The chart above, drawn from Everett Rogers, Diffusion of Innovations Fig. 7-1 (5th ed. 2003), shows the adoption of hybrid seed corn by farmers in two Iowa communities. The dashed line on the bottom shows the number of adoptions by year. The solid line on top shows adoption on a cumulative basis. The first farmer in the sample adopted hybrid seed corn in 1927. 15 years later, in 1941, the last four farmers made the switch.
The dashed line is a real-world example of a Rogers Diffusion Curve. See Post 004 (discussing curve); Post 007 (discussing adopter types). Likewise, the solid line is a real-world example of the S-shaped curve. The farmers switched to hybrid seed corn because it was more bountiful, disease resistant, and drought resistant than traditional methods. The chart above is useful because it shows the common diffusion pattern of (1) a prolonged period of slow adoption, even for a highly advantageous innovation; and (2) a short period of rapid adoption. Cf. Post 016 (showing histogram with long innovator tail).
In the case of the Iowa farmers, the prolonged period of slow adoption was not a random event. Few if any farmers would have adopted hybrid seed corn but for agronomists from Iowa universities. The agronomists were necessary to help the innovator and early adopter farmers understand and use this new technology. When some of the more influential early adopter farmers met with success, they shared their experiences with other farmers. As the benefits of the innovation were experienced by the early majority farmers, adoption spread like a social contagion through the two Iowa communities.
In this real-life example, the university agronomists were the change agents. And the influential early adopter farmers were the opinion leaders. This post (020) explains the crucial role played by these two types of actors. It also emphasizes how these concepts apply to the current challenges facing the legal industry.
Post 020 is part of Legal Evolution’s foundational series on diffusion theory. Readers seeking to influence innovation within the legal industry will be more successful if they obtain and apply this background knowledge. Care has been taken to make this information non-technical and accessible.
Before the farmers could adopt hybrid seed corn, they needed “awareness-knowledge”, which is knowledge that such a technology exists. However, there can be a considerable lag between awareness-knowledge and actual adoption of a new innovation. This dynamic is shown in the chart below, which is based on the same study of Iowa farmers. (This chart was first shown in Post 008 to help illustrate Rogers’ rate of adoption model).
For the typical farmer, roughly six years elapsed between hearing about hybrid seed corn and adopting it. In addition to inexperience and uncertainty with hybrid seeds, the lag time was due to the sheer novelty of the innovation, which was rooted in laboratory science and at odds with longstanding views regarding how to grow the best corn. See Post 008 (discussing how complexity and cultural incompatibility can impede adoption of innovations).
Yet, in the early 20th century, agricultural production was a matter of national security, as World War I had driven home the importance of a secure and bountiful domestic food supply. Farmers had also become a formidable legislative lobby. Thus by 1920, there were more than 3,000 agricultural extension workers funded by a mix of federal, state, and county agencies.
Remarkably, despite the benefit of a large and well-financed change establishment delivering an unalloyed benefit to farmers, the uptake was far from rapid. The key sociological question was “why?” The parallel applied research question was “can the rate of adoption be accelerated?” The answer to the latter question was yes, thus creating foundational research that would eventually result in a general theory for how innovation diffuses.
Early versus Middle-Late Diffusion: The S-shaped Curve
Diffusion theory is part of an applied research tradition that seeks to enable change strategies that work in a controlled and predictable way. The core insight is that the diffusion of an innovation is a process that occurs through a social system. See Post 004 (discussing Rogers Diffusion Curve). In most cases, the process begins with a need or problem and a desire by some members of the social system to find and implement a solution.
For the purposes of this post, we can divide the diffusion process into two stages: an early stage, characterized by a relatively long period of slow adoption (base of the S-shaped curve that starts with the long innovators tail); and a middle-late stage, characterized by rapid adoption over a relatively short period (the steep portion of the S-shaped curve followed by a plateau).
Between these two stages, the early stage is far more tenuous and fragile. This is because it requires a member of the social system to (1) obtain knowledge of an innovation, (2) evaluate its relative benefits and costs, (3) make an affirmative adoption decision, (4) successfully implement the innovation, and (5) confirm the existence of the desired results. In substance, this is a time-consuming and potentially expensive experiment that could fail. Obviously, only a sub-segment of any population would be willing and/or able to bear this risk.
In the diagrams above, the early stage would roughly correspond to the 1924 to 1933 time period. Many farmers had heard about hybrid seed corn, but only a handful had adopted it. The early stage typically comes to an end when the social system’s opinion leaders become part of the adopter group and can vouch for the innovation’s effectiveness. Rogers writes, “[T]he [cumulative] diffusion curve is S-shaped because once opinion leaders adopt and begin telling others about an innovation, the number of adopters per unit of time takes off in an exponential curve” (p. 300).
The middle-late stage of diffusion begins with the rapid ascension of the S-shaped curve (1934 to 1941). In Diffusion of Innovations, Rogers discusses the concept of “critical mass”, which is the point at which enough adoption has occurred that further adoption becomes “self-sustaining.”
[T]he heart of the diffusion process is the modeling and imitation by potential adopters of their near peers’ experiences with the new idea. In deciding whether or not to adopt an innovation, individuals depend mainly on the communicated experience of others much like themselves … . The subjective evaluations of an innovation flow mainly through interpersonal networks. (p. 330).
On a micro-level, change is occurring because individuals are observing each other and responding to social proof. Each individual in the social system has a “threshold” of proof needed to spur change. Once the middle-late stage of diffusion is reached — i.e., the steep part of the S-shaped curve — the adoption process become less deliberative and more imitation of people in their close social network. Thus adoption moves like dominoes from early adopters to the early majority to the late majority to the laggards. Although thresholds operate at an individual level and vary by adopter type, at a system level, their aggregate effect is to create a critical mass that leads to a tipping point.
In the case of culturally novel and complex innovations, critical mass is seldom reached without the participation of opinion leaders. Thus, it is important to understand their characteristics and attributes.
Opinion leaders are rarely innovators and are not necessarily early adopters. Their relative position among the five adopter types depends upon the norms of the social system. Within the tradition-bound legal industry, the opinion leaders may be members of the early majority, refusing to adopt change without a very high standard of proof.
Roger defines opinion leadership as “the degree to which an individual is able to informally influence other individuals’ attitudes or overt behavior in a desired way with relative frequency.” Thus, among corporate law firms, Cravath Swaine & Moore is clearly an opinion leader. See, e.g., Cravath Raising Starting Salaries to $180,000, WSJ, 6/6/16 (reporting that “[c]hange is likely to spawn a wave of copycat moves”). Likewise, Harvard Law leads the way in legal education. See Harvard Law is now accepting the GRE. Could other schools follow?, Boston Globe, 3/21/17. Yet, neither institution is widely viewed as an early adopter. In less conservative social systems, however, the overlap between opinion leaders and early adopters would be significantly larger. Cf. Post 007 (discussing the influence and sway of early adopters).
A key feature of opinion leaders — and one that usually renders innovators unfit for the role — is their strong conformity to social system norms. Respect for norms is necessary to obtain the trust and allegiance of other adopter types. Note that the value at play here may be less about innovation than power and influence, as opinion leaders can be disregarded or toppled. Rogers writes:
The interpersonal relationships between opinion leaders and followers hang in a delicate balance. If an opinion leader becomes too innovative, or adopts a new idea too quickly, followers may begin to doubt his or her judgment. One role of the opinion leader in the social system is to help reduce uncertainty about an innovation … . To fulfill this role, an opinion leader must demonstrate prudent judgment decisions about adopting new ideas. So the opinion leader must continually look over his or her shoulder and consider where the rest of the system is regarding new ideas. (p. 319)
On balance, however, opinion leaders tend to be distinguished by several attributes, at least as compared to other members of the social system. Opinion leaders tend to have:
- greater connections to the outside world (more “cosmopolite”)
- greater exposure to diverse media
- higher levels of social engagement
- higher socioeconomic status
- more innovative than followers
- greater exposure to change agents.
Regarding point #6, below is a bar chart showing the average number of change agent contacts per year for a group of farmers in Brazil. It is drawn from an agricultural diffusion study conducted by Rogers and other researchers.
The key takeaway from this chart is that change agents are sources of innovative ideas. Rogers demonstrates the empirical connection between the Mark Granovetter’s Strength of Weak Ties theory and access to high-impact information. In Granovetter’s well-known study of how people found employment, connections to far-flung cliques and social groups, albeit weak, were far more powerful than local networks of friends and family. Thus, peripheral “weak” ties tend to be more informationally rich than the dense connections at the center of the social system.
Change agents and their ideas enter a social system through these weak ties. Although change agents find the greatest receptivity with innovators, their success often hinges upon their ability to influence opinion leaders.
A change agent is defined as “an individual who influences clients’ innovation-decisions in a direction deemed desirable by the change agency” (p. 27). Their biggest impact is felt during the tenuous early stage of diffusion.
In the agricultural study, the change agents were government-funded university agronomists who were hired to help farmers adopt new technology. The goal was to boost agricultural production. However, in other contexts, change agents could be public health workers trying to reduce the spread of HIV; teachers introducing new curricula and materials to public schools as part of a broader “new math” movement; or salespeople selling enterprise software to large organizations. Indeed, this last example became the basis for the Silicon Valley classic, Crossing the Chasm (1991), which I’ll discuss in the next and final foundational post.
In cases of complex or novel innovations, change agents are necessary to fill gaps in technical knowledge and know-how. These change agents typically have a significantly greater technical competence than members of the “client” social system. Unfortunately, this superior know-how often creates communication and cultural gaps that are difficult to bridge. This phenomenon is very much present in the legal industry circa 2017 as lawyers and legal educators struggle to learn new work methods grounded in data, process, and technology. The gap is undoubtedly the most visible with artificial intelligence.
The Tradeoff between Information Impact and Communication Ease
Communication and cultural gaps are most likely to occur when change agents are very dissimilar from members of the social system. A straightforward example would be lawyers needing to learn technical information from data scientists, software developers, and process engineers. This dissimilarity is referred to as “heterophily” (the technical term used in diffusion theory). Although there is an enormous breadth of knowledge in these pairings, and thus the latent potential for high-impact knowledge transfer, communication tends to be slow, arduous, and uncomfortable. Thus, except among innovators and early adopters, persistence in heterophilous pairings is rare.
Conversely, when two individuals are very similar (homophily), such as two lawyers who attended the same law school and work in the same area of law, any communication gap is likely to be small or non-existent. Unfortunately, that pairing is unlikely to transmit high-impact information, as their base of knowledge is too similar. Cf. Scott Page, The Difference (2008) (economist demonstrating that diverse teams outperform homogenous teams on tasks requiring creativity and innovation). Thus, in a very real sense, law firms, legal departments, and law faculty cannot be leaders in innovation if their information gathering and strategizing is substantially limited to high-level meetings among lawyers. Remarkably, many will try.
The diagram below illustrates the conundrum.
On the far left side of the diagram, the transfer of high-impact information is impeded by significant communication and cultural gaps between change agents and members of the social system. Simply stated, they are too dissimilar to connect. On the far right side, communication is easy and fluid, but there is little or no novel information to share.
, when an effective change agent works with innovators and early adopters and eventually receptive opinion leaders, a knowledge-rich exchange is possible (center left). After that, diffusion follows the example of opinion leaders, adopting sequentially early majority, late majority, and laggards (center right). See Post 007 (profiling the five adopter types).
Effective Change Agents
The theory of change agents may seem relatively simple. However, when the desired change is complex and impinges on social and cultural norms, the change agent’s job is enormously difficult. Rogers observes:
As a bridge between two differing systems, the change agent is a marginal figure with one foot in each of two worlds.
In addition to facing this problem with social marginality, change agents also must deal with the problem of information overload, the state of an individual or a system in which excessive communication inputs cannot be processed and utilized, leading to breakdown. …. By understanding the needs of the clients, the change agent can selectively transmit to them only information that is relevant. (p. 368-69).
My own interest in diffusion theory was borne of my six years at Lawyer Metrics. See Post 004. As an applied research company, we created data analytics tools for legal service organizations. Although the company had PhD social scientists who could build highly sophisticated quantitative models, our biggest challenge was finding ways to present data that lawyers could process, understand, and accept. On many occasions, we quipped that the statistical work was simple by comparison.
As I survey the legal landscape in 2017, I see the same challenges affecting many legaltech start-ups. Most early stage entrepreneurs emphasize the technical features of their product, because they know and love its full range of capabilities. Yet, this perspective places them at a high risk of failure.
Below is a model of change agent effectiveness based on Chapter 9 of Diffusion of Innovations. Suffice it to say, it fully aligns with my professional experience.
The original rate of adoption model in Post 008 listed five categories of variables that influence the rate of innovation adoption. The fifth category was “Efforts of Changes Agents.” The model above provides additional detail for that category. Cf Post 011 (discussing importance of the first category, “Perceived Attributes of Innovation,” to explain the difference between fast and slow innovations, even when the innovations at issue can save human life).
- Making contact with clients (+). Frequent contact builds familiarity and creates opportunities to establish credibility and trust.
- Client orientation (+). Is the change agent trying to solve the clients’ problem or trying to advance their own agenda (e.g., make a sale)? If the change agent is listening, they can learn ways to modify and improve their innovation.
- Client empathy (+). A change agent is more effective when she or he can see the world through the eyes of the client.
- Homophily with clients (+). Can the change agent look and act like an insider? In the legal industry, change agents with law degrees generally have an easier time because of a common experience and background with most clients.
- Credibility in the clients’ eyes (+). Can the change agent fluidly answer tough questions? If the client must trust the change agents’ judgment, do the change agents possess the credentials and background to understand the underlying innovation?
- Working thru Opinion Leaders (+). Rogers observes, “The time and energy of the change agents are scarce resources” (p. 388). Engaging opinion leaders is the most efficient path to systemwide success.
- Improving technical competence of clients (+). Clients dislike long-term dependency on change agents. Thus, effective change agents often make education the cornerstone of their efforts, which builds trust and enables clients to make future adoption decisions on their own.
The Legal Productivity Problem
I started Legal Evolution because I believe the legal industry has a very serious problem of lagging legal productivity. This problem is (a) causing ordinary citizens to forgo access to legal advice; (b) fraying relationships between corporate clients and outside counsel; and (c) causing a collapse in demand for law school graduates. See Post 001. From a social welfare perspective, this is a very precarious situation.
Solving the legal productivity problem is going to require the uptake of new innovations. If you want to be an effective change agent, perhaps in the cause of your own innovation, you would benefit from learning the basic principles of diffusion theory and deploying them in an analytically rigorous way.
What’s next? See The Legal Services Innovation Index (021)