Finding the Bell Curve of Meaning

A process for supporting the emergence of shared language in broad collaborative communities

by Kaliya Young & Margo Johnson

Introduction

The identity community focused on user-centric ID gathers twice a year at the Internet Identity Workshop (IIW), an event at the crossroads of more than a dozen open standards working groups focused on various aspects of digital identity.

At IIW29 in October of 2019, a session was called to assess the creation of a glossary to support the convergence of meaning around key terms that were being used in relationship to the emerging decentralized identity and self-sovereign identity technologies. The small group convened there was led by Kaliya Young (Identity Woman), Margo Johnson (Transmute Industries), and Drummond Reed (Evernym), and included a variety of perspectives from other contributors like Adrian Gropper (HIE of One) and Juan Caballero (Spherity).

We considered several sources ahead of beginning the process towards glossary creation. We looked at the past history of glossaries in the community, including the Identity Gang Lexicon from 2005, which was created in the very early days of the community that would come together over the following fifteen years (Kaliya and Drummond were active at that time). There was the Sovrin Glossary(2016; 2019; 2019), which in its current form contains 248 terms as a potential source we could draw terms from. We also spoke with Daniel Hardman about his approach to glossary work, where one defines key concepts with a functional description but no name, and then when all the concepts are outlined collectively, names are decided all at once upon firm concepts.

We also drew on the work of Eugene Kim, who focuses on high performance collaboration, one of the key elements for which is the social processes of developing “Shared Understanding” and “Shared Language” are social processes:

• Shared Understanding is when two people or groups use different words for the same things or the same words for different things, but because of enough social contact, they can actually understand what is being communicated.
• Shared Language emerges if you have sufficient contact between groups/people such that their word choice becomes decidedly aligned.

We were mindful of this when we chose the process for this glossary work. And we wanted it to be inclusive, so we listened carefully to the current diversity of language that was already being used throughout the community. The meaning of key words wasn’t ours to fix or define; we aimed to establish a composite of the digital identity vernacular as we encountered it.

Community members urged us to create a glossary, pointing to confusion “in the market” about what key terms meant. Ultimately, we learned that this request represented a variety of impulses, which spawned some confusion about our efforts. It was clear from the beginning that the work should not seek to define terms at the level where they are being used within technical working groups where significant time was being devoted to defining key terms locally within a given technical specification. The energy of the group got behind better understanding how words are used in every day language and conversation in the marketing level documents.

As such, we aimed to invite alignment around key words, to find the “bell curve of meaning,” and to invite convergence towards the middle of this curve as a result of our research process. Our hope was to maximize the ability to inform and engage business audiences interested in adopting/purchasing this technology. This meant we would not be setting out to define hundreds of terms; rather, we chose to focus on a subset of the most widely used terms.

Methodology

First, our group created and circulated a survey throughout the broader digital identity community to identify key terms to focus on. The group chairs designed the survey by brainstorming a starting list of terms (based on working knowledge of the space), and respondents also had the option to add their own. Respondents ranked the provided terms from highest to lowest priority. The chairs reviewed these results and chose three of the top-ranked terms as a starting point for the glossary process.

In the case of this research those terms were “agent”, “wallet”, and “credential.”

One other term — “self sovereign identity” — was actually rated as the most important by this process. However, the group ultimately decided that this hot-button term would be best approached after a more foundational understanding of supporting terms was in place.

Having defined three targets for building clarity and consensus, we then created and shared a second survey asking respondents about their working descriptions for each of the three terms, as well as how those terms relate to each other. We worked from an assumption that understanding the relationships between the terms would help us better understand their meaning. Several additional questions about the company business model were also included to provide context about respondents. This survey was distributed via multiple mailing lists to a broad swath of community groups working in decentralized identity, verifiable credential, blockchain, and data privacy space.

Analysis

After several weeks, the second survey had twenty seven responses from various companies, and our group was able to start with community data analysis efforts. The chairs took some time to consider different analysis options, prioritizing those that were digital, inclusive, participatory, fast, and free. We wanted to minimize any barrier to community members engaging in the process. Finally, we designed a framework in Google Sheets whereby qualitative responses to each question were inductively sorted into emergent groups that we named and modified in real-time, helping us understand common themes as well as areas of divergence. You can picture this as a digital version of hundreds of post-its all over a wall, being moved around by different people until they eventually fall into groups, and those groups are given a big post-it name after an iterative process of redistribution and renaming. This is a creative combination of human-centered design practices such as affinity diagramming, design synthesis, and insight combinations.

Using this analytical framework, the glossary group met weekly for an hour over several months to read through the survey responses together and create groupings. Additional participants rotated in and out throughout the process, offering different perspectives and helping to refine the approach. The approach was messy at times, with responses falling into multiple groups or defying categorization altogether. However we persisted, and noted interesting discussions and reflections that emerged in the process. Once we finished grouping all the responses, we combed through the data once more, cleaning it up a bit, merging groups where there was high overlap and clarifying the final description for each group.

Upon completing the grouping process we counted the number of responses in each group across each term and relationship type to determine which concepts came up the most frequently, potentially indicating the highest level of alignment across respondents. We also looked at the number of groups per term/relationship as an indication of convergence or divergence in responses.

Looking at this qualitative and quantitative information, the group took the final step of writing working descriptions for each of the three key terms. This involved looking at the 2–4 groups for each term with the highest number of related responses and constructing a phrase from their group descriptions. This was a surprisingly smooth process for the review team — in particular for the terms “wallet” and “credential” — where there was a higher level of consensus. One notable challenge we faced was with the term “agent” based on the large number of groups included, and while we were able to construct a more complex phrase, the difficulty arriving at that point became a finding in its own right. Interestingly, while the term agent had the highest level of divergence and outlier definitions, it also had one group — “representative of the subject” — with the highest number of relevant responses in the whole survey at 21. This could be interpreted as meaning that the term agent actually has a high level of convergence around this core representation property, and then individual implementations diverge in other agent features.

Aggregated Descriptions (Agent, Wallet, Credentials)

Credentials: Credentials provide structured standards for accessing identity data

Wallets: Provide storage of keys, credentials, and secrets, often facilitated or controlled by an agent.

Agent:

SIMPLE: Dictionary definition: “a person or thing that takes an active role or produces a specified effect”

NUANCED: An agent is a software representative of a subject (most often a person) that controls access to a wallet and other storage, can live in different locations on a network (cloud vs. local), and can facilitate or perform messaging or interactions with other subjects.

For the relationships between the terms, we found that the simplest way to share the results was to create a diagram showing the relationships (see below) and to list out all of the group titles there. This helped to show the diversity and multi-directional nature of those relationships.

Sharing Results

Our group built a simple, publicly visible presentation to share details of the process, results, and next steps. In this deck, we emphasized descriptive rather than prescriptive results, noting that this was one way of better understanding our shared landscape and taking actions together to build market understanding and product language.

We presented this deck and answered community questions within the same community groups where the original work was spawned,, as well as running a session at the 30th Internet Identity Workshop (full notes here). The feedback was generally positive.

Reflecting on the process, what remains to be seen is how the community will utilize these terms, and whether we will see a further convergence towards shared meaning as a result of this work. We have already seen this work referenced in discussions when shared vocabulary issues are raised, and we know that several publicly traded companies active in the industry are also sharing this internally. We are also preparing to follow up and share the results with respondents to complete the feedback cycle.

Conclusion and Next Steps

From an information-sharing perspective, the group hopes to continue to share this work out into related communities, and encourage further conversations and actions. We particularly want to encourage conversations at a market level, connecting standards to businesses today.

Moving forward, the glossary group hopes to do another round of definitions using the same process. We also want to continue to explore specific impacts of this work to gauge its continued value for the broader community.

We hope that other communities can learn from our experience, and use this process or iterations of it to support the emergence of shared language in their communities. From a research process perspective, we would love to see other groups leverage this methodology, see how it works for them, and offer process and marketing improvements that can benefit everyone. If you do pick up this method and try it, please let us know how it works for you.

If you are interested in joining the community glossary group please sign up here: Fill out the form here.