The Civil Society Game

Why Rational Organisations Keep Producing Irrational Systems

South Africa’s civil society sector is often accused of being fragmented, territorial and too competitive. Fair enough. But put an organisation into a system where funding is scarce, relationships are currency, attribution determines who gets the next grant, collaboration carries risk and survival is never quite guaranteed, and then ask what a rational organisation would do. It will protect relationships, guard information, compete for attention and cooperate selectively. Not because the people running it are particularly selfish, but because they have learned the game. The uncomfortable possibility is that much of what we call dysfunction in civil society is actually perfectly rational behaviour, and that our social investment architecture repeatedly produces precisely the behaviour it then complains about.

This is where game theory becomes useful. Not because civil society needs an economics lecture, but because it forces us to look at behaviour without first turning it into a moral judgement. Game theory asks what actors are likely to do when their outcomes depend not only on their own choices, but on the choices of everyone around them. Civil society is full of these situations.

An NGO deciding whether to share a donor relationship, a funder deciding whether to back something genuinely new or something it already knows, two organisations deciding whether to pitch together or separately, a coalition deciding who carries the work and who gets the credit. Even communities are making calculations about whether another consultation is worth their time.

None of these decisions happens in a vacuum. Everyone is reading the room, remembering what happened last time and trying not to be the person who carries the risk while someone else walks away with the advantage.

Take five organisations working in the same field. All five know that better coordination could reduce duplication. They could share data, refer opportunities, build common infrastructure or stop commissioning five variations of the same research. Collectively, this makes obvious sense. Individually, the calculation is more complicated.

If I share a donor relationship and you do not, you have gained something while I have surrendered something. If I open my programme model and you package it better than I do, I may have strengthened the sector and weakened my own organisation. If we jointly pitch for funding but your brand ends up owning the relationship, collaboration may have produced a better proposal while making my own pipeline more fragile.

So organisations hedge. They collaborate where the risk is manageable and protect where it is not. They share enough to remain useful and hold back enough to remain viable. There is nothing particularly surprising about this. It is exactly what rational actors tend to do in conditions of uncertainty.

The problem appears when everybody makes the sensible decision at the same time.

That is how we end up with multiple organisations building parallel databases, running separate monitoring systems, approaching the same funders and producing different reports on roughly the same problem. Nobody sat in a room and decided that duplication was a good idea. It emerged from a series of individually reasonable choices.

Then, having designed a system in which individual organisations are expected to survive independently, we stand back and complain that the sector is fragmented.

The funder is not outside this game either. Funders operate with incomplete information. They do not know with certainty which organisation will deliver, which impact claims are inflated, which leadership team is stronger than its proposal, or which ambitious intervention will unravel once implementation begins. So they look for signals: track record, governance, known leadership, evidence, existing partners, strong reporting, scale, visibility. Again, nothing irrational about this.

The interesting part is what happens next. Organisations learn which signals attract money. If scale is rewarded, everyone becomes scalable. If innovation is rewarded, familiar work is repackaged as innovation. If partnership is rewarded, proposals gather partners. If professional reporting signals competence, scarce unrestricted money gets spent on increasingly polished reports.

The system teaches actors what successful behaviour looks like, and actors adapt.

This is not necessarily dishonesty. More often, it is learning. Every sector has its codes, and civil society has become fluent in its own. The risk comes when the signal begins to separate from the substance. A metric that started as evidence becomes a performance requirement. A partnership that started as cooperation becomes a collection of logos. A strategy document that was meant to guide decisions becomes something produced because credible organisations are expected to have one.

The same logic applies to communities, although the sector often treats them as though they sit outside the game. They do not. Communities learn too. They know which organisations arrive with energy and disappear eighteen months later. They know when a consultation can actually influence something and when the real decisions have already been made. They know what kinds of stories attract attention and which promises tend to survive beyond the launch.

If participation repeatedly produces very little, declining to participate becomes a perfectly sensible response. If telling your story generates another report but not much else, eventually you become less interested in telling it. What an institution describes as disengagement may simply be experience.

This is why I am increasingly unconvinced that civil society has a collaboration problem in the way we usually describe it. It has an incentive problem.

We keep treating behaviour as though it were primarily a question of organisational character. NGOs must share more, think longer-term, break down silos and put collective outcomes ahead of institutional interest. All reasonable ambitions. But institutions cannot run indefinitely on good intentions.

If data creates competitive advantage, expect organisations to guard it.

If donor relationships affect survival, expect those relationships to be protected.

If attribution influences future funding, expect people to fight for attribution.

If the grant ends in March, it is difficult to persuade an organisation to behave as though April is somebody else’s problem.

That does not make the behaviour desirable. It simply means we should understand where it comes from.

We have a habit in development of moralising behaviour that is actually structural, and then trying to correct it with another workshop. A collaboration workshop will not overcome an incentive structure that makes collaboration risky. A call for systems thinking will not create long-term behaviour if all the money remains short-term. Telling people to share will not make sharing rational if the cost of doing so sits with one actor while the benefit flows to another.

This is where the conversation becomes more interesting, because the question changes.

Instead of asking why organisations behave as they do, we can ask what would make better behaviour rational.

That is a mechanism-design question.

If we want organisations to refer work to one another, referral needs to create value rather than simply hand an opportunity away. If we want common data, the shared system has to be more useful than the private spreadsheet. If we want common infrastructure, funding architecture cannot keep rewarding duplication. If we want specialisation, every funding call cannot require every organisation to do everything. If we want collective outcomes, then our attribution models cannot continue pretending that complex social results belong neatly to one institution.

None of this is easy, but it moves us into much more useful territory. Instead of asking people to behave against their own interests for the sake of the system, we start asking whether the system can be designed so that institutional interest and collective interest are less frequently at odds.

South Africa has another complication. The actors are not only playing different strategies, they are playing on different clocks.

A corporate may be working against an annual budget, ESG commitments or B-BBEE requirements. A foundation may have a three-year strategy. Government works through financial years, political terms and administrative cycles. An NGO may be thinking about whether salaries are covered six months from now. A community may have been living with the same structural problem for twenty years.

All of these actors can sit around the same table, agree on the same problem and still make completely different decisions. They are solving for different things over different time horizons.

We tend to interpret this personally. Government lacks urgency. Corporates are short-term. NGOs are defensive. Communities have lost trust. Sometimes those descriptions are true, but they are not always especially useful.

A better question is what each actor stands to gain, lose or carry, and over what period they are making that calculation.

Once you start looking at the system this way, a lot of supposedly irrational behaviour becomes easier to understand.

For years, civil society strengthening has focused heavily on improving individual organisations: better governance, better strategy, stronger boards, better data, stronger fundraising, better leadership. There is value in all of that. But I am not sure the organisation itself is still the most interesting unit of intervention.

The more interesting work may sit between organisations.

Who carries risk. Who owns the data. Who controls the relationship. Who gets credit. Who gets paid. Who benefits when another actor succeeds. Who loses something by sharing. Who can afford to wait and who cannot.

Those are not soft questions about partnership. They are the architecture of the system.

And that may be where the next generation of social investment needs to focus more deliberately.

Not on persuading rational organisations to behave irrationally for the common good, but on designing a system in which the rational choice produces a better collective result.

This is how we #GrowZA