I left off my last post considering when you might want to pursue systemic change rather than marginal change. In this post I'm going to take a closer look at that.

A system's key elements are: a group of actors (with some interests in common), their interactions, and the outcomes that result from those interactions.

A single actor can be part of more than one system at once, and systems can change over time. They can also vary in size.

Consider the 'donating to charity' system ( the example I discussed in my last post). Its actors include donors, analysts, and aid recipients, and the main outcome of the system we're interested in is how much money gets to the recipients.

A marginal approach to change in that system would simply try to increase the amount of money donated (perhaps by taking a job as an investment banker and earning to give).

A systemic approach to change here might try to suggest that the money should be going to a different set of recipients – maybe by arguing for a new metric against which to assess charitable effectiveness.

If successful, this systemic approach would change a parameter of the system (who the recipients are), which is a potentially powerful way of changing the system's outcomes.

But when should we pursue marginal change and when systemic?

In my last post I suggested that there were at least two factors we would need to bear in mind when making that choice: expertise and risk/reward mindset.

Choosing marginal change or systemic change requires expertise appropriate to the task. If you want to earn to give, you need to have the expertise required of, say, an investment banker. If you want to argue for different charitable aid recipients, you need to have the expertise to analyse charitable effectiveness and present your arguments in a compelling manner.

As to mindset, marginal change can be more predictable and controllable. The reward is readily calculable and the risk is comparatively low. Systemic change, because of the complexity of even very small systems, is much riskier – i.e. far more likely to fail. However, it is possible that the reward (relative to the effort put in) may be greater.

For example, it may take only one report on charitable effectiveness to channel a significant proportion of donations away from some organisations and towards others – that may well be far more money than one person could expect to earn to give over the course of their working life. (This is an idealised example, of course.)

What about qualities of the system itself?

It's not just about expertise and risk/reward mindset. Those terms describe the characteristics I think are important for us to possess if we want to pursue systemic change. But we also need to consider the conditions of the system.

A system can be more or less amenable to change, and we ought to take that into account when choosing whether or not to pursue such change.

What makes a system amenable to change?

A key aspect is complexity. The more complex the system, the less likely it is that we can predict the effects of our actions on it, which makes identifying suitable levers of change extremely difficult.

Generally, the larger the system (so the more actors and/or interactions it has) the more complex it will be.

In terms of what a small group of individuals or organisations can hope to achieve, large complex systems probably aren't worth bothering with – the complexity is simply too great.[1]

On the other hand, a system needs to be dense enough to be worth trying to influence. In other words, there need to be enough actors with close interactions. If we're talking about a few actors who interact only quite loosely and occasionally, or with little mutual influence, then what we have is barely a system at all. Trying to exert influence in that kind of an environment is like pulling a lever that isn't properly attached to anything – the effect just won't be very big.

So assuming we have a system that is just the right size – neither too complex nor too ill-defined – we need to have enough knowledge of that system to identify the levers of change we want to use. Gathering that information ourselves might not be possible since it could require substantial resources, so it may be that we have to wait for someone else to gain that understanding.

We also need to be able to access the levers of change that we identify. In many systems a majority of those levers may belong to national or local government. That means we need to be able to convince others to press those levers for us.

That in turn means that politics becomes a factor. Is the general political environment likely to be in favour of the change(s) we're seeking? Do we have any influence with the rights parts of the government? Can we work with someone who does? (This may mean that a political career could be advantageous to pursuing systemic change. See 80000 Hours for further thoughts on the ethical effectiveness of a political career.)

An example

To illustrate the above, I'm going to return briefly to the topic of my own research – human trafficking in the UK.

Human trafficking is a poorly understood problem. There is little robust empirical data on the scope or extent of the problem anywhere, and while the UK has better data than many other countries it's still not much to go on.

However, understanding has been increasing over the past ten years or so.

The anti-trafficking 'system' in the UK consists of traffickers, victims, law enforcement, victim support NGOs and various branches of the Government. They produce several outcomes that we're interested in, such as criminals convicted, and victims rescued.

Ten to fifteen years ago it could be said that this system barely existed. There was little awareness of trafficking amongst law enforcement, immigration and social services, and there were fewer active NGOs than there are today.

That's not to say that trafficking didn't exist – it did. But there wasn’t a system that we could hope to influence.

Since then, the UK's trafficking response has developed substantially – such that we can now call it a system. There are enough actors involved with a sufficient density of connections and interactions.

As that system has developed, many NGOs and other actors have identified weaknesses within it that they thought needed to be changed in order to help victims and obtain more convictions of traffickers. In other words, they identified what they thought of as likely levers for systemic change.

However, until relatively recently, few of those changes were successfully introduced.

That changed somewhat in 2013, when a think-tank called the Centre for Social Justice released a report ('It Happens Here: Equipping the United Kingdom to fight modern slavery') on the UK's anti-trafficking response. That report led almost directly to the introduction of the Modern Slavery Bill in Parliament – a Bill that has now become law.

The Modern Slavery Act introduces several new measures that we can see as operating levers of systemic change. A good example is the clause mandating that companies above a certain size report on efforts they are taking (or not) to combat modern slavery in their supply chains. By producing new information that may influence how large numbers of actors conduct themselves (e.g. consumers in their buying choices), the way the system works may change.

Why did this Bill get introduced and become law? Obviously the reasons are complex but I think we can see it as a relatively successful example of systemic change that illustrates the importance of the factors I discussed above.

First, the anti-trafficking system has, over the last few years, reached the right size – cohesive enough to identify as such but not too large that its complexity could not be grasped.

Second, sufficient knowledge of that system existed. Several major NGO reports had thoroughly mapped it over the previous few years and the CSJ report did so again – working on the basis of over 200 interviews with relevant actors. That enabled the CSJ to confidently identify the levers of change.

Third, the general political environment was fortuitous. Another Bill had been thrown out, leaving Parliamentary time available, and the issue of trafficking (or 'modern slavery' as it has come to be termed) is bipartisan enough that a Bill had a chance of succeeding in the limited time available.

Fourth, the CSJ had good connections with the Government and so were able to communicate their findings effectively.

So that's alright then?

A caveat or two may be in order.

The vagaries of the political process have meant that although the Modern Slavery Act has on the whole been welcomed as a success, it has not addressed all the points that the CSJ report – and many other actors – would have liked it to. Several potential levers of change remain undisturbed.

Further, although the CSJ's report, building on other work, constructed the best map of the UK trafficking system we have yet had, it was of course not perfect. Our knowledge will continue to improve and as it does it may well change what we perceive to be the relevant levers of change.

Finally, statistical data (for example on number of victims rescued and numbers of traffickers convicted) remain poor, which means that it will be difficult to evaluate the success of the new law.

Any conclusions?

I think we can cautiously see this as a good example of successfully engineered systemic change, albeit with the caveats noted above.

It suggests that the conditions under which systemic change is an appropriate choice are quite limiting, however. The choice to pursue such change is therefore one that we should take only after careful consideration.