Not all concepts are better-illuminated by more analysis, and demands for definitional precision can accidentally carve off important associations. But I thought I’d try to make some headway, and my attempt is now out in a 14,000 word philosophy journal article.i Since that’s written in academic-ese for a different audience who have very different assumptions, and it had to survive a competitive process not optimised for accuracy and understanding, it seemed worth writing a shorter version for people who already have some familiarity with the basic notion and its value.
***
Original conjecture: Slack is the absence of binding constraints on behaviour.
Some immediate puzzles arise with this definition.
First, why ‘binding constraints’? How can a constraint not bind? Isn’t ‘bind’ a synonym for ‘constrain’?
Second, isn’t this basically just ‘freedom’?
Third, if it’s an absence, how can you have more or less of it? Whether I’m 10m outside my house or 1km away, I’m equally absent. I’m further away in the second case, but now we’re talking my distance from the house, not an ‘absence’.
I think the original conjecture is dead right, but we need some apparatus to see why.
***
First, no, not all constraints bind. Constraints foreclose options, but they don’t all meaningfully affect decisions or behaviour because not all options mattered in the first place.
We can distinguish between ‘being under a constraint’ and ‘being meaningfully affected by a constraint’. Note ‘meaningfully affected’ can be interpreted probabilistically – even if adding a constraint doesn’t change your behaviour in the actual sequence, it can be kind of thing that makes it much less likely you’d do that behaviour (say, by making it certain that you wouldn’t X in the likely event that a second constraint was added).
***
Second, slack isn’t just freedom, though there’s overlap. Open a philosophy textbook and you’ll see the following taxonomy.
Negative freedom obtains when there is a lack of constraints. Core intuition: if I’m in a room and you lock the door, I’m unfree to leave the room, even if I choose to stay in the room, and even if I don’t know you’ve locked it.
Negative freedom is what classic liberals and libertarians focus on. Some people think having this is not enough to count as ‘free’ in the sense important to various political disputes; you also need certain internal conditions or processes to obtain, and then you will have ‘positive freedom’. Core intuition: If the door is unlocked but you choose to stay in the room only because you’ve been ‘brainwashed’ (common caricatures likely not real), you’re still unfree. Alternatively: if you’re an addict with a literally irresistible compulsion (common caricatures likely not real) you’re not free to refrain from using. You thus need to the presence of certain rational capacities.
People who think the core intuition captures something right then inflate this and run with it, saying that you’re also ‘unfree’ if you have common political beliefs (‘false consciousness’), or worse, that all these restrictions on your negative freedom that our regime is now imposing are actually making you more free. Unsurprisingly, ‘positive freedom’ is controversial and perhaps shouldn’t be called ‘freedom’ at all.
Then there’s Republican freedom. You have this if you are not dominated. You are dominated if you are subject to the arbitrary will of another. Core intuition: If a slave master sincerely tells his slaves they can do as they please without interference, and then doesn't interfere, but the slaves legally remain slaves, the slaves are still unfree. They may have high negative freedom—there might be few actual constraints on most things they want to do—but since the slave master could change his mind at any time and start immediately imposing constraints, the slaves remain unfree.
The upshot is that freedom does seem to be a function of what constraints are in place, either in actuality or their liability to being imposed. So why does ‘slack’ seem to capture something different?
Here’s why. Consider three ways you can fail to be bound by constraints:
The constraint gets removed (typically increasing your freedom directly).
You gain resources which provide a buffer against the constraint.
You gain options to work around the constraint.
The distinction between (2) and (3) isn’t sharp: resources can create options, and options can get you resources. But there’s clear examples of each. When I gain the ability to trade my skills and labour, I haven’t gained resources, but I have gained options. Alternatively, when I sleep in my house, I don’t do so as an alternative option to sleeping under bridges and getting fined; rather, the house insulates me against needing to even consider the trade-offs of sleeping in different public locations. It means that constraint does not meaningfully affect my behaviour.
Another difference is that freedom, as a matter of emphasis, tends to be concerned with constraints that are socially imposed. In contrast, slack is concerned with all manner of constraints, even those resulting from commitments we’ve voluntarily taken on, which we usually don’t think render us ‘unfree’.ii
***
Answering the third puzzle requires a digression: What’s a hole?
It’s very hard to articulate what holes are. Holes aren’t ‘made’ of anything; indeed, they’re almost necessarily the space where something isn’t. They can move, and grow and shrink, but you can’t capture this by describing the behaviour of the atoms in the hole themselves, you need to talk about the ‘host’: the thing around the hole. Holes also aren’t just a function of how much ‘stuff’ makes up the host, you need to talk about how that stuff is arranged.
So it goes with slack. Slack is indeed an absence. But there can also be more or less of it because this ‘absence’ is determined by the ways in which constraints interact with one another. It’s about what they leave. You can have few constraints and yet little slack because one constraint binds very tightly, acting as a bottleneck that everything else now needs to work around. You can have many constraints and yet plenty of slack because they don’t interact in a way that binds much.
This also explains why—even though we talk about forms of slack and gaining or losing slack—not every increase in resources that count as forms of slack will increase a particular agent’s slack, because it doesn’t help insulate that agent against the constraint that’s binding most strongly, or the constraint that’s our current topic of discussion. Note also that removing a constraint can cause someone to end up with less slack, because that removal then enables and then requires agents to compete by spending their slack.
So, let’s reconsider the conjecture. Slack is indeed the absence of binding constrains on behaviour. This is both precise and non-puzzling when one understands how:
Not all constraints bind, even if all constraints foreclose options.
Slack isn’t just freedom, because freedom is (mostly) about the presence or absence of (socially-imposed) constraints, whereas slack is about what remains after constraints (both socially imposed and otherwise) interact.
Slack does come in degrees, despite also being an absence, because it is a function of how those constraints interact, not simply their sheer number. You can gain more slack via the removal of a constraint, but you can also gain more slack via the accumulation of buffers or options that leave the constraint in place.
The original definition didn’t need refining; only our understanding of the ways that constraints interact and impact us did.
i ZM and SS's names are on the article to reflect rightful attribution of credit for core ideas, not necessarily to reflect complete endorsement this attempted refinement.
ii We sometimes say things like ‘I’m not free to hang out then because I need to attend my daughter’s recital’ but we don’t mean this in the same sense that you’re not 'free' to vote twice in one election or drive on the wrong side of the road. We just mean that there’s a conflicting activity that we’re choosing and we expect this to not be seen as an unreasonable choice.
Slack by Zvi is a much-loved post which captures something important. But the core idea’s contours can be hard to pin down, and people attempting to get a tighter grip haven’t converged on a better characterisation. Slack is the distance to your constraints. No, it’s your spare capacity. No, it’s your room to manoeuvre. No, it’s resources you can use for an advantage.
Not all concepts are better-illuminated by more analysis, and demands for definitional precision can accidentally carve off important associations. But I thought I’d try to make some headway, and my attempt is now out in a 14,000 word philosophy journal article.i Since that’s written in academic-ese for a different audience who have very different assumptions, and it had to survive a competitive process not optimised for accuracy and understanding, it seemed worth writing a shorter version for people who already have some familiarity with the basic notion and its value.
***
Original conjecture: Slack is the absence of binding constraints on behaviour.
Some immediate puzzles arise with this definition.
I think the original conjecture is dead right, but we need some apparatus to see why.
***
First, no, not all constraints bind. Constraints foreclose options, but they don’t all meaningfully affect decisions or behaviour because not all options mattered in the first place.
Consider the quip that the law, in its majestic equality, forbids the rich and poor alike from sleeping under bridges without penalty. The law does constrain the rich from sleeping under bridges without penalty; this option gets foreclosed. But this constraint doesn’t bind them because they weren’t going to do that anyway: they were going to sleep in their houses regardless of whether the law exists or not.
We can distinguish between ‘being under a constraint’ and ‘being meaningfully affected by a constraint’. Note ‘meaningfully affected’ can be interpreted probabilistically – even if adding a constraint doesn’t change your behaviour in the actual sequence, it can be kind of thing that makes it much less likely you’d do that behaviour (say, by making it certain that you wouldn’t X in the likely event that a second constraint was added).
***
Second, slack isn’t just freedom, though there’s overlap. Open a philosophy textbook and you’ll see the following taxonomy.
Negative freedom obtains when there is a lack of constraints. Core intuition: if I’m in a room and you lock the door, I’m unfree to leave the room, even if I choose to stay in the room, and even if I don’t know you’ve locked it.
Negative freedom is what classic liberals and libertarians focus on. Some people think having this is not enough to count as ‘free’ in the sense important to various political disputes; you also need certain internal conditions or processes to obtain, and then you will have ‘positive freedom’. Core intuition: If the door is unlocked but you choose to stay in the room only because you’ve been ‘brainwashed’ (common caricatures likely not real), you’re still unfree. Alternatively: if you’re an addict with a literally irresistible compulsion (common caricatures likely not real) you’re not free to refrain from using. You thus need to the presence of certain rational capacities.
People who think the core intuition captures something right then inflate this and run with it, saying that you’re also ‘unfree’ if you have common political beliefs (‘false consciousness’), or worse, that all these restrictions on your negative freedom that our regime is now imposing are actually making you more free. Unsurprisingly, ‘positive freedom’ is controversial and perhaps shouldn’t be called ‘freedom’ at all.
Then there’s Republican freedom. You have this if you are not dominated. You are dominated if you are subject to the arbitrary will of another. Core intuition: If a slave master sincerely tells his slaves they can do as they please without interference, and then doesn't interfere, but the slaves legally remain slaves, the slaves are still unfree. They may have high negative freedom—there might be few actual constraints on most things they want to do—but since the slave master could change his mind at any time and start immediately imposing constraints, the slaves remain unfree.
The upshot is that freedom does seem to be a function of what constraints are in place, either in actuality or their liability to being imposed. So why does ‘slack’ seem to capture something different?
Here’s why. Consider three ways you can fail to be bound by constraints:
The distinction between (2) and (3) isn’t sharp: resources can create options, and options can get you resources. But there’s clear examples of each. When I gain the ability to trade my skills and labour, I haven’t gained resources, but I have gained options. Alternatively, when I sleep in my house, I don’t do so as an alternative option to sleeping under bridges and getting fined; rather, the house insulates me against needing to even consider the trade-offs of sleeping in different public locations. It means that constraint does not meaningfully affect my behaviour.
Another difference is that freedom, as a matter of emphasis, tends to be concerned with constraints that are socially imposed. In contrast, slack is concerned with all manner of constraints, even those resulting from commitments we’ve voluntarily taken on, which we usually don’t think render us ‘unfree’.ii
***
Answering the third puzzle requires a digression: What’s a hole?
It’s very hard to articulate what holes are. Holes aren’t ‘made’ of anything; indeed, they’re almost necessarily the space where something isn’t. They can move, and grow and shrink, but you can’t capture this by describing the behaviour of the atoms in the hole themselves, you need to talk about the ‘host’: the thing around the hole. Holes also aren’t just a function of how much ‘stuff’ makes up the host, you need to talk about how that stuff is arranged.
So it goes with slack. Slack is indeed an absence. But there can also be more or less of it because this ‘absence’ is determined by the ways in which constraints interact with one another. It’s about what they leave. You can have few constraints and yet little slack because one constraint binds very tightly, acting as a bottleneck that everything else now needs to work around. You can have many constraints and yet plenty of slack because they don’t interact in a way that binds much.
This also explains why—even though we talk about forms of slack and gaining or losing slack—not every increase in resources that count as forms of slack will increase a particular agent’s slack, because it doesn’t help insulate that agent against the constraint that’s binding most strongly, or the constraint that’s our current topic of discussion. Note also that removing a constraint can cause someone to end up with less slack, because that removal then enables and then requires agents to compete by spending their slack.
So, let’s reconsider the conjecture. Slack is indeed the absence of binding constrains on behaviour. This is both precise and non-puzzling when one understands how:
The original definition didn’t need refining; only our understanding of the ways that constraints interact and impact us did.
i ZM and SS's names are on the article to reflect rightful attribution of credit for core ideas, not necessarily to reflect complete endorsement this attempted refinement.
ii We sometimes say things like ‘I’m not free to hang out then because I need to attend my daughter’s recital’ but we don’t mean this in the same sense that you’re not 'free' to vote twice in one election or drive on the wrong side of the road. We just mean that there’s a conflicting activity that we’re choosing and we expect this to not be seen as an unreasonable choice.