UC Berkeley, School of Information
December 15, 2020
Also available as: pdf
This is a section of a chapter on findings from a published dissertation: Enacting Privacy in Internet Standards.
Individuals who participate in the process are the fundamental unit of the IETF organization and the IETF’s work. The IETF has found that the process works best when focused around people, rather than around organizations, companies, governments or interest groups. That is not to say that these other entities are uninteresting - but they are not what constitutes the IETF.
– A Mission Statement for the IETF (Alvestrand 2004)
The procedural principle of individual affiliation is frequently cited and discussed by participants in IETF standard-setting, as well as in its documentation and by its leadership. Individual affiliation takes on a sort of mythical status: everyone knows and talks about it, and knows that it isn’t quite true, but also that it has some weight and history behind it. According to my interviewees with substantial experience at IETF, it’s a “narrative” or a “story” or a “theory”:
there is the theory that all people at the IETF are participating in their individual capacity and are not representing their employer. I say that is a theory because of course in reality most people there are acting consistently with the interests of their employer, but, I mean, especially in the early days of the IETF kind of back into the ’80s and ’90s [..] engineers who were there […] their employers were forward-looking enough to assign them to essentially go contribute to the IETF and to not be heavily pursuing a corporate agenda.
Participants note examples (sometimes very specific, sometimes general) where an employee will take a different position from their employer’s direct interest, both at IETF and W3C, or where they themselves handle the tension of aligned but not identical goals for their work. These separations are attributed to a few factors:
Sometimes you see someone who does something that’s clearly not in their employer’s interest and you just think, “wow, that’s amazing.” And sometimes there are really serious experts who are distinguished enough that they have a lot of freedom in their job, or they work in research or a lab or something and they’re not constricted by the fact that they already have features built into a product.
That was [Company’s] goal. Yeah. It’s not my goal. My goal was to not break the Internet. It’s most of what I do, is not break the Internet.
Employees at times have this latitude because of an interest from their employer in supporting standards development work: sometimes that’s because the company has a particular product or business goal dependent on improving standards in an area, but also it can be because a company wants “insight into what was going on” (or similarly “active awareness”) or “visibility” by having their employee in a prominent role with “active involvement.”1 There are significant similarities here to the boundary organization collaboration of open source software foundations (O'Mahony and Bechky 2008): where companies can fund employees to work on “areas of convergent interest” in an open source project, while allowing that project to maintain its own practices.2
The examples above of individual participation share a sense that an individual’s role in the larger Internet community has, sometimes, a priority over their role as an employee or representative. Long-time standard setting participants note that this experience can influence or moderate how an individual behaves because of their longer-term, cross-company interest.
Most people don’t stick to a single company forever. And so, that colors behavior. That works as long as you are within what is typically called a community or within an industry in the broadest sense.
Engagement in the collaborative process of developing the infrastructure that is the Internet requires or at least benefits from this long-term and larger-scale commitment. These situations in technical standard-setting seem especially relevant for engineering professionals, rather than others in business, product development or legal teams, where financial considerations or the client’s position are more primary to the job. And it coincides with technical standard-setting organizations being an arena where engineers are often given latitude to take public positions without requiring more extensive sign-off from the rest of their organization.
The closest analog that I heard outside of engineering was from ad industry participants who sometimes refer to the future direction of the industry as a whole. In particular, this seemed to be attributed to leadership, in the sense of senior executives of large companies or leadership of industry trade associations, rather than a perspective of typical participants or employees of member companies.
And so you can cut through sort of, you know, junior-level perceptions of what they feel their goal is, and by moving it to that board level, they truly do care about the ecosystem, right? They want all businesses to flourish.
A focus on ecosystem or platform or more than a single company or interest can be compatible with this sense of forward-looking planning and more collaborative behavior. In the case of high-level firm leadership or technical seniority, it might in both cases be motivated by the possibility of moving between firms and organizations and having an interest in positive relationships and reputation. Consider the trends described in Regional Advantage (Saxenian 1996) and the Silicon Valley culture of employees easily moving, and cross-pollinating ideas, between competing chipmakers.
However, a more cynical explanation for an individual’s divergence from an organization’s interest comes up in the context of Do Not Track:
there’s a class of kind of advocates, mostly on the industry side, probably entirely on the industry side, at least from my point of view, who have a vested interest in gumming things up and in friction when it comes to efficient government action, and, honestly, a lot of times they’re not doing it, I think, even in the best interests of their clients or particular companies. They’re doing it because it makes them more important. It raises their profile, makes them more essential. It makes them the kind of main roadblock, which makes them the main broker that you have to pay attention to, and of course it leads into billable hours. And so I feel like there’s this class of D.C. advocate, D.C. lobbyist who fits that bill, and I think I felt like the Do Not Track room was full of enough of those people that it derailed a lot. I know I’m making a lot of assumptions here that probably other people would disagree with, but that was kind of my take.
That is, while we might identify cases where an individual will diverge from an employer’s direct interests because of a community interest, there may also be cases where an individual may diverge in order to benefit themselves directly, in terms of power or financial interest. In that case, divergences can lead to more obstructive behavior, relative to what the client might prefer, or relative to what the broader community may need.
And in the context of negotiations over Do Not Track, there are also objections to these divergences or a preference or respect for direct representation of an employer’s or client’s interests. Respect for representing a client or a position is tied to the idea of being a good faith and principled participant who believed what they were saying or advocated for a genuine interest, even when the speaker disagrees with the identified person.
two or three people in the Do Not Track process, who were on the other side of me on every issue, who were not terribly pleasant to work with, but at the same time I think were basing their opinions strongly on ideals or strongly on what they thought one particular company wanted or needed
We almost never agreed substantively on the issues, and [he] might skewer me about half a dozen things in the press or on Capitol Hill. [He] and I have testified on the same panels in front of Congress. […] We had a job to do. We both believed in what we were doing, and I think we both did a really good job.
These are positive evaluations of behavior not just because someone’s statement might be reasonable or principled, but because they were advocating for a company’s or an organization’s interest and that was their “job to do.”
(See also: good faith vs bad faith and participant antagonism, as detailed in the earlier section on process.)
Under the perspective that employees should represent their employers faithfully, there are also cases where a lack of internal coordination can create a situation where a representative doesn’t know what other decisions may have been made internal to a firm. This can cause a change in position in a standard-setting negotiation or even conflicting statements from individuals in different teams who are employed by the same company. While this doesn’t get described in the same way as bad faith, it can lead to practical frustration.
People in the room have views different than their companies. […] the people who were representing Microsoft at DNT did not know that Microsoft was about to change its position [re: default settings].
Or, describing inconsistent positions and a lack of shared information from experience (this quote refers to employees from a different tech company):
he was pretty down on the lack of specificity of the API and I was kind of like, you work with the person who edited this thing, right, really? […] honestly, what are you doing here? Can you guys go in the corner and strategize for a little bit before you get up and say this? […] Your coworker over there who’s actually building the product thinks the exact opposite of you.
This phenomenon is common in standard-setting, though it’s ironic to hear Microsoft as one example, as the informal reputation I had seen built up was that Microsoft was more likely to discuss and agree on something internally before taking a position at a standards body, where Google was more likely to have different teams that didn’t coordinate about their products and employees might take opposite positions at the standards body and resolve them after the fact. People have preferences in either direction – where the lack of internal coordination can be frustrating, it can also be appraised for speed or transparency – but organizational cultures differ on how much coordination happens internally and separately from the more public standard-setting process.
Some have positioned themselves as fundamentally opposed to companies that do not sufficiently coordinate employee actions. Regarding potential changes to cookie functionality in Firefox, the Interactive Advertising Bureau CEO Randall Rothenberg described Mozilla in a trade press interview (Ebbert 2013):3
Mozilla is obviously a very factionalized organization. It’s like mob rule. It’s very difficult if you’re a rational player. […] It’s not really clear if there is a Mozilla itself, other than the radical players who seem to have the ability to control what does or does not go into the browser.
This “mob rule” reference is perhaps frustration regarding the model of open source software development or the ability of non-employee developers (here called “radical players”) to contribute code to the Firefox browser. Perhaps analogous is similar frustration expressed (again, with caveats, in the press (DePillis 2013), from an advertising industry lawyer) about the input of developers and developer associations in a NTIA-convened multistakeholder process about mobile app transparency: “Developers are the people we hire to do software coding. That’s like saying the painter of a retail store makes decisions about the paint.”
That engineers or developers may contribute to decision-making about software, and may have the ability to do so without the hierarchical approval more common in other industries, may contribute to some of the surprise or mismatch of expectations here. As described in The Ethics of Engineering, previously, engineering is inherently ethically-laden and there is an impulse towards integration of larger concerns into the practice of engineering. If engineering is inherently about deciding on and bringing into being the good life, then the choices of engineering will be significant in a way that isn’t simply the direction of their employer.
In addition to a concern over individuals not acting in concert with their organizations, some participants seem to consider the equivalent status of individuals as an affront, or impractical, because of its informality or the lack of appropriate representation.
And so you had somebody there representing a 100-billion-dollar industry and you’ve got Jonathan Mayer who was what, 22 at the time? And very smart. But the idea that he was going to show you, it was just … it was really – there were so many dynamic issues at play.
It can at times, as in this case, be difficult for participants to put directly into words this mismatch feeling. It may also be related to discontent (this same participant calls it “absurd”) with the informality of procedural steps like a “hum” to gather a sense of the room.4
Lower barriers to participation may be preferable for the idea of access, a potential boon for legitimacy. But low barriers may also prove frustrating because of who may choose to participate.
one thing that’s very odd, is who’s on the call and who’s not on the call, right? So, if an individual in some country decides to show up every week, they get to speak. And there’s a sort of formal equality to their participation with the participation of people who’ve invested a billion dollars in a particular thing. Or, the same as one of the public interest groups that’s invested years and years of effort in the space, and then some loud-mouth, who just like feels like showing up, gets the same formal role in the process as the dedicated group that’s well-staffed and very thoughtful. So, one way to say it is, you don’t have to have skin in the game to be on those calls
“skin in the game” is an especially evocative description of this challenge, as it provides a similar metaphor to “stakeholder,” that some commitment, investment or ownership is a reason itself for legitimacy in participation. Individuals may not have the same ability to demonstrate that weight that someone affiliated with an organization can. This is a sharp contrast to the legitimacy concerns raised about implementers (and especially market leaders) having too great a power in an interoperability-focused consensus process.
One view of governance is that of balancing the interests of stakeholders: what makes a decision legitimate and valuable is the positions of important sub-groups that have a significant stake in the outcome. Distinct, and often quite different, is a process where legitimacy and value comes from efficient and effective analysis of arguments’ validity by the key experts in the area – we might call this the technocratic view.5 Internet standard-setting has typically taken a part of both approaches through its interoperability focus: implementer weight is especially important in determining acceptable outcomes but insight is valued for finding and evaluating technical feasibility.6
That individuals participate in a process rather than organizations is on one level obvious: it’s people who are in the room or on the calls who are talking and debating. Companies and organizations don’t take actions like that, although it’s very common to anthropomorphize them in our language, perhaps based on a mental model of firm hierarchical decision-making.7 That individuals are affected by their affiliation with organizations, influenced by corporate priorities even if distinct in their own goals and how they interpret them is also clear. What’s of interest, though, is how those individuals within (or perhaps mediated by) organizations build relationships, collaborate, compete and come to decisions in a consensus standard-setting process.
The representational view of individuals in a multistakeholder process minimizes their personal perspectives, expertise or interests: the role of a representative is to accurately represent the positions she has received from the represented group and to argue most effectively for the represented group’s interests. And for the stakeholder-balancing view of multistakeholder process and its legitimacy, representation is a natural fit for what’s expected from each participant.
But a collaborative view of individuals in a multistakeholder process maximizes their expertise in a problem-solving orientation, where they have a high-level goal based on (or at least aligned with) their organization’s priorities and have autonomy to use their own perspective to navigate towards a practical outcome. For a technocratic view of processes where quality of argument is key, encouraging individuals to collaborate with their own expertise seems most fitting.
I argue that the Internet standard-setting process attempts to accommodate both the stakeholder-balancing and the technocratic view of the process and both the representational and collaborative views of participation. There is audible frustration from anyone expecting purity of either perspective. A representationalist will be angry that an individual without sufficient stake needs to be addressed; a collaborativist will be disappointed that crass concerns about someone’s business model interfere with a more rational or beneficial outcome. But the potential benefits of using consensus multistakeholder technical standard-setting to make progress on tech policy challenges rely on both the pragmatic feasibility of finding outcomes acceptable to key stakeholders and the potential innovation of problem solving among a heterogeneous group of people with varied backgrounds and expertise.
It is possible that to the extent that the architecture and implementation of the Internet has been a success for liberal – in the sense of Postel, but also perhaps Mill – outcomes, it is the combination of those views of a process for individuals that has contributed to its success.
There is an imperfect but apparently substantial alignment between organization-centered and individual-centered perspectives on process. Consider the following categories.
| organization-centered | individual-centered | |
|---|---|---|
| unit of participation | organization | individual |
| scope of work | policy | technology |
| task | decision-making | implementation |
| purpose of a process | balancing of stakeholder interests | technocratic |
| form of process | negotiation | problem-solving |
| role for individual | representation | collaboration |
Not all disagreements or conflicts described in the qualitative analysis above line up with these categories. For example, some might agree with the individual’s role as representational, but disagree about who should be represented (herself, one’s employer, the larger cross-organizational community to which one belongs). But the merging of these two categories does explain some regular confusion. Some object to a technical standard-setting body as the wrong venue for making policy decisions, preferring legislative or administrative governments for that purpose; others object to the engineers being involved in decision-making rather than sticking to implementation.8 The tension of impulses towards separating engineering from ethics or policy and the impulse towards more directly integrating it reflects this distinction in the particular field of engineering.9
One approach is to explicitly recognize the difference between these categories and that they’re both relevant and then try to divide them up. P3P designers tried to make that explicit with vocabulary vs policy choices, and the Tracking Protection Working Group also divided deliverables into compliance and preference expression syntax. Some involved in the DNT process and some that I spoke with would have advocated to push that separation further, separating technology and policy and enabling a tussle over which compliance policy would be chosen or accepted.
What if, instead, we recognized not only the differences but also the increasing blurring of these boundaries and embraced that merging? We might, as in the case of DNT, see processes where stakeholders discussed and learned technical details, business practices and policy implications in a combined setting and we might see more people with combined technical and policy expertise, as discussed in the following section on patterns of participation.
These quotes in passing are from two different interviewees.↩︎
This connection has been detailed previously in Doty and Mulligan (2013), and there are particular connections here since there is overlap between those open source projects and standard-setting organizations.↩︎
This quote is not from a research interview and is not similar in style to a research interview discussion; incendiary statements of this kind are part of active campaigning, and not uncommon from Rothenberg.↩︎
Mentioned in brief in a popular press piece about the Do Not Track process: “‘Do Not Track’ Web System Stuck In Limbo” (2012).↩︎
Sunstein describes the tension, which applies to all public law but especially administrative law, as between the technocratic and democratic view, of expertise and accountability to the public will (2014).↩︎
That it might be possible to bridge these procedural and substantive views of legitimacy, or both rationality of argument and the material conditions of the world in making decisions, is not novel. For example, Habermas’ view of discourse ethics maintains that under the right conditions of autonomy and speech, procedural requirements can guarantee just outcomes, and Froomkin argues that IETF standard-setting is a rare (unique!) example of this idealized practical discourse within a particular community (2003). This work is not focused on testing this hypothesis, although there is plenty of evidence here about strategic action to counter that argument if others are interested in it. Instead, I am simply trying to explain the different approaches that people have to a multistakeholder process and the sources of the conflicts for those who do not accept a combined view and how they respond to individuals and their unclear relationships to larger stakeholder organizations.↩︎
See also the concept of “institutional synecdoche,” as described previously in Internet Standard-Setting and Multistakeholder Governance.↩︎
There are also contrasting views of engineering as a profession and the relationship of the individual engineer to her employer. Following that representational view, the engineer works on behalf of the client, implementing to their exact specifications and setting aside any of her own value judgments for the needs of the client (which might be the employer, or the end user). Following the collaborative view, the engineer considers not only the needs of the client, but also her own values and her insights into the design and how it will interact with others.↩︎
Again, see, previously, The Ethics of Engineering.↩︎