Call of an Optimist

Rules are important. They tell people how to behave, what to expect of others, and where the line is.

So it’s pretty important that people can actually understand your rules.

Most rule-makers start out by attempting something fairly straightforward, only to then find that there are so many edge cases. Content moderators, lawmakers, corporate policymakers, anyone trying to define acceptable behavior ends up on this path. And when they do start to glimpse the many edge cases they failed to handle, they all decide “well, I guess we’ve got to add nuance to the rules to accommodate this.”

I think they are wrong. I think they should bite the bullet, keep the rules simple, and either accept some edge cases or solve them another way. Not because this results in perfect rules, but because the costs triggered by complicated rules are MUCH greater than the costs of those missed edge cases.

Simple doesn’t mean stupid, though. With the right basic framework, it’s possible to build rules which everyone can understand and follow, which leave a lot less room for bad edge cases than you might think.

Start with distinguishing two types of rules: Laws and Policies.

Laws are rules which must be followed. Violations are not permitted, so if someone appears to have violated a law, their only recourse is to argue that their actions didn’t really violate the law, based on the law’s specific wording or intent.

Policies are rules which should be followed. Violations are discouraged, but if someone appears to have violated a policy, they do have a recourse – they must simply justify that the violation was in society’s best interests.

Laws are appealing, because they feel simple and definitive. But the problem is that they encourage people to not just nitpick the individual wording, but to expand the definitions over time. A Law that says “don’t murder”, when confronted with questions like “wasn’t that self-defense, though?”, is forced to erase the word “murder” and replace it with “kill anyone intentionally unless it’s a necessary act of self-defense.” Then someone else says “what’s necessary, though”, and before you know it, your ‘simple’ Law takes 500 pages to fully write down because of all the nuance.

The reason you are stuck in this trap is that, by writing a Law, you’ve created a binary. Following the Law is always good, disobeying is always bad – therefore the exact wording of the Law is the highest authority.

That might sound good, but most Lawmakers don’t actually want to shackle themselves to a quirk of their past wording. And they are correct – if you find that you wrote a Law poorly, it’s a terrible thing to continue following it!

In practice, most things people treat as Laws would be better as Policies. A Policy saying “Don’t murder” doesn’t need to define exactly what “murder” means. Any time it’s ambiguous whether someone committed a murder, you don’t ask them to debate the meaning of M-U-R-D-E-R, you ask them “was the action you took beneficial for society?” If no, then whether or not it was murder, it was bad. If yes, then by definition society is happy they took it.

Many of you readers might be rolling your eyes here. Yes, you might say, obviously it’s better to approve things which are good for society and disapprove things that are bad for society. The point of Laws, you would argue, is that most people can’t tell which is which, and so the Laws are there to tell people in advance whether they can do something. To do otherwise would leave everyone in constant uncertainty about whether what they were doing is permissible!

To which I respond…good. That’s a feature, not a bug. If you’re a member of any society or community or whatever, you should be constantly asking yourself if what you’re doing is benefiting the group. If you’re not willing to spend the effort actually thinking for yourself whether your actions are benefiting others, you probably don’t deserve whatever those others are giving to you. And if you’re so stymied by uncertainty that you’re unwilling to take an action, well, there are LOADS of ways to get a sense whether your community would respond well to an action aside from forcing them into an irrevocable promise based on exact wording. If the only way you can get approval is through a hyper-specific contract with precise, careful details spelled out across so many pages…consider WHY it’s so hard to get approval any other way, and whether that says something about whether you should in fact be doing what you’re doing.

Some of you still won’t be satisfied. You’ll argue that the real problem isn’t just certainty or fear – it’s the cost of adjudication. To know if someone broke a Law, consult the facts of the case and the written Law, and the answer just falls out. To know if someone should be punished for violating a Policy, you need not just the facts and the text of the Policy, but also the ability to model different future states based on whether they did/didn’t take the action, to understand whether or not it was actually net beneficial. That’s way harder to do, which means adjudication will cost more and back up the court system forever, won’t it?

Here my answer is admittedly a bit more speculative, though I do believe it firmly – I think Policies are actually, in aggregate, cheaper to adjudicate than Laws, for three reasons:

1. Laws do in fact need complex adjudication too – it just is part of the up-front legislation process. But that’s a HUGE bottleneck to all kinds of things we want to build rules around, so I’m counting it as a “cost” paid on the Law side of things, which naturally makes the Policy side look better in comparison.
2. In the world of Laws, people read the exact wording, see a loophole, and then feel empowered to act on it. This results in some people getting away with doing things we didn’t want them to do (which is another “cost” on the Law side of things), and sometimes results in people with more money/weight literally redefining the meaning of words (since words aren’t actually exact no matter how hard we try), which results in both unfair treatment and ‘drift’ of the meaning of Laws over time.
3. Finally, I predict that fewer people will do questionable things in a world of Policies. In the world of Laws, if you believe you see a true loophole, then in some sense you feel safe to act even if you know that most people will disapprove. In the world of Policies, it’s harder to believe you’ll get away with doing something obviously against the society’s interest. Some people will still violate the Policies in the hopes that they don’t get caught, but far fewer people will violate Policies assuming they can talk themselves out of it than the number of people who navigate around our Laws this way.

Laws make sense when you absolutely, definitively know exactly what you’re talking about, and know that real-world circumstances will not evolve any faster than you can update the Law. This is a really rare situation, so Laws rarely make sense. And writing a Law and then treating it like a Policy is the worst of all worlds – now you’re directly breaking a promise (that you’ll enforce the Law consistently) that was made implicitly when you chose to write the rule as a Law.

The more uncertain the future is, the more you’re trying to be robust to corruption or loopholes, and the more you want to empower creativity – the more you should rely on Policies.