At the root of effective altruism is a concern with how best to bring about social change. However, effective altruism discussions rarely distinguish overtly between two different kinds of change: marginal and systemic. It's really important we make this distinction – because it has the potential to profoundly influence some of the choices that we face, and to lead to very different effects.

What is marginal change?

By 'marginal', I don't mean 'minimal'. I'm using the word in the economic sense to describe a change that is incremental relative to some other possible change. What does that mean exactly? Say you're about to graduate from university and you have two possible career options. One is in publishing and the other is working for a development charity. Your one criterion for job choice is the amount of positive social impact you can have.

You decide that you will likely have more impact as a charity worker than as a publisher and you take that option accordingly. Your social impact is higher than it was when you were a student – that's a marginal increase. And it's higher than it would have been if you'd made a different choice – to go into publishing. The difference in social impact between these positions is a marginal difference.

A lot of effective altruism discussion revolves around examples like this. Most are to do with three different kinds of marginal change, measured against potential social impact. First, change relative to something else you could be doing. Should you change career from A to B or stay where you are? What would the marginal change be in your likely social impact potential?

The second kind is change relative to what someone else might do. I'll borrow an example from 80,000 hours: should you go to work in your own country or volunteer as a doctor in a developing country? Assuming a high demand for such positions (so someone else will take the position if you don't), the marginal difference you make is equal to how much better you are than whoever else would have taken the job. Maybe you save two more lives a week, for example.

Questions about where you should donate fall into a third category, where you're asking about the difference in the potential social impact of two third parties. Should you give your money to organisation A or B? What you're asking is: what is the marginal difference between the impact A would have (by using your money) and the difference B would have?

The main feature of marginal change is that it is commensurable. That is, you can assess the difference between the change represented by two possible choices using some common unit – a unit that in this case represents social impact. You might decide to use a unit like quality adjusted life years (QALYs), for example. So, comparing two possible jobs, or two possible donation recipients, you try to enumerate the possible QALYs you will be able to add to the world with each course of action, and you choose accordingly.

What about systemic change?

Systemic change doesn't feature in these sorts of examples – and in general is less prominent in effective altruism discussions (it's not absent entirely, but when it is discussed it can often be implicit). So, what is it and why does it matter?

Systemic change is hard to define precisely, but we can start by suggesting that it's not just large amounts of marginal change. To borrow another example from 80,000 hours, let's say you convince several hundred high net-worth individuals to contribute large amounts to specific, highly effective charitable organisations. The amount of change you could bring about is potentially enormous – but I would argue that it's still marginal change.

Why? Because this is basically a quantitative approach. You're not questioning your overall goal of getting as much money to these organisations as you possibly can. Nor are you questioning who those organisations are. In other words, you're not challenging the system parameters. More on that in a moment, but first we need to define what the 'system' is that systemic change transforms.

It's important to remember that although 'system' sounds big, it doesn't have to be. A system can be small or large, just as marginal change can be small or large. Here's one definition of system: a group of actors with related interests who interact with one another in pursuit of certain goals and end up producing a measurable outcome.

That's pretty abstract, so here's an example to illustrate.

Let's take the 'donating to charity' system. We can think of this as just one part of the overall effective altruism system (another property of systems: they can be part of other, larger systems). In the 'donating to charity' system we have a few groups of actors – amongst others there are donors, recipients and analysts.

The donors want to donate to the most effective charity, the analysts want to work out which organisations are the most effective, and the recipients want funding to pursue their particular objectives. From an effective altruism perspective, this system's outcome is money directed towards particular causes.

A system like this offers opportunities for both marginal and systemic change. Imagine that you want to focus on getting money to effective organisations. You've got some choices as to how best to do that. Some of those involve marginal change, and others are systemic.

Let's say you go down the marginal route and concentrate on donating. Your choices are things like: do you become an investment banker and donate a set percentage of your income, or do you become a persuader, trying to convince other people to donate their income to charity? Or do you try to combine both? If so, how? Those choices accept the basic metric (amount donated) and the system parameters (which organisations are, broadly, best to donate to; the actors involved, etc.). In other words, they involve working within the system.

But what if you're more inclined towards a systemic approach? Now your choices involve challenging some parameter of the system. You could become an analyst and work on methodologies for measuring charitable effectiveness. By altering which organisations are assessed as particularly effective you change a key element of the system – those organisations that interested donors will target most. (Note how analysts can change the system while also forming a part of it).

Alternatively, another systemic choice could be to work towards introducing a mandatory charitable donation tax on everyone who earns above a certain amount. You're not challenging the metric used (amount donated) but you're creating a new method of affecting it, and so affecting other actors' choices. If that kind of tax were introduced, it might well mean that you're less likely to go round trying to persuade wealthy people to donate more. It also has the potential to create new choices – for example, you now have the option of forming pressure groups to try to lower the income threshold at which that tax kicks in.

That kind of radical transformation, challenging metrics and choices within a specific sphere (however small or large that may be), is what systemic change is all about.

Why might we want it?

Systemic change isn't a panacea. The discussion I've given here has been tilted towards the idea of systemic change out-performing marginal change. But that's not necessarily true, for three reasons.

First, systemic change might be smaller in scale than marginal change. Maybe changing the organisations assessed as effective is less important than persuading some of the world's wealthiest individuals to donate billions to a slightly less effective set of causes.

Second, systemic change is context-dependent and its effects may be difficult to foresee. What if a mandatory charitable donation tax actually reduced absolute levels of donation to a given set of causes by increasing complacency or even hostility amongst the wealthiest donors?

Third, systemic change can require a greater investment of resources to bring about. A tax reform such as the one I’ve described here would be politically extremely difficult to achieve. Devoting that amount of time, expertise and resources to an alternative marginal option might result in far more significant results.

On the other hand, systemic change does have the potential to introduce powerful transformations. It's a question of leverage – and changing system parameters can undoubtedly be a very powerful tool.

So why does this matter?

We need to be aware that these two types of change are quite distinct. Depending on the context, they may be more or less appropriate and have different degrees of potential influence. If you find yourself in a situation where there is limited potential for further marginal change it might be that systemic change offers a more promising set of choices.

Marginal and systemic change also require different skillsets and attitudes.

Analysing systems for leverage points (points at which systemic change can be introduced) and possessing the technical know-how to operate them emphasises certain skills over others. Introducing tax reform is obviously quite different to persuading individuals to donate. And appreciating that your efforts to find such leverage points might be in vain requires a certain risk/reward mindset. Pursuing your tax reform would likely require a substantial investment of resources, and even then may never happen – or, worse, could be counter-productive.

What's important overall is to recognise that marginal and systemic change represent distinct approaches to tackling social problems and producing positive social impact. When tackling any particular problem we should bear both in mind as possible options, consciously choosing whether to work with an existing system or challenge its parameters, taking into account our personal strengths and the relevant context. Ultimately, that choice could make all the difference.