Systemic Theories of Change ADMC 2020 - Design Dialogues

16
22st DMI: Academic Design Management Conference Impact The Future By Design Toronto, Canada, 5-6 August, 2020 Copyright © 2020. Copyright in each paper on this conference proceedings is the property of the author(s). Permission is granted to reproduce copies of these works for purposes relevant to the above conference, provided that the author(s), source and copyright notice are included on each copy. For other uses, including extended quotation, please contact the author(s). Design management for wicked problems: Towards systemic theories of change through systemic design Ryan J. A. MURPHY a , Peter JONES b a Memorial University of Newfoundland; b OCAD University Design and design management is increasingly called upon to respond to the world’s complex, dynamic problems. Yet, no standard methodology exists to help designers understand, model, and design solutions for these wicked problems. Program theory uses theory of change models to develop linear pathways of outcomes to show how a change initiative will have its desired effects. However, critics of these models accuse them of being reductively linear. Systems thinkers use causal loop diagrams to create maps of systems that show their behaviour in their full, dynamic complexity. However, these diagrams are sometimes overwhelming and impractical. In this paper, we combine these tools with a novel technique from systemic design called “leverage analysis” to help identify crucial features of a complex problem and help designers develop practical theories of systemic change. Keywords: Wicked problems, systemic design, leverage analysis, modelling.

Transcript of Systemic Theories of Change ADMC 2020 - Design Dialogues

22st DMI: Academic Design Management Conference

Impact The Future By Design

Toronto, Canada, 5-6 August, 2020

Copyright © 2020. Copyright in each paper on this conference proceedings is the property of the author(s). Permission is granted to reproduce copies of these works for purposes relevant to the above conference, provided that the author(s), source and copyright notice are included on each copy. For other uses, including extended quotation, please contact the author(s).

Design management for wicked problems: Towards systemic theories of change through systemic design Ryan J. A. MURPHY a, Peter JONES b a Memorial University of Newfoundland; b OCAD University

Design and design management is increasingly called upon to respond to the world’s complex, dynamic problems. Yet, no standard methodology exists to help designers understand, model, and design solutions for these wicked problems. Program theory uses theory of change models to develop linear pathways of outcomes to show how a change initiative will have its desired effects. However, critics of these models accuse them of being reductively linear. Systems thinkers use causal loop diagrams to create maps of systems that show their behaviour in their full, dynamic complexity. However, these diagrams are sometimes overwhelming and impractical. In this paper, we combine these tools with a novel technique from systemic design called “leverage analysis” to help identify crucial features of a complex problem and help designers develop practical theories of systemic change.

Keywords: Wicked problems, systemic design, leverage analysis, modelling.

RYAN J. A. MURPHY, PETER JONES

2

Introduction It is no secret that modern organizations are tackling some of the most complex challenges in recent history.

Managing the ongoing COVID–19 pandemic is simply the latest and most prominent example of Rittel and Weber’s (1973) wicked problems. Conventional approaches to problem-solving often fail to generate satisfying progress on challenges of this class. Recently, therefore, leaders have turned to design as a discipline that may be able to turn the tide in complex problem-solving.

This paper presents several contributions to the theory and practice of designing and managing change initiatives in complex systems. First, we emphasize the potential for design to help address wicked problems, and we centre the design management discipline’s role in leading that work. We then show how designers can apply a novel technique from systemic design called leverage analysis to the management of complex changemaking. We combine leverage analysis with the concept of theory of change (ToC) from program evaluation practices to present a hybrid technique we call systemic theories of change. Through a demonstration using a modelled real-world system, we illustrate the utility of systemic theories of change in designing and managing complex changemaking projects.

In section 2, we explain the techniques of leverage analysis and ToC, then define a general approach to combining these two techniques. In section 3, we demonstrate the usefulness of systemic theories of change in a demonstration. Finally, in section 4, we discuss the implications of this technique and point to several directions for research and practice to examine and advance it further.

Design Management and Complex Problems In this research, using the transdisciplinary nature of design management, we integrate existing techniques for

managing complex change from the disciplines of systemics and program evaluation. Making significant progress on wicked problems requires advancing these techniques beyond disciplinary silos. To fully comprehend wicked problems requires “students of complexity” (McGowan et al., 2014, p. 37) who appreciate the need to synthesize multiple—sometimes conflicting—perspectives in complex work via multiple methodologies and epistemologies (Thering & Chanse, 2011). As a practice, design (and design management) is uniquely positioned for transdisciplinarity: the union of different disciplines into a novel, collaborative intellectual framework (Jensenius, 2012)1. Design and design research methodologies can be deliberately planned as design action research (Swann, 2002), participatory (Slavin, 2016), dialogic (Jones, 2018), or co-creative (Sanders & Stappers, 2008). Systems approaches often seek requisite variety of perspectives, or to “get the whole system into the room” (Weisbord & Janoff, 2007) to engage the members of the system to talk to one another. Design training teaches learners to adopt transdisciplinary approaches to understanding complexity such as sensemaking (Klein, Moon, & Hoffman, 2006) and emergence (Chen & Crilly, 2016; Van Alstyne & Logan, 2007). Design also relates easily to other disciplines, so much so that other disciplines often see design as an “applied” version of their understandings (Buchanan, 1992). Further, sub-disciplines of design integrate other disciplines in seeking an ever-more comprehensive worldview. Transition design aims to equip designers to facilitate contextual transitions for sustainability in society (Irwin, 2015). Systemic design recruits systems thinking and theory for advanced design to better provide designers with the ability to navigate complex adaptive systems (Ryan, 2014; Jones, 2014; Sevaldson & Jones, 2019). Design management further integrates design with management science, strategy, leadership, innovation, changemaking, and other business-related fields to advance the design of business and the business of design (Clark & Smith, 2010; McBride, 2010; Buchanan, 2015).

Moreover, a variety of authors have elevated design’s role in addressing wicked problems. In the early 1990s, Buchanan (1992) argued that design was a new liberal art that helps to connect and render visible the interplay between different phenomena, people, and concepts—one that is uniquely suited to imagining the impossible. Kolko’s 2012 book, Wicked Problems: Problems Worth Solving, presents the case for designers to work on innovations with impact (Kolko, 2012). In 2014, a group of design leaders calling themselves “The DesignX Collaborative” proposed adapting design tools and redirecting the focus of design itself to address the “complex and serious problems facing the world today” (Friedman et al., 2014). In a follow-up paper to the 2014 missive, Norman and Stappers (2015) enumerate the kinds of challenges the DesignX Collaborative had identified and outline nine properties shared by these challenges. Ito’s (2017) “Resisting Reduction” manifesto calls for designers

1 See also (Thering & Chanse, 2011; Brown, Harris, & Russell, 2010). Cf. Stember's (1991) informative comparison of multi- and interdisciplinarity, and Oxman's (2016) for a treatise on antidisciplinarity—the notion that knowledge no longer is created within truly disciplinary bounds.

Design management for wicked problems

3

to become more aware of how their designs integrate with ecological, economic, and especially technological systems to facilitate flourishing over arbitrary growth. There has even been a growth of studios or “labs” focused on using design for systems change (Boyer, Cook, & Steinberg, 2013; Boyer, Cook, & Steinberg, 2011; Hassan, 2014; Bellefontaine, 2012; Westley, Goebey, & Robinson, 2012).

As these scholars and practitioners show, there is a need to design—and manage—change in complex systems. The objective of this paper is to provide a novel methodology for the management of design in complex change initiatives. To do so, we bring together approaches from program and evaluation, systems thinking, and graph theory. In the next section, we provide a brief background on what we have borrowed from each of these fields.

Towards Systemic Theories of Change

Theories of Change Although management scientists have discussed the concept of a “theory” of change for decades (e.g., as to

how organizational change happens; Quinn & Cameron, 1988), the phrase here refers specifically to an approach to managing a changemaking initiative developed and used in program evaluation. A ToC, in this sense, is a “blueprint” for change programs (Mackinnon, 2006). Theories of change describe the logic of a change initiative. They capture and make explicit the assumptions and expectations built into change initiatives ("Theory of Change: A Practical Tool," 2004). Theories of change represent situational analyses of the focal system: what happens, why, and via what mechanisms (Funnell & Rogers, 2011, p. 151). Users of ToCs often model them as logic models, such as “outcome chains” (Funnell & Rogers, 2011, pp. 176–195) or “so that chains” ("Theory of Change: A Practical Tool," 2004), where the desired goal is at the end of the chain and the outcomes that lead to it precede from it (Funnell & Rogers, 2011, p. 176).2

Sketching a basic ToC is a straightforward process. Program evaluators work with stakeholders to consider the initiative’s desired impact in the focal problem or system, and to identify the strategies the initiative employs to influence that change. They then work backwards from this impact and forwards from the strategies to determine (1) outcomes that would lead to that impact, (2) objectives that, if met, would lead to those outcomes, (3) what activities from the strategies would accomplish those objectives, and (4) what inputs are required to complete those activities ("Theory of Change: A Practical Tool," 2004). The result of this process is conventionally a linear map of outcomes.

The linear nature of ToC models is powerful in the design and management of change initiatives. Because these models are linear, it is easy to see what the desired goals may be, what actions the organization should take, and what assumptions are crucial to the model’s logic. However, this linearity may fail to capture the true complexity of the system it represents (Abercrombie, Boswell, & Thomasoo, 2018, p. 5). By reducing the complexity of a system to a linear chain of actions and outcomes, leaders may fail to recognize hidden complexity that can stand in the way of results (Alford, 2017b). In the next section, we will describe an approach from systems developed help designers understand and take advantage of this complexity.

2 Note: some authors separate how change happens and how an initiative’s actions create change into separate models as a “theory of change” and a “theory of action” (Funnell & Rogers, 2011, p. 34). In this paper, we use the phrase theory of change to refer to both, as is common in practice ("Theory of Change: A Practical Tool," 2004; Mackinnon, 2006).

RYAN J. A. MURPHY, PETER JONES

4

Figure 1. An example Causal Loop Diagram (CLD) illustrating the concept of a “Success to the Successful” pattern common in education. Dashed arrows represent a negative relationship between phenomena.

Causal Loop Diagrams Sometimes called systems maps or influence maps, causal loop diagrams (CLDs) are models of systemic

behaviour (Kim, 1992). A CLD illustrates how different phenomena in a system interact, often leading to feedback loops and other kinds of systems archetypes. In turn, these models help leaders see how these interactions lead to emergent, counterintuitive behaviour (Braun, 2002; Forrester, 1995). Leaders may, therefore, use CLDs to try to deconstruct and understand these often-unpredictable and counterintuitive behaviours.

Developing a basic CLD is also straightforward. Begin with some specific, focal event, such as “increasing poverty.” Then, within the scope of the system you are studying, think of all the phenomena that could directly cause that event (e.g., “job loss” and “increasing cost of living”), and draw an arrow from the causing phenomena to the caused phenomena. Then, think of all the phenomena that are caused by that event (e.g., “worsening health” and “loss of access to paid services”), and draw a line from your focal event to those newly added phenomena. Also, draw arrows between newly added phenomena and any others you’ve previously added (e.g., worsening health → increasing cost of living). Repeat this process of adding new causes and effects and drawing causal linkages with every phenomenon you’ve just added. Continue this repetition until the model seems comprehensive.

However, as you may infer from the explanation above, CLDs are often complex. It can be challenging to know when to stop adding new phenomena—a problem called boundary framing (Jones, 2014). It can be difficult to know where to “start” or “end” in the system because every element you add could be as actionable or as consequential as the last (the problem of ordering; Jones, 2014). While there is undeniable value in representing and decomposing system complexity, these and many other complications can make CLDs impractical.

Yet, you may also infer that CLDs and ToC models have a lot in common. Both describe the causal structure of a system, usually to intervene within the system. Both are illustrative models represented by boxes of phenomena and arrows linking those phenomena together. Could there be some way of combining these approaches so that the merits of one overcome the flaws of the other? In the next section, we will discuss several approaches to this synthesis that already exist.

Existing approaches to systemic theories of change Other scholars and practitioners have previously recognized the synergy between ToC models and systems

models. First, we must acknowledge that scholars and practitioners in program theory have long acknowledged the need to embed complexity in ToC models (Funnell & Rogers, 2011, pp. 48–49). Funnell & Rogers (2011, p. 159) encourage designers to conduct a situation analysis, for instance, to fully understand a problem before developing an initiative (and they suggest that designers update this situation analysis as they learn from working within the system). They also indicate that designers of ToC models can use branching and nesting to accommodate complex outcome structures.

Design management for wicked problems

5

Nonetheless, ToC models can oversimplify the logic of a problem. To address this oversimplification, several authors have suggested combining systems approaches and ToC models. For instance, Alford (2017a; Alford, 2017b; Alford, 2017c) discusses the limitations of conventional ToC models, and how CLD features such as feedback loops and delays can improve the usefulness of these models. He then describes how a designer might develop a ToC from a CLD or a CLD from a ToC. Both of these directions, he says, can result in more valuable models from which to design and manage change strategies.

Similarly, Abercrombie, Boswell, and Thomasoo (2018) provide a guide for using ToC in systems change work. They, too, note the apparent similarities between the tools: “Theory of change was originally developed to model and evaluate complex change initiatives, so it is clearly related to systems change” (Abercrombie et al., 2018, p. 5). They describe five pitfalls of ToC in systems change work. ToCs can cause leaders to neglect context, ignore their responsibility to change, maintain linear cause-effect thinking, protect their plans instead of adapting to change, and frame change as technical sets of inputs and outputs rather than driven by people as relationships (Abercrombie et al., 2018, p. 5). In turn, they provide five rules of thumb inspired by systems thinking that respond to these pitfalls (Abercrombie et al., 2018, p. 6).

Our approach does not conflict with that of these authors. Instead, we provide a novel method to understanding the structure of a system to help designers develop a ToC model from a CLD to create a systemic theory of change (SToC). We introduce this method, leverage analysis, in the next subsection.

Leverage Analysis Leverage analysis is an analytical technique for understanding the features of a complex problem using graph-

based analysis of the structure of the problem (Murphy & Jones, 2020). It emerged at the intersection of systems thinking, design, data science, and graph theory as a way of reconciling the traditional tensions between the quantitative rigour of systems dynamics and the fuzzy systems comprehension capabilities of systems thinking.

To use leverage analysis on a complex problem, designers must begin with a finished CLD. Once a CLD is developed, and stakeholders reach consensus that it represents the problem at hand as accurately as possible, the model is ready for leverage analysis. Leverage analysis involves running any of several kinds of algorithmic evaluations on the structure of the model, adopted from graph theory (and especially from social network analysis; Murphy & Jones, 2020). These algorithms surface features of the model that can be difficult to observe otherwise, such as which phenomena in the system lie between the most other phenomena (betweenness centrality; Freeman, 1977). Table 1 shows the different kinds of leverage analysis and the features they surface in a systems model.

Table 1. Leverage measures and their use in systemic design.

Measure Description Use in Systemic Design Degree Number of connections Changes to high-degree phenomena will

translate more quickly to more phenomena; high-degree phenomena are more sensitive to changes throughout the system

Indegree Number of incoming connections High-indegree phenomena receive change from many other elements; they are indicators of systemic change and representative of the system’s volatility

Outdegree Number of outgoing connections Many other elements change in response to shifts in high-outdegree phenomena

Betweenness Frequency of participation in the shortest path between two other elements

High-betweenness phenomena are gateways or bottlenecks for change; change strategies must consider how to address these gateways

Closeness Average length of the shortest paths between the given vertex and every other vertex in the graph

High-closeness phenomena are resilient or independent, resisting change coming from elsewhere in the system; likewise, the system may resist change coming from these phenomena

Eigenvector Connectedness to other well-connected elements

High-eigenvector phenomena are powerful; these are good candidates for leverage across the rest of the system

RYAN J. A. MURPHY, PETER JONES

6

Reach Number of elements within [x] steps of the given element

High-reach phenomena connect deeply to the rest of the system; other relatively disconnected elements will change in response to shifts in these phenomena

Reach efficiency Reach divided by the degree of a given node High-reach efficiency phenomena quickly and efficiently propagate change throughout the rest of the network

Eccentricity

Distance away of the furthest node High-eccentricity phenomena are far away from others, implying localization of outcome or intervention; eccentricity may be used to structure and target “neighbourhoods” of phenomena that are structurally close together

It is important to note that the use of this analytical technique is not a shortcut to problem-solving, we do not

mean that mean designers can “click here to save the world” (Morozov, 2013). The technique reveals features of the focal system that designers must further investigate through systems and design research. The results of the analysis should be interrogated and vetted by system stakeholders before acting upon them. For instance, designers may present the initial results of leverage analysis to stakeholders for critique. If any of the insights or conclusions generated by the technique are questionable, the designers and stakeholders should work together to understand the results and to re-examine the structure of their model for any flaws. Then, leverage analytics should be re-run, and the results discussed again until stakeholders reach consensus that the results are valid and useful.

Moreover, the results of leverage analysis are not free of interpretation. As the demonstration in section 5 shows, designers and stakeholders must reflect upon the rankings provided by leverage measures. Some results will be senseless, especially around the system’s boundary. For example, suppose you have developed a systems model that depends heavily on a regional budget or economy that does not include detailed modelling of broader economic forces. Leverage analysis may suggest that phenomena related to “the value of exported goods” is an accessible point of intervention. Obviously, regions rarely have granular control of the value of their goods. However, because the designers of the model decided to draw a boundary around the system at this point, the forces that determine the value of exported goods are largely left out. Leverage analysis algorithms cannot know this, so they return a meaningless result. (For an example of this behaviour, see Murphy, 2016, p. 59).

Similarly, designers may use the same leverage measure to investigate different features of the system. High-closeness values can highlight phenomena that are both useful indicators of systems change and potential barriers to change. Another aspect of leverage analysis open to interpretation is the threshold a designer uses to determine whether to include a phenomenon in a collection of systems features based on leverage measures. In the demonstration in section 5, for example, we list approximately the ten most highly-ranked phenomena for each leverage measure to develop our Systemic Theory of Change. This choice of ten items is arbitrary as it simplifies the demonstration. We sometimes display more than ten, particularly when phenomena are tied in value for the given leverage measure. In practice, however, designers may choose a different threshold for inclusion in a given list of systems features. Some approaches to selecting this cut-off include: (1) selecting a number of the highest- or lowest-ranked phenomena, as we have; (2) selecting a cut-off value for the phenomena based on the actual reported centrality measure; or (3) actively reviewing each potential candidate from the top of the list down and removing them from the list if they do not logically fit (as discussed earlier in this paragraph), until the designer has a satisfying number of phenomena in that collection of features. In general, leverage analysis gives designers a useful place to start, but using the results still requires logical interpretation and argument.

In the next section, we show how designers can use leverage analysis on CLDs to create SToCs.

Creating a Systemic Theory of Change As discussed earlier, a Theory of Change model typically consists of chains of outcomes beginning with strategic

activities enacted by designers and ending with the ultimate results they aim to achieve. ToCs may have any number of stepped outcomes in between these activities and goals.

As you can see from Table 1, some leverage measures identify key leverage points in the system—places where a little effort may yield significant results (Meadows, 1997). Others identify phenomena that may be easily shifted by design interventions. Others identify phenomena that may act as barriers for change or that are likely to be good indicators that change is happening.

Design management for wicked problems

7

The development of a Systemic Theory of Change (SToC) using a CLD is, therefore, quite straightforward. When a conventional CLD is too complex to easily interpret “start” and “end” points and other features relevant to the design and management of an intervention, designers can use leverage analysis to identify ToC pathways in the model.

We begin by identifying our goal(s): the ultimate objectives in the system that designed interventions will address. Then, calculate the model’s leverage measures.3 Designers may then begin analysing the results.

First, phenomena that rank highly in three categories are candidates for systemic leverage. Those are (1) eigenvector (phenomena that are close to other highly connected elements), (2) reach (phenomena with the smallest pathways to every other phenomena), and (3) eigenvector-weighted reach (highly connected phenomena with the shortest paths to all other phenomena). Small degrees of change in these phenomena will likely yield more significant change in the system than efforts placed elsewhere. However, high-leverage phenomena are not necessarily easily manipulated. Therefore, look at phenomena that rank highly in closeness and remove those from your list of initial phenomena.

The remaining list of leverage points may still not be the easiest phenomena to work with in the system. Instead, phenomena that rank the lowest on closeness (phenomena with the longest paths to all other phenomena) measure may be more accessible. The resulting list of high-access and high-leverage phenomena offer starting points for our SToC.

The next step is to choose indicators of change. Since systems are dynamic and often feature long-term delays, measuring progress by only looking for change in our ultimate goals may not be reasonable. Phenomena ranking highly on indegree (those that have many incoming connections) and closeness (those with the shortest distance to the rest of the system) leverage measures are candidates for systemic signals of change.

Our penultimate step in the development of a SToC is the identification of bottlenecks and barriers. Again, phenomena that rank highly on closeness measures are likely to resist change as many competing phenomena influence them. At the same time, high-betweenness (those phenomena that appear the most frequently between all others) phenomena are gateways in the system. These phenomena are located along many pathways of change in the system, and they may represent bottlenecks in any change theory.

Finally, designers may craft the SToC. To do this, find your fulcra: the high-access and high-leverage phenomena identified earlier. Working with each of these phenomena in turn, chart a path of connections and other phenomena to the goals identified at the outset (while being mindful of the barrier and bottleneck phenomena). Include some systemic signals—high-indegree phenomena—as potential indicators that your strategies are working. There may be multiple pathways from any given fulcra to any given goal. The existence of multiple pathways is good—it gives you numerous options from which you may build your SToC.

Each of the pathways you identify are analogous to the outcome chains of conventional ToC models. You may, therefore, adopt all or a subset of these as the outcome chains of your SToC. It is now your task to look at these chains of outcomes and design interventions that traverse each chain. Designing interventions may mean, for instance, developing new solutions that address the fulcra at the beginning of these chains. It may also mean designing programming that accounts for any crucial bottlenecks and barriers.

The resulting map shows linear pathways—systemic change sub-theories—representing semi-linear theories of change that chart a course within the complex nonlinear structure of the system. It begins with strategic interventions that act on systemic opportunities, which in turn influence high-leverage phenomena in the system. It arcs through outcomes that are influenced by these phenomena, potentially passing through signals that the change is working and avoiding barriers or bottlenecks that may resist your strategy. Finally, it ends on the ultimate goal of the initiative.

In the next section, we briefly demonstrate this technique using a previously-developed CLD.

Demonstration Here we present a simple demonstration of how SToCs may be developed and used. We base this example on a

CLD (figure 2) of Canada’s innovation system created by Fascinato, McCallum, Berman, & Murphy (2016). The model visualizes systemic opportunities for advancing innovation and entrepreneurship in Canada. The creators collected data for the model from a series of interviews with entrepreneurs and investors.

3 The systems mapping software Kumu.io has built-in algorithms for computing these scores via the Social Network Analysis feature. Please see Murphy & Jones, 2020 for a complete discussion of these measures and how they are derived.

RYAN J. A. MURPHY, PETER JONES

8

Figure 2. Canada’s innovation system Causal Loop Diagram.

This model of Canada’s innovation system serves as a useful demonstration as the model is reasonably complex;

it is not easy to decipher at a glance. The process of developing a SToC based on the model should show how useful it is to highlight and use model features with the help of leverage analysis.

Identifying goals We begin the process of developing the SToC from the model by selecting goal phenomena. This selection

involves logical argument: designers must interpret the model to identify the phenomena that align with their initiative’s vision or mission. In the present case, our goal phenomena are entrepreneurship, chosen because this map was developed to identify ways of strengthening the nation’s innovation ecosystem to bolster entrepreneurial success.

Finding leverage Next, we calculate eigenvector, reach, and eigenvector-weighted reach leverage measures. We list the top ten

phenomena ranked by each of these measures in order in table 2. These measures point to phenomena in which change is likely to yield significant shifts elsewhere in the system.

Table 2. Potential high-leverage systemic interventions.

High-eigenvector phenomena High-reach phenomena High-eigenvector-weighted-reach phenomena

entrepreneurship businesses scaling businesses scaling successful fundraising startup investment availability of incubators entrepreneurial events/groups strong businesses successful exits entrepreneurial culture access to capital startup investment potential startups availability of incubators entrepreneurial culture

Design management for wicked problems

9

brain drain successful exits entrepreneurship businesses scaling entrepreneurial culture access to capital fear of risk strong Canadian economy brain drain efficacy mindset low SES strong businesses successful exits brain drain potential startups

Opportunities for intervention Once we have a list of potential fulcra, we must find phenomena that are easily accessed to use as levers on

those fulcra. To do this, we look for the phenomena with the lowest-ranking closeness values. Later, when we chart outcome chains to our goals, we will begin with interventions targeting these accessible phenomena and seek pathways from them to the high-leverage phenomena identified above.

Table 3. Potential high-access phenomena.

Low-closeness phenomena K12 teachers parents efficacy mindset university research entrepreneurial programs & education commercialization of campus research campus accelerators/incubators potential startups “shiny” startups weak domestic investment community commercialization of campus research

As the phenomena on this list should serve as inspiration for potential programming directions, it is essential to

curate a list of realistically opportunistic phenomena. In general, no one can directly influence everything. Therefore, designers must be careful to focus the design of changemaking actions and activities on phenomena they can influence directly and to design a theory of change that connects those actionable phenomena to high-leverage phenomena, shifting the whole system. So, we must examine our list of systemic opportunities and remove anything we really cannot directly influence. On the above list, that is potential startups, as it is unlikely that we will start our businesses to enable systemic change. Removing these uncontrollable phenomena leaves us with the nine items in table 4.

Table 4. Final list of high-access phenomena.

Low-closeness phenomena K12 teachers parents efficacy mindset university research entrepreneurial programs & education commercialization of campus research campus accelerators/incubators “shiny” startups weak domestic investment community commercialization of campus research

Seeking signals The next step is to identify high-indegree and high-closeness measures to try to find useful indicators of

systemic change. Phenomena with high values in these leverage measures are more inwardly connected, relatively, than others in the system.

Table 5. Potential signalling phenomena.

High-indegree phenomena High-closeness phenomena startup investment strong businesses

RYAN J. A. MURPHY, PETER JONES

10

entrepreneurship startup investment entrepreneurial culture businesses scaling access to capital low SES fear of risk access to capital strong Canadian economy successful exits brain drain failed executions campus entrepreneurship natural resources potential startups entrepreneurial culture efficacy mindset strong Canadian economy

Once again, we must review our results to remove anything impractical or unrealistic in the real world. In this case, at this stage, it is challenging to use phenomena like fear of risk, potential startups, efficacy mindset, entrepreneurial culture, strong businesses, failed executions, and natural resources as signals of change because they represent immaterial concepts. In turn, we remove those from the list, leaving us with the following table:

Table 6. Final list of signalling phenomena.

High-indegree phenomena High-closeness phenomena startup investment startup investment entrepreneurship businesses scaling access to capital low SES strong Canadian economy access to capital brain drain successful exits campus entrepreneurship strong Canadian economy

Beware of barriers and bottlenecks Before we finalize a SToC, we must use closeness and betweenness measures to identify phenomena that may

act as barriers or bottlenecks in change strategies.

Table 7. Barrier and bottleneck phenomena.

High-closeness phenomena High-betweenness phenomena strong businesses entrepreneurship startup investment startup investment businesses scaling successful fundraising low SES businesses scaling access to capital entrepreneurial culture successful exits strong businesses failed executions brain drain natural resources access to capital entrepreneurial culture strong Canadian economy strong Canadian economy access to investors

Developing the Systemic Theory of Change Finally, with the lists above, we must chart pathways between potential interventions (i.e., those listed in table

2) and the goals, being sure to travel through high-leverage and signalling phenomena, and avoiding or deliberately addressing barriers and bottlenecks.

As you have likely already noticed, some interpretation is necessary. One issue is that many phenomena represent different systems features at the same time. For instance, our goals—entrepreneurship and strong Canadian economy are each a candidate for leverage, signals, and barriers/bottlenecks. A designer may want to use one as a signal or bottleneck within a SToC that targets the other.

Perhaps the most critical issue is that no obvious answers emerge. It is still up to the designer to review the results of the SToC process and make judgements to develop the final SToC model. Visualization tools can help make this easier. For instance, we can use Kumu.io to highlight and flag the different kinds of systems features (figure 3).

Design management for wicked problems

11

Figure 3. Visualizing leverage analysis in a CLD.

It is easy to see how patterns emerge in the system using this kind of visualization. For example, high-access phenomena appear to cluster around the high-leverage phenomena fear of risk and efficacy mindset. The idea that parents and educators may help elevate entrepreneurial youth can be abductively (Kolko, 2010; Peirce, 1998) connected to the high-leverage phenomena availability of incubators in the form of growing school-based entrepreneurial incubators. We model a possible systemic change sub-theory using this idea in figure 4.

RYAN J. A. MURPHY, PETER JONES

12

Figure 4. A systemic change sub-theory with school-based entrepreneurial incubators.

Another pattern we identify is a cluster of barriers. The first pertains to successfully-seeded startups: startup

investment, access to capital, successful exits, successful fundraising, businesses scaling, access to investors, and brain drain. This cluster is highly connected: each connects to at least one other in the cluster directly. This connectivity is true in practice, too—a robust entrepreneurial ecosystem depends on an engine of growing startups that attract attention and investment to the ecosystem, and Canada’s entrepreneurs are often invited to the investment opportunities in the USA (ergo, brain drain; Fascinato et al., 2016). An initiative that provides matching funds to within-nation entrepreneurial investment might be a powerful way of addresses these barriers simultaneously (see figure 4 for a relevant sub-theory).

Design management for wicked problems

13

Figure 5. A systemic change sub-theory focused on addressing a set of barriers and bottlenecks.

Both figures 4 and 5 look like conventional ToC models. However, they include the feedback loops that drive the

entire model. More importantly, we derived them from a systemic context with an appreciation for the dynamic complexity within which they will operate.

6. Discussion In this paper, we showed how a novel approach to understanding the structure of complex problems—leverage

analysis—could be applied to a CLD to develop a SToC. We believe this approach is powerful, making it easier to identify systemic change sub-theories in the most complex of CLDs. In turn, designers can compose several such well-aligned sub-theories into a theory of change for the management of their designed interventions.

We believe, however, that our present implementation is rudimentary. The process we have defined does not take advantage of all of the features of leverage analysis (namely the elements of structural dominance, such as level and cycle partitions; Murphy & Jones, 2020). The process also does not intentionally account for feedback loops or systems archetypes (Braun, 2002). In the future, we hope that this technique is advanced further to include these aspects of systems structure.

Though the goal of using leverage analysis to create SToC models was simplifying the complexity of conventional CLDs, we were not wholly successful. While leverage analysis highlights critical features of systems models, it does so by adding information to the model, which arguably adds as much complexity as it takes away. The process still requires a skilled designer who can organize and interpret the results to make effective use of them.

The demonstration that we have provided is just that—a demo of a novel approach to theorizing about and practicing systems change. The most obvious limitation is that this technique remains unproven in a real-world setting. Ideally, a future research project will explore these ideas and test them further. One way to do this is through engaging stakeholders in the development of a system CLD, working to develop a SToC based on the CLD, and then conducting interviews to assess whether the results are valid. (Researchers could even compare a SToC process with a conventional ToC modelling approach, and participants could review and rank the resulting models’ effectiveness.)

RYAN J. A. MURPHY, PETER JONES

14

Conclusion In this paper, we have merged ToC models with systems models (CLDs) under the banner of the design

management of complex problems. Design management is increasingly called upon to address wicked problems. To this end, there exists no standard management tool with which to design effective solutions to these intractable challenges.

Many interventions in wicked problems depend on the use of ToC models. Most of these models are linear in form. Ergo, no matter how complex the outcomes pathway they represent may be, this linearity likely fails to capture the counterintuitive dynamics common in complex adaptive systems (Gharajedaghi, 2011, pp. 111–119; Sterman, 2009, chapter 8). The linear, reductive nature of ToC models may tempt designers instead to use systems maps, such as a CLD. However, the complex dynamics of such models do not yield easily to the hierarchies of outcomes that make theories of change useful. An effective theory of change is a pathway that leads from some origin(s) that changemakers can act upon through a chain of outcomes to the ultimate impact they hope to have upon the system. Nonlinear models of systems, by definition, do not have such start- and end-points.

An opportunity therefore lies in this tension between the useful hierarchies of theories of change and the nonlinear complexity of systems models. By combining ToC and CLD models into a SToC, designers can simultaneously appreciate the feedback dynamics of complex problems while surfacing a hierarchical structure of influence in the system. SToCs map the complex interactions in a system from higher-leverage actions to higher-impact results, maintaining the rich complexity of systems models while providing actionable insights for the design of interventions.

References Abercrombie, R., Boswell, K., & Thomasoo, R. (2018). Thinking big: how to use theory of change for systems change.

New Philanthropy Capital. Alford, C. (2017a). 6 principles for building effective theories of change and the Kumu features that make it

possible. In Too Deep. https://blog.kumu.io/6-principles-for-building-effective-theories-of-change-and-the-kumu-features-that-make-it-possible-af71857cea9f

Alford, C. (2017b). How systems mapping can help you build a better theory of change. In Too Deep. https://blog.kumu.io/how-systems-mapping-can-help-you-build-a-better-theory-of-change-4c85ae4301a8

Alford, C. (2017c). Two approaches for combining theories of change and system maps. In Too Deep. https://blog.kumu.io/two-approaches-for-combining-theories-of-change-and-system-maps-5b556091c880

Bellefontaine, T. (2012). Innovation Labs: Bridging Think Tanks and Do Tanks. Policy Horizons Canada. Boyer, B., Cook, J. W., & Steinberg, M. (2011). In studio: Recipes for Systemic Change.

http://www.helsinkidesignlab.org/pages/studio-book.html Boyer, B., Cook, J. W., & Steinberg, M. (2013). Legible Practices: Six Stories about the Craft of Stewardship.

http://www.helsinkidesignlab.org/peoplepods/themes/hdl/downloads/Legible_Practises.pdf Braun, W. (2002). The System Archetypes. In.

http://www.albany.edu/faculty/gpr/PAD724/724WebArticles/sys_archetypes.pdf Brown, V. A., Harris, J. A., & Russell, J. Y. (2010). Tackling wicked problems through the transdisciplinary

imagination. Earthscan. Buchanan, R. (1992). Wicked Problems in Design Thinking. Design Issues, 8(2), 5. 10.2307/1511637 Buchanan, R. (2015). Worlds in the Making: Design, Management, and the Reform of Organizational Culture. She Ji:

The Journal of Design, Economics, and Innovation, 1, 5-21. 10.1016/j.sheji.2015.09.003 Chen, C.-C., & Crilly, N. (2016). From modularity to emergence: a primer on the design and science of complex

systems. Clark, K., & Smith, R. (2010). Unleashing the Power of Design Thinking. Design Management Review, 19(3), 8-15.

10.1111/j.1948-7169.2008.tb00123.x Fascinato, D. T., McCallum, R., Murphy, R. J. A., & Berman, M. (2016). Canada’s Real Innovation Gap: Leverage

points & opportunities for change in the Canadian innovation system. https://fulcra.design/innovation-systemics Forrester, J. W. (1995). Counterintuitive behavior of social systems. collected papers of Jay W. Forrester.

http://live.iugaza.edu.ps/NR/rdonlyres/Sloan-School-of-Management/15-988Fall-1998-Spring-1999/34262D53-CE8F-41B4-947D-4F815D81E9DB/0/behavior.pdf

Freeman, L. C. (1977). A Set of Measures of Centrality Based on Betweenness. Sociometry, 40(1), 35-41. 10.2307/3033543

Design management for wicked problems

15

Friedman, K., Lou, Y., Norman, D., Stappers, P. J., Voûte, E., & Whitney, P. (2014). DesignX: A Future Path for Design. https://jnd.org/designx_a_future_path_for_design

Funnell, S. C., & Rogers, P. J. (2011). Purposeful Program Theory: Effective Use of Theories of Change and Logic Models. Jossey-Bass.

Gharajedaghi, J. (2011). Systems thinking: managing chaos and complexity: a platform for designing business architecture. Morgan Kaufmann.

Hassan, Z. (2014). The social labs revolution: a new approach to solving our most complex challenges. Berrett-Koehler Publishers, Inc.

Irwin, T. (2015). Transition Design: A Proposal for a New Area of Design Practice, Study, and Research. Design and Culture, 7(2), 229-246. 10.1080/17547075.2015.1051829

Ito, J. (2017). Resisting Reduction: A Manifesto. Journal of Design and Science. 10.21428/8f7503e4 Jensenius, A. R. (2012). Disciplinarities: intra, cross, multi, inter, trans. Alexander Refsum Jensenius.

https://www.arj.no/2012/03/12/disciplinarities-2 Jones, P. H. (2014). Systemic Design Principles for Complex Social Systems. In G. S. Metcalf (Ed.), Social Systems and

Design (pp. 91-128). Springer Japan. http://link.springer.com/chapter/10.1007/978-4-431-54478-4_4 Jones, P. (2018). Contexts of Co-creation: Designing with System Stakeholders. In P. Jones & K. Kijima (Eds.),

Systemic Design: Theory, Methods and Practice (pp. 3-52). Springer Japan. http://link.springer.com/10.1007/978-4-431-55639-8_1

Kim, D. H. (1992). Guidelines for Drawing Causal Loop Diagrams. The Systems Thinker, 3(1), 5-6. Klein, G., Moon, B., & Hoffman, R. R. (2006). Making sense of sensemaking 1: Alternative perspectives. IEEE

intelligent systems, 21(4), 70-73. http://www.computer.org/csdl/mags/ex/2006/04/x4070-abs.html Kolko, J. (2010). Abductive thinking and sensemaking: The drivers of design synthesis. Design Issues, 26(1), 15-28.

http://www.mitpressjournals.org/doi/abs/10.1162/desi.2010.26.1.15 Kolko, J. (2012). Wicked Problems: Problems Worth Solving: A Handbook & A Call to Action. AC4D. Mackinnon, A. (2006). Mapping Change: Using a Theory of Change to Guide Planning and Evaluation.

http://grantcraft.org/wp-content/uploads/sites/2/2018/12/theory_change.pdf McBride, M. (2010). Design Management: Future Forward. Design Management Review, 18(3), 18-22.

10.1111/j.1948-7169.2007.tb00227.x McGowan, K. A., Westley, F., Fraser, E. D. G., Loring, P. A., Weathers, K. C., Avelino, F., Sendzimir, J., Chowdhury, R.

R., & Moore, M.-L. (2014). The research journey: travels across the idiomatic and axiomatic toward a better understanding of complexity. Ecology and Society, 19(3). https://www.jstor.org/stable/26269642

Meadows, D. (1997). Leverage Points: Places to Intervene in a System. Donella Meadows Institute. http://www.donellameadows.org/wp-content/userfiles/Leverage_Points.pdf

Morozov, E. (2013). To Save Everything, Click Here. PublicAffairs. https://play.google.com/store/books/details?id=fdggBahA1qsC&source=gbs_api

Murphy, R., & Jones, P. (2020). Leverage analysis. FormAkademisk - forskningstidsskrift for design og designdidaktikk, 13(2). 10.7577/formakademisk.3384

Murphy, R. J. A. (2016). Innovation Education OCAD University. http://openresearch.ocadu.ca/id/eprint/1344 Norman, D. A., & Stappers, P. J. (2015). DesignX: Complex Sociotechnical Systems. She Ji: The Journal of Design,

Economics, and Innovation, 1(2), 83-106. 10.1016/j.sheji.2016.01.002 Oxman, N. (2016). Age of Entanglement. Journal of Design and Science. Peirce, C. S. (1998). Abduction. In Collected Papers of Charles Sanders Peirce (Vol. 7, pp. 136-144). Quinn, R. E., & Cameron, K. S. (1988). Paradox and transformation: Toward a theory of change in organization and

management. Ballinger Publishing Co/Harper & Row Publishers. Rittel, H. W. J., & Webber, M. M. (1973). Dilemmas in a general theory of planning. Policy Sciences, 4(2), 155-169.

http://link.springer.com/article/10.1007/BF01405730 Ryan, A. (2014). A Framework for Systemic Design. FORMakademisk–research journal for design and design

education, 7(4). https://journals.hioa.no/index.php/formakademisk/article/view/787 Sanders, E. B.-N., & Stappers, P. J. (2008). Co-creation and the new landscapes of design. CoDesign, 4(1), 5-18.

10.1080/15710880701875068 Sevaldson, B., & Jones, P. (2019). An Interdiscipline Emerges: Pathways to Systemic Design. She Ji: The Journal of

Design, Economics, and Innovation, 5(2), 75-84. 10.1016/j.sheji.2019.05.002 Slavin, K. (2016). Design as Participation. Journal of Design and Science. http://jods.mitpress.mit.edu/pub/design-

as-participation

RYAN J. A. MURPHY, PETER JONES

16

Stember, M. (1991). Advancing the social sciences through the interdisciplinary enterprise. The Social Science Journal, 28(1), 1-14. 10.1016/0362-3319(91)90040-B

Sterman, J. D. (2009). Business dynamics: systems thinking and modeling for a complex world. Irwin/McGraw-Hill. Theory of Change: A Practical Tool. (2004). https://www.aecf.org/m/resourcedoc/aecf-theoryofchange-2004.pdf Thering, S., & Chanse, V. (2011). The Scholarship of Transdisciplinary Action Research: Toward a New Paradigm for

the Planning and Design Professions. Landscape Journal, 30(1), 6-18. 10.3368/lj.30.1.6 Van Alstyne, G., & Logan, R. K. (2007). Designing for Emergence and Innovation: Redesigning Design. Artifact, 1(2),

120-129. 10.1080/17493460601110525 Weisbord, M., & Janoff, S. (2007). Get the Whole System in the Room. The Journal for Quality and Participation,

30(3), 4-8. http://search.proquest.com/docview/219115072/abstract/34CBD29280604DB0PQ/1 Westley, F., Goebey, S., & Robinson, K. (2012). Change lab/design lab for social innovation. Waterloo: Waterloo

Institute of Social Innovation and Resilience, University of Waterloo. http://www.marsdd.com/wp-content/uploads/2014/05/Paper_FINAL_LabforSocialInnovation.pdf