At a co-design workshop I facilitated in Cape Town in 2023, I posed a question to the room. The attendees were all scientists and researchers, and they had all, without hesitation, said that they wanted more diversity. More diversity in their networks, their publications, their convenings. So I asked them: would you be open to non-PhD holders being in the room? To community health workers? Would you be open to village elders, to other forms of knowledge that do not arrive with institutional credentials?

It gave them pause.

That pause is where I want to start, because I think it is the most honest thing that happened in that room. These were people of good faith who genuinely believed they wanted diversity. And they did, up to a point. The limit became visible the moment I named what diversity might actually require.

What are we really asking for when we say we want diversity? Are we asking for different bodies, or are we asking for different knowledge? Are we asking for people who look different but think within the same disciplinary frameworks, carry the same credentials, and ask the same kinds of questions? Or are we prepared for the genuine disruption that comes when you open the room to people whose ways of knowing do not map neatly onto peer-reviewed categories? If we do not interrogate these questions before we start building, the diversity we invite into the room becomes tokenism. People of color brought in to represent a community they are no longer quite speaking for. Community health workers invited to validate a program already designed. The room looks different. Nothing changes.

That workshop became one of the founding moments for what eventually grew into CSIDNet, the global network for climate-sensitive infectious disease research. The question I had pushed the room to sit with was the same question I then had to sit with myself as I began co-designing the community. Wellcome Trust had identified a need: a genuine community of practice capable of shifting the geopolitics of who produces and who uses CSID knowledge, with the breadth of voice that ambition actually requires. More diversity. Broader representation. All the right language. But what would we actually build for that?

If we anchored membership in the “CSID researcher” identity, we would reproduce the same dynamic I had watched play out in Cape Town. The researcher identity carries institutional weight and concentrates in predictable geographies: it skews Global North, skews academia, skews toward people already inside the networks that precede the network you are trying to build. So rather than anchoring the community there, we introduced a third framing alongside researchers and tech practitioners: the end user. Not a formal membership category, but a design lens, a way of asking before we had convened anything who this knowledge was ultimately supposed to serve and whether those people would recognize themselves when they found us.

Community health workers. Practitioners running programs in high-burden settings. Nonprofits at the intersection of climate and public health. Policymakers whose work depends on CSID research but who would never describe themselves as producing it. The end user framing opened the door to all of them, and because it did, they were in the room when we began co-designing what CSIDNet would actually become.

That last part is the structural argument. Who you bring in to help build the community is who the community gets built for. The people who feel like they shaped something are the people who feel like it belongs to them. This is not a philosophical point; its consequences are very difficult to correct after the fact, because founding members shape the norms, the language, and the implicit expectations of participation in their own image, and that culture compounds in ways a diversity initiative added later simply cannot undo.

The architecture we built includes eight volunteer-led committees and around 80 core members holding active governance roles on rotating mandates. Those structures were not designed primarily to deliver programs. They were designed to give people genuine stake in how the community runs, which is a different thing. A volunteer sitting on a committee is making decisions, not doing courtesy labor, and they carry organizational memory and relational trust forward in ways that a paid secretariat of three and a half people cannot.

This is also a sustainability argument, though perhaps not the one funders usually want to hear. Wellcome’s catalytic investment will not fund CSIDNet indefinitely, and a community whose only carriers of institutional knowledge are the staff does not survive the grant cycle. But eighty people with real relationships across disciplines and geographies, who feel genuine ownership because they helped build the thing, are a different kind of foundation. Exit to community was in the design from the beginning, not because we were pessimistic about the funding, but because a community you cannot gracefully exit was never really a community at all.

None of this begins with a diversity statement. It begins with a pause. It begins with the willingness to ask not just whether you want diversity, but what the limits of the diversity you are actually prepared for are, and then to build the structural conditions that answer takes you to. The Cape Town room went quiet when I asked about village elders. I think that silence is more productive than most diversity frameworks I have encountered, because it is at least honest about where the real work starts.