large number of groups that are not given the
necessary consideration by the existing processes and
whose requirements are thus left out. This is not only
due to a lack of sensitivity to such discrimination, but
also to the fundamental systematics of the HCD process,
which consists of an iterative interplay between
downscaling and upscaling (e.g. Henze et al. 2011,
Henze 2012). This downscaling can also affect
(depending on the context of use), for example, women,
BIPoC, left-handed people, blind people, short-sighted
people, people who wear glasses, tall people, short
people – and many others (Coleman & Lebbon 1999,
Newell & Gregor 2000). We want to emphasize that
discrimination in the context of HCD is not limited to
the “traditional” categories of discrimination, but is
even more multifaceted in individual contexts of use.
Thus, the principle of multiple discrimination described
by the term intersectionality (Crenshaw 1989, McCall
2005) can also applied to HCD (Schlesinger et al. 2017,
Windsong 2018, Rankin & Thomas 2019).
However, the HCD according to ISO 9241-210 is also
only one example of a human-centered design process
characterized by upscaling and downscaling.
Ultimately, other processes, such as design thinking and
the Double Diamond are also characterized by a
constant alternation between convergent (downscaling)
and divergent (upscaling) methods (British Design
Council 2005, Johansson-Sköldberg et al. 2013, Dorst
2015, Carlgren 2016, Park & McKilligan 2018).
Therefore, this paper focuses on the basic systematics
and the underlying problem: How can a (human-
2
Especially ISO 9241-210 (Thomas et al. 2017, DIN EN ISO 2010).
3
It should be briefly pointed out that the concept of human-centered
design is quite critically discussed: On the one hand, "human" is a
term that sounds empathetic and empowering in principle, but is
completely undefined for the specific context and requires a more
precise process to ensure that all needs are covered as much as
centered) design process focus while keeping users with
specific needs in mind?
For this purpose, we will first deal with the rather static
downscaling and upscaling during the life cycle of
artifacts before we will then take a closer look at the
interplay in the context of design processes.
“DESIGN FOR ALL”
Analyzing the target group is, according to our
observation, still the most neglected aspect in industrial
practice. When interdisciplinary teams are asked to
create a particular artifact (e.g., an online store, an app
or a gesture-based interface) as part of hands-on
activities, they too often start with directly designing the
artifact. Experienced teams differ from inexperienced
teams not only in the solutions they design, but more
importantly in the questions they (don’t) ask.
Figure 1: User research changes the target group: (a) Target
group without user research vs. (b) with user research
The questions asked by the more experienced and
reflective teams first gather information about the
context of use (e.g.: Who are the users? What are their
tasks? What are their tools? How does their
environment look like? Or: In short, what exactly is the
problem? Why do you need an online store at all?). One
could counter that there are indeed processes in design
that try to terminate exactly these questions – for
example, the often-cited design thinking (Thoring &
Müller 2011, Plattner 2013, McKilligan et al. 2017,
Brown & Katz 2019). But even such processes, which
even pose the question of the problem at an early stage
possible - this is, after all, discussed in this paper. Likewise, in theory,
yes, the term provokes the question of design that includes non-human
interests. Above all, however, the practice of HCD is repeatedly the
focus of criticism, since in reality human-centered means rather profit-
centered and people are regarded more as consuming beings.
No 9 (2021): NORDES 2021: MATTERS OF SCALE, ISSN 1604-9705. www.nordes.org
and involve the user, are problematic if the scaling
process described later is not carried out correctly.
Teams who skip these considerations, design with the
implicit hypothesis in mind that the online store is “for
everyone”. In Figure 1, these two user groups are
compared: When comparing the left side (without user
research) with the right side (with user research) of
Figure 1, one thing stands out: Some groups are missing
on the right side. In this (and the following) figures,
each geometric shape stands for individual users with
certain characteristics. This insight helps to initiate a
critical reflection of the hypothesis “the design is for
everyone” by addressing, for example, the following
questions:
• Is the store also for minors?
• Is the store also for welfare recipients?
• Is the store also for pensioners?
• Is the store also for illiterate people?
• Is the store also for people without a car?
• Is the store also for people in the countryside?
DOWNSCALING (ON TARGET)
These questions make a valuable contribution to the
next step: The right focus. Only with the right focus
design can solve problems and reach the target group –
as shown in Figure 2. Focusing increases the total
amount of users: When designing for an unspecified
target group (left side), then only a low percentage of
people will be addressed. If the target group is clearly
defined (right side), the pie gets smaller, but the piece
gets bigger (even in absolute terms); we refer to this
strategy as downscaling on target.
In consequence, a differentiated examination of the target
group leads to a comprehensive understanding of the context
of use. Different methods of user research, for example
interviews, focus groups and surveys help to develop a better
understanding of the target group (Väänänen-Vainio-Mattila et
al. 2008, Rohrer 2014, Robinson et al. 2018). The results are
typically made usable within the design team through personas
(Chang et al. 2008, Miaskiewicz & Kozar 2011, Schulz &
Fuglerud 2012).
In the context of this paper, which is dedicated to the
downscaling and upscaling that takes place in design
processes, this approach has some weaknesses: The
more realistic and closer to reality these personas are
designed, the higher the risk that this representation of
reality will be mistaken for reality in the further course
(Junior & Filgueiras 2005). This leads to various
challenges: As personas depict prototypical users,
personas (despite their foundation in research) focus
primarily on the greatest common denominator. This
strong exaggeration of the commonalities leads to the
danger that stereotypes develop, a pigeonhole thinking
evolves, uniformity dominates in the further design
process and individual facets are lost. Numerous
existing artifacts from other contexts demonstrate this
problem: Car interiors adapted to an average man’s
body (while increasing the risk of injury and death for
those who deviate greatly from that body-especially for
women, whose specific characteristics are not taken into
account) can serve as one of countless examples
(Criado-Perez 2020).
Figure 2: Focusing increases the total amount of users: (a)
Design for All vs. (b) Downscaling on target
The tension between facets that are relevant for
abstraction and those that are irrelevant is not resolved
by the persona approach; therefore, we will present a
possible resolution of this tension in the chapter on
differentiated downscaling.
UPSCALING (OFF AND ON TARGET)
Downscaling is – as just described – the decisive
process in order to be able to carry out the process of
design in a focused manner. The opposite principle,
upscaling, on the other hand, is not relevant until much
later: For example, the four phases of ISO 9241-210 are
iterated multiple times; this means that several iterations
are necessary until a solution is available that can be
used effectively, efficiently and satisfactorily in the
specified context of use (i.e., in particular by a concrete
user group) in practice. Only at this point – when the
problem is sufficiently solved for a specific group –
strategies for upscaling are relevant.
In practice, upscaling often starts earlier (for example,
due to economic constraints); this is fundamentally
extremely detrimental to the process going forward:
design teams lose the necessary focus and, in the worst
case, find themselves again faced with the challenge of
having to design “for everyone.” Furthermore,
broadening the target group on the basis of a usable
solution succeeds more easily - design processes can
then concentrate on the additional requirements to be
considered, and thus remain focused despite the
upscaling. This upscaling can basically be done in two
ways – as shown in Figure 3.
No 9 (2021): NORDES 2021: MATTERS OF SCALE, ISSN 1604-9705. www.nordes.org
Figure 3: Different strategies for upscaling: (a) Off-target and
(b) on-target
On the left side better coverage of the target group
(upscaling on target) makes use of established methods.
The second approach on the right side (upscaling off
target) is broadening the target group: The context of
use expands to include users with new needs not
previously considered. Design provides the adequate
methods to methodically support this broadening and to
further develop the artifact accordingly. In the context
of design, (bottom-up) approaches to broadening can
emerge, but this broadening can also come from
outside; broadening can also come (top-down) from
changing the business model.
These downscaling and upscaling processes are
characterized by their rather static nature: The processes
comparatively rarely take place in the lifecycle of an
artifact. Permanent downscaling is typically required
once at the beginning of the design process, and
upscaling also occurs with rather low frequency
(sometimes not at all). However, the design process
itself also makes intensive use of the mechanisms of
upscaling and downscaling. Thus, in the next chapters
we will focus on the mechanisms within the design
process itself.
DOWNSCALING IN THE CONTEXT OF HCD
When downscaling in design aims to represent reality as
accurately as possible, distortions arise. It is well known
from research that ultimately there are always two
categories of hurdles: On the one hand, there are errors
that affect a large proportion of users - and on the other
hand, there are errors that affect only a comparatively
small group of users (or even only one user) (Kujala et
al. 2001, Lindgaard et al. 2006, Wang et al. 2020).
Against the background of limited resources, design will
thus always inevitably focus on the first category.
Measures of optimization thus always refer to the
achievement of the greatest possible effects for the
largest possible group of users - the larger the group of
people affected and the more serious the hurdle, the
greater the attention paid to this hurdle in the course of
the design process. This approach ensures both the best
possible use of resources and the best possible overall
effect. Ultimately, this is nothing more than an
application of the pareto principle (Dunford et al. 2014),
(Kiremire, 2011): With appropriate prioritization and
focus, 20% of the budget required to eliminate all
hurdles can already eliminate 80% of the hurdles –
always in relation to the totality of all users.
So far, we have used the term Design for All in a
shortened form as a synonym for “design for an
insufficiently analyzed target group”. This
contextualization may initially give the term a negative
connotation, but this is relativized by the clarifications
we have just made: Methodically correct Design for All
goes into the breadth, and does not follow the Pareto
principle. Design for All is not design for an unspecified
general public, Design for All is the consideration of all
requirements within the clearly specified target group.
The inadequacy of the Pareto principle in the course of
Design for All is thus not in contradiction to intimate
downscaling, it is rather a strong argument for its
necessity: The construct “all requirements of the target
group” is only specific, measurable, accepted, realistic
and scheduled if the target group has been sufficiently
specified in advance in the course of downscaling.
The idea that design should work for everyone is, of
course, ingrained in design discourse. Universal design
in particular (Mace 1985; Center for Universal Design
1997) has attempted to formulate rules that attempt to
create the basis of a design that works for all possible
users. These focuses, among other things, on physical
and cognitive limitations – but do not elaborate further,
and especially not in detail, on how consideration of
such factors should be reflected in the design process.
The idea that marginalized persons should also be taken
into account is thus formulated – whereas the concrete
implementation recommendation is missing. For a
design that tries to exclude any form of group-focused
enmity and – see intersectionality – the combination of
several characteristics, the approach is also not suitable
because the focus is on inclusion and not on avoiding
discrimination.
UNDIFFERENTIATED DOWNSCALING
Design neglects a multitude of requirements when it
makes complex target groups manageable with the help
of downscaling. When Nielsen in 2000 postulated that
“Five users are enough”, his statement was critically
and intensively discussed within the HCD community
(see Faulkner 2003, Woolrych & Cockton 2001, Spool
& Schroeder 2001). Even if seven, ten, fifteen or twenty
users have to be tested in practice, downscaling is still
crucial in order to make design processes manageable:
The prototypical users (personas) serve as a template for
the selection of suitable subjects. Just like the design
process itself, the selection that takes place in the course
of downscaling also focuses on “the 80 percent”. Only
those hurdles that occur in at least two of the usability
No 9 (2021): NORDES 2021: MATTERS OF SCALE, ISSN 1604-9705. www.nordes.org
tests have a realistic chance of being eliminated in the
further course.
We refer to this process of downscaling in the further
course of this publication as undifferentiated
downscaling, since the consideration of specific needs is
neither intended nor desired. Figure 4 illustrates how
specific needs play no role in the selection of subjects.
To this end, we have extended our visual representation
and additionally use unfilled geometric shapes. These
shapes represent users within the respective user group
with individual barriers. During undifferentiated
downscaling, these individual barriers are ignored.
Figure 4: Individual barriers are ignored during
undifferentiated downscaling
This observation has a central and obvious limitation:
People in real life are not as uniform as the persona
identified in the design process. The persona is an
abstraction; therefore, against the background of the
previous findings, the question inevitably arises whether
the right aspects are now part of this abstraction – and
whether the aspects not taken into account have been
neglected for the right reasons. If one has doubts about
the basic validity of the persona approach, there is an
alternative interpretation: When the shapes are the
known aspects of the target group and the fillings are
the unknown aspects, the bottom line remains the same -
the unknown aspects are not considered during
downscaling.
Depending on the specific requirements considered in
each case, the percentages vary greatly. For some of the
groups considered at the beginning (e.g., women), the
percentage is significantly higher than shown, while for
other groups (e.g., blind people) it is lower. Figure 4
therefore initially only makes clear that specific needs
occur with different frequency depending on the context
of use considered and are initially left out of the
downscaling required for the design process.
This practiced process of downscaling is the enemy of
any specific requirements. But in practice, this effect
can even be exacerbated: If in the further course after
downscaling (of the sample) a generalization of the
findings (upscaling) takes place, this process acts like a
target group filter, as shown in Figure 5: When using
the dark gray circular area (the result from the
downscaling, see Figure 4) for the upscaling attempts
(all light gray circular areas), individual barriers do not
reappear during the process of upscaling. Under this
focal lens, only the requirements lying in the overlap
point remain – and the originally finely differentiated
target group becomes narrower. In theory, user research
does not change the target group - but in practice,
strategic and operational decisions are often based on
these findings. Implicitly, at many points in the process,
the “stamped reality” from Figure 5 might be used
instead of the “real reality” from Figure 4.
Figure 5: Subsequent downscaling and upscaling distorts the
reality
In particular, qualitative findings from the research
shape the further orientation of designers and
developers in a decisive way. The effect shown in
Figure 5 is not a defect in the system, but ultimately a
desired effect of design: the focus should shift from the
world of thought of the designers to the world of
thought of the users. Therefore, regarding the
undifferentiated downscaling we must conclude: It is
methodically correct and leads to a representative user
study. At the same time, however, it is also the reason
why we experience a multitude of systems in practice
that do not work for users with individual barriers.
DIFFERENTIATED DOWNSCALING
As explained in detail in the previous chapter,
downscaling is not a priori non-discriminatory. On the
contrary: Downscaling currently practiced in the context
of design processes (which is also mandatory in the
course of manageability) is always discriminatory. By
focusing on the highest common denominator, design
processes ignore the specific needs – especially of
smaller marginalized groups. For the sake of clarity and
precision: In our further considerations, a small
marginal group is a group with individual requirements
that affect less than five percent of the total. Thus, on
the one hand, these requirements are well below the
threshold of 20 percent (of the Pareto principle) and, on
the other hand, it is unlikely that subjects from this
No 9 (2021): NORDES 2021: MATTERS OF SCALE, ISSN 1604-9705. www.nordes.org
group are already accidentally recruited during
undifferentiated downscaling.
Thus, obvious candidates for such groups are motor and
cognitive impairments of any kind. However, some of
the aspects mentioned at the beginning (e.g. left-handed
people, BIPoC, women) would generally not be
covered. However, since the relevant basis is the context
of use, the specific requirements of women, for
example, can also be covered by the five-percent hurdle;
think, for example, of specialist applications for
occupational groups still dominated by men. Men,
however, can as well be affected by the five-percent
rule, for example when specialized applications for
educators are designed
4
. From these findings, a better
downscaling strategy can be derived; we refer to this as
differentiated downscaling because of the great
importance of a differentiated approach.
Figure 6: Downscaling with a stronger focus on marginalized
groups
As shown in Figure 6, marginalized groups with
specific needs must be overrepresented in the design
process in order to be adequately addressed: The filled
square (general needs) is replaced by an unfilled one
(specific needs). The same is done for the specific needs
represented by the unfilled circle (note, however, that
only one circle is replaced). All of these specific needs
that are overrepresented in differentiated downscaling,
would fall by the wayside in undifferentiated
downscaling. Or to put it another way: If specific needs
are quantitatively so serious that they are already taken
into account in undifferentiated downscaling, they do
not have to be overrepresented in differentiated
downscaling. All other specific needs, however, only
4
In Germany, the proportion of male kindergarten teachers exceeded
the five percent hurdle (5.2 percent) for the first time in 2015 (2014:
gain access to the sample through their intentional
overrepresentation.
Two different cases can be observed among these
specific needs: On the one hand, quantitative minorities
(e.g., cognitive and physical barriers) are permanent
beneficiaries of differential downscaling. On the other
hand, temporary minorities (e.g., BIPoC and women)
also benefit from differential downscaling. This also
results in two slightly different effects: In the first case,
differentiated downscaling serves a better representation
of reality with all its facets – since successful design is
decided precisely by these facets. In the second case,
differentiated downscaling serves to reduce
discrimination and inequalities that lead to current
underrepresentation. In this way, differentiated
downscaling also makes a substantial contribution to
breaking through the chicken-egg problem: As long as
groups are underrepresented, they are given special
consideration by the five-percent rule.
The five-percent rule makes itself partially superfluous
through its consistent application. Therefore, the
differential analysis of the downscaling process is not a
one-time activity; rather, the design process must
regularly validate the validity of the five-percent rule –
and, if necessary, include new groups. In practice, this
has very concrete implications, for example: A
government agency has 1,000 employees, one of whom
is blind. If a new application for booking business trips
is to be introduced, then he must be included in the
design process. Or if this authority introduces a tool for
internal project management for its 50 managers (48
male, 2 female), then the differentiated downscaling
ensures that at least one female manager is included in
the design process. Or if there are three BIPoC working
in a logistics center with 500 employees, then at least
one should be included in the design process here as
well.
UPSCALING IN THE CONTEXT OF HCD
As we have just explained, there are a number of pitfalls
in downscaling. But upscaling can equally lead to a
distortion of reality. The right strategy is also crucial
here to avoid falling off the horse on the other side:
Differentiated downscaling should not lead to a
situation where consensual requirements (“the 80
percent”) are no longer appropriately prioritized and
focused.
4.8 percent) (see https://de.statista.com/infografik/14678/maennliche-
paedagogische-fachkraefte-in-kitas/).
No 9 (2021): NORDES 2021: MATTERS OF SCALE, ISSN 1604-9705. www.nordes.org
This fear is not taken out of the air, but several aspects
lead to the fact that the basic problem with upscaling in
design processes is less serious: First, even after the
consideration of the marginalized groups, subjects
without specific requirements still remain in the target
group – and are the focus of the design process. Their
consolidated requirements are therefore still taken into
account (on the basis of the Pareto principle). Second,
the marginalized groups do not only contribute specific
requirements, so these subjects also play a crucial role
during further consolidation. And third, general and
specific requirements are often mutually dependent. For
example, BITV requires accessibility and usability.
UNBALANCED UPSCALING
The broad masses thus benefit from the requirements of
special groups. From the practice of accessibility
assessment, for example: Accessibility analysis and
improvement makes interfaces better for everyone. This
philosophy is for instance proposed by the design and
consulting company IDEO; they recommend to pay
special attention to the extremes in design processes
5
. In
concrete terms, this means that the sample created by
differentiated downscaling is unbalanced – it
disproportionately represents the marginalized groups.
If this distribution is adopted in the course of upscaling,
then the marginalized groups are also disproportionately
represented in the target group focused on in the further
process. We refer to this effect as unbalanced upscaling
and illustrate it in Figure 7 (left): The resulting specific
needs are overrepresented during this kind of upscaling
process.
Figure 7: Different strategies for upscaling: (a) Unbalanced
upscaling vs. (b) Balanced upscaling vs. (c) Balanced
upscaling with inverted downscaling
The danger of unbalanced upscaling is that different
intentions are mixed together – and contradictions can
arise in the further design process. The objective of
downscaling and classical design processes is the most
effective and efficient use of resources (achieving as
much as possible for as many as possible). The
objective of differentiated downscaling was the best
5
see: https://designthinking.ideo.com/resources/extremes-and-
mainstreams-design-toolkit-by-ideo-org
possible addressing of individual barriers. If these two
aspects are now placed next to each other in the context
of unbalanced upscaling, they inevitably compete with
each other. This means that both directions of objectives
thus also become the subject of prioritization and
focusing. It sounds paradoxical at first: Although the
marginalized groups are given additional weight by
unbalanced upscaling, this weight is usually not
sufficient to achieve a sufficiently high priority for these
concerns.
BALANCED UPSCALING
The greatest weakness of unbalanced upscaling thus
does not lie in the overrepresentation of the concerns of
the special target groups – on the contrary, the weakness
arises from the systematics of the process of
prioritization and focusing. This systematics is
necessary due to the limitation of resources. As our
proposed strategy should be capable of maintaining the
two different intentions of the design process, these
specific needs may not get lost during upscaling. This
requires first of all a removal of overrepresentations in
the course of upscaling. With the maxim “as much as
possible for as many as possible” the specific needs
have to be put in relation to the population; this leads us
to balanced upscaling, as shown in Figure 7: By
replacing individual needs by general needs (Figure 7,
middle) and additionally taking into account the insights
from differentiated downscaling (Figure 7, right),
upscaling is capable of inverting the downscaling
process.
This strategy results in two advantages: When
prioritizing hurdles, existing systematics can be
retained. Based on the recognition that the broad masses
benefit from the needs of special groups, general
improvements can be achieved with the resources
originally available for design processes; in practice
designers regularly observe the following effect: What
is unusable for special needs groups is often usable by
the general public only with great difficulty (Astbrink et
al. 2003, Keates & Clarkson 2003, Borys et al. 2013).
When design processes increase effectiveness for
special groups, they increase efficiency and satisfaction
for the general public at the same time. Secondly,
balanced upscaling makes use of the separation of
concerns: Balanced upscaling explicitly rejects the
hypothesis of resource neutrality. If available resources
are to be distributed between two diametrically opposed
intentions, then two good intentions enter into
competition and conflicts are pre-programmed. Instead
of moderating these conflicts in the context of design
No 9 (2021): NORDES 2021: MATTERS OF SCALE, ISSN 1604-9705. www.nordes.org
processes on the concrete case such conflicts have to be
resolved in advance. Additional resources must
therefore be allocated to the additional activities up
front. While the interleaving in downscaling leads to the
resource-saving integration of the additional activities,
these different interests have to be balanced in
upscaling.
In consequence, balanced upscaling follows the
previous process of prioritizing and focusing the
requirements on the basis of the Pareto principle with
the resources available for design. Since it is based on
the differentiated downscaling, specific requirements
that are “majority-driven” are also taken into account.
At the same time, an additional budget is provided to
address the specific needs in the process. During this
process synergies are considered, but all specific needs
are equally significant. We would like to emphasize that
this requirement leads to a fundamentally different
systematic for prioritizing needs: Inclusive design
decisions are based on the lowest common multiple
rather than the highest common denominator. Thus, the
second budget does not follow the didactic of efficiency
and effectiveness, but rather the principle of equal
opportunity.
These marginal adaptations of the previous processes
are fundamental; and marginal adaptations lead to
central effects. This is clearly illustrated by the (already
discussed) example of accessibility vs. usability: First,
part of the budget for usability no longer has to be
diverted for better accessibility; instead, the budgets are
planned separately and backed up with concrete goals
and metrics. Second, usability measures no longer have
to be covered by the accessibility budget; instead, the
budget can also be used for specific requirements that
are eligible for majority support. Third, individual
barriers do not compete with general requirements; the
principle of equal opportunity is separated from
increasing effectiveness and efficiency.
DISCUSSION
In the end, what can this publication achieve in this field
of tension? Many of the topics have already been
analyzed and discussed in the field of UUX from
different angles and under consideration of different
facets. However, the systematics of downscaling and
upscaling presented and explained in this paper and its
application to design processes has not been done in this
form before. Thus, the paper makes a valuable
contribution to resolving the tension between general
needs and individual barriers. This can be achieved on
the one hand by separating downscaling from upscaling
and on the other hand by using different approaches for
integrating specific needs.
This publication is intended to help practitioners in the
field of UUX, for example, to differentiate between
usability and accessibility on the one hand and to
exploit synergies on the other. At the same time, these
findings and methodologies can be transferred to other
design disciplines and be used for supporting specific
goals and concrete strategies (advancement of women,
accessibility, etc.).
Of course, self-critical reflection also includes the fact
that the mandatory prerequisite postulated in the context
of balanced upscaling (additional budget) is not part of
the solution but part of the problem in many practical
issues. We are aware of this problem, although this
publication at least provides a substantive
argumentation basis for claiming additional budgets.
Nevertheless, the design teams should never be forced
to make difficult trade-offs that cannot be handled with
design, regardless of the concrete framework conditions
– even if resources are strictly limited. If economic
reasons really do make trade-offs unavoidable, then they
must be made at the management level. A competition
between the two goals can only be resolved – even with
limited overall resources – through separate budgets;
even if, in the worst case, this means that something has
to be diverted from the existing budget.
A second hurdle may arise in practice from the presence
of a large number of marginalized groups with specific
needs. In the extreme case, the sample is filled
exclusively with representatives of marginalized groups
– and is not even quantitatively sufficient for all
marginalized groups to be represented. While the
aforementioned intersectionality can sometimes lead to
additional challenges, it is a valuable phenomenon here
that can significantly reduce the effects: For example,
female BIPoC provide a particularly large number of
specific needs in design processes, or large left-handed
people help identify edge cases in a particularly targeted
way. If this strategy also does not lead to a resolution of
the conflicting goals, the overall sample can
alternatively be enlarged in differentiated downscaling
instead of replacing individual subjects. This
enlargement is not the ideal solution, since it increases
the effort for the design process, but it is a compromise
that can be achieved (especially if this concept is not
used in an excessive form) – a compromise that can help
prevent discrimination, sexism, and racism in and
through design solutions.
CONCLUSION
On the one hand, the strategy we propose allows HCD
processes to focus and narrow down the issues (in the
course of differentiated downscaling), while remaining
open to prioritized generalization of findings (in the
course of balanced upscaling). Downscaling itself –
despite its discriminatory effects – is not negative;
downscaling is necessary to maintain focus in the design
process. Downscaling makes complex realities
manageable; personas (properly done) are as important
No 9 (2021): NORDES 2021: MATTERS OF SCALE, ISSN 1604-9705. www.nordes.org
in practice as maps – they simplify a complex,
multidimensional and differentiated world. This
simplification is a necessary condition for orientation
and practical usability of these tools. However, our
persona map ultimately only draws our attention to the
aspects and facets that are particularly relevant.
With the systematic linking of downscaling and
upscaling, this paper helps us to fulfil this purpose – to
maintain the structuring and focus-supporting guard
rails in the further course of design processes. Instead of
an arbitrary section on reality, the differentiated
downscaling directs the view to the special “sights”. At
the same time, reflection on the processes of
downscaling will also help to ensure that this issue
receives greater attention in future discourses on design
methods.
Legal foundations support the process of finding
bottom-up synergies in the area of tension (general
requirements vs. individual requirements). Although
template-like and standardized requirements have a
particularly strong resonance in practice due to their
ease of application, they are not entirely harmless: They
can be mistaken as a top-down approach. Thus, on the
basis of our considerations of downscaling and
upscaling, these regulations should even more clearly
point out that no useful shortcuts or top-down solutions
make a differentiated examination of the context of use
dispensable. In order to effectively avoid playing off
groups that are discriminated against in different ways,
bottom-up strategies such as differentiated downscaling
in combination with balanced upscaling have to be
implemented.
REFERENCES
Algermissen, L., Dermann, G., & Niehaves, B. (2005).
Barrierefreiheil für Webseiten von Bund, Ländern
und Gemeinden. Wirtschaftsinformatik, 47(5),
378–382.
Astbrink, G., & Beekhuyzen, J. (2003). The synergies
between universal design and user-centred design.
Brisbane, Australia: Griffith University School of
Computing and Information Technology.
Borys, M., Plechawska-Wójcik, M., Miłosz, M., Luján-
Mora, S., & Caffiau, S. (2013). Synergy effect in
GUI usability and accessibility education
improvement. 2013 IEEE Global Engineering
Education Conference (EDUCON), 311–317.
British Design Council (2005). The Design Process: The
double diamond design process model, London
Brown, T., & Katz, B. (2019). Change by design: how
design thinking transforms organizations and
inspires innovation (Vol. 20091). HarperBusiness.
Carlgren, L. (2016). Design thinking in innovation, in
practice: the case of Kaiser Permanente. EURAM
Conference Proceedings. European Academy of
Management, June 1-4, Paris,.
Center for Universal Design (1997). The Principles of
Universal Design, Raleigh, NC: North Carolina
State University.
Chang, Y., Lim, Y., & Stolterman, E. (2008). Personas:
from theory to practices. Proceedings of the 5th
Nordic Conference on Human-Computer
Interaction: Building Bridges, 439–442.
Coleman, R., & Lebbon, C. (1999). Inclusive design.
Helen Hamlyn Research Centre, Royal College of
Art.
Crenshaw, K. (1989). Demarginalizing the Intersection
of Race and Sex: A Black Feminist Critique of
Antidiscrimination Doctrine, Feminist Theory and
Antiracist Politics,. University of Chicago Legal
Forum, 1, 139–167.
Criado-Perez, C. (2020). Unsichtbare Frauen: Wie eine
von Daten beherrschte Welt die Hälfte der
Bevölkerung ignoriert, München: btb Verlag
DIN EN ISO (2010). 9241-210: Ergonomie der
Mensch-System-Interaktion-Teil 210: Prozess zur
Gestaltung gebrauchstauglicher interaktiver
Systeme
Dorst, K. (2015). Frame innovation: Create new
thinking by design. MIT press.
Dunford, R., Su, Q., & Tamang, E. (2014). The pareto
principle.
Faulkner, L. (2003). Beyond the five-user assumption:
Benefits of increased sample sizes in usability
testing. Behavior Research Methods, Instruments,
& Computers, 35(3), 379–383.
Henze, N. (2012). Hit it! an apparatus for upscaling
mobile HCI studies. In CHI’12 Extended Abstracts
on Human Factors in Computing Systems (pp.
1333–1338).
Henze, N., Pielot, M., Poppinga, B., Schinke, T., &
Boll, S. (2011). My app is an experiment:
Experience from user studies in mobile app stores.
International Journal of Mobile Human Computer
Interaction (IJMHCI), 3(4), 71–91.
Johansson-Sköldberg, U., Woodilla, J., & Çetinkaya, M.
(2013). Design thinking: past, present and possible
futures. Creativity and Innovation Management,
22(2), 121–146.
Junior, P. T. A., & Filgueiras, L. V. L. (2005). User
modeling with personas. Proceedings of the 2005
Latin American Conference on Human-Computer
Interaction, 277–282.
Keates, S., & Clarkson, P. J. (2003). Countering design
exclusion: bridging the gap between usability and
accessibility. Universal Access in the Information
Society, 2(3), 215–225.
No 9 (2021): NORDES 2021: MATTERS OF SCALE, ISSN 1604-9705. www.nordes.org
Kiremire, A. R. (2011). The application of the pareto
principle in software engineering. Consulted
January, 13, 2016.
Kujala, S., Kauppinen, M., & Rekola, S. (2001).
Bridging the gap between user needs and user
requirements. Advances in Human-Computer
Interaction I (Proceedings of the Panhellenic
Conference with International Participation in
Human-Computer Interaction PC-HCI 2001),
Typorama Publications, 45–50.
Lindgaard, G., Dillon, R., Trbovich, P., White, R.,
Fernandes, G., Lundahl, S., & Pinnamaneni, A.
(2006). User Needs Analysis and requirements
engineering: Theory and practice. Interacting with
Computers, 18(1), 47–70.
Mace, R. (1985). Universal design, barrier-free
environments for everyone, Los Angeles:
Designers West 33(1), pp.147–152.
McCall, L. (2005). The complexity of intersectionality.
Signs: Journal of Women in Culture and Society,
30(3), 1771–1800.
McKilligan, S., Dhadphale, T., & Ringholz, D. (2017).
Speed Dating with Design Thinking: An empirical
study of managers solving business problems with
design. The In-Ternational Association of Societies
of Design Research Conference, Cincinnati.
Miaskiewicz, T., & Kozar, K. A. (2011). Personas and
user-centered design: How can personas benefit
product design processes? Design Studies, 32(5),
417–430.
Newell, A. F., & Gregor, P. (2000). User sensitive
inclusive design – in search of a new paradigm.
Proceedings on the 2000 Conference on Universal
Usability, 39–44.
Park, H., & McKilligan, S. (2018). A systematic
literature review for human-computer interaction
and design thinking process integration.
International Conference of Design, User
Experience, and Usability, 725–740.
Plattner, H. (2013). An introduction to design thinking.
Iinstitute of Design at Stanford, 1–15.
Rankin, Y. A., & Thomas, J. O. (2019). Straighten up
and fly right: rethinking intersectionality in HCI
research. Interactions, 26(6), 64–68.
Robinson, J., Lanius, C., & Weber, R. (2018). The past,
present, and future of UX empirical research.
Communication Design Quarterly Review, 5(3),
10–23.
Rohrer, C. (2014). When to use which user-experience
research methods. Nielsen Norman Group.
Schlesinger, A., Edwards, W. K., & Grinter, R. E.
(2017). Intersectional HCI: Engaging identity
through gender, race, and class. Proceedings of the
2017 CHI Conference on Human Factors in
Computing Systems, pp.5412–5427.
Schulz, T., & Fuglerud, K. S. (2012). Creating personas
with disabilities. International Conference on
Computers for Handicapped Persons, pp.145–152.
Spool, J., & Schroeder, W. (2001). Testing web sites:
Five users is nowhere near enough. CHI’01
Extended Abstracts on Human Factors in
Computing Systems, pp.285–286.
Thomas, V., Remy, C., & Bates, O. (2017). The limits
of HCD: Reimagining the anthropocentricity of
ISO 9241-210. Proceedings of the 2017 Workshop
on Computing Within Limits, pp.85–92.
Thoring, K. & Müller, R. M. (2011). Understanding
design thinking: A process model based on method
engineering. DS 69: Proceedings of E&PDE 2011,
the 13th International Conference on Engineering
and Product Design Education, pp.493–498.
Väänänen-Vainio-Mattila, K., Roto, V., & Hassenzahl,
M. (2008). Towards practical user experience
evaluation methods. Meaningful Measures: Valid
Useful User Experience Measurement (VUUM),
19–22.
Wang, Y., Luo, L., & Liu, H. (2020). Bridging the
Semantic Gap Between Customer Needs and
Design Specifications Using User-Generated
Content. IEEE Transactions on Engineering
Management.
Windsong, E. A. (2018). Incorporating intersectionality
into research design: An example using qualitative
interviews. International Journal of Social
Research Methodology, 21(2), 135–147.
Woolrych, A., & Cockton, G. (2001). Why and when
five test users aren’t enough. Proceedings of IHM-
HCI 2001 Conference, 2, 105–108.