More Notes on Simple Rules

post by Zvi · 2019-05-21T14:50:00.305Z · LW · GW · 1 comments

Contents

1 comment

 

Original Post by Robin: Simple Rules

Previously: Simple Rules of Law

Sarah Constantin on Twitter (if you are doomed to be on social media at all, you should follow her) offers a commentary thread of alternate explanations for the pattern pointed out in Robin’s post. It contains some ideas that I didn’t cover and deserve to be addressed, so this post will do that, and capture her arguments in an easier-to-find-in-the-future location. Quotes are from the thread.

1.) Low trust: nobody believes a “fair” rule would actually be applied fairly, they aren’t considering the possibility of a genuinely impartial rule (and sometimes they’re right)

In this scenario, people would prefer rule of law via a simple and fair rule to arbitrary decision making. But when someone proposes such a rule, or tells them they will be bound by such a rule, they do not believe it. They think that this is code for them having to abide by the rule when it suits those in power, only to have those in power ignore the rule when they would rather something else happen. So you need to obey the rules or get blamed for it, and others don’t. Or, alternatively, the rules are an additional thing to be scapegoated for when the political winds call for it, without taking away any of the other methods. If you obey the rules they get you for ignoring what really matters. If you focus on what really matters they get you for not obeying the rules.

This happens often enough that the rules are now just another way to get gas-lit. Following the rules truly becomes a loser’s position, in every sense.

I’d consider this a ‘good’ objection, similar to the Goodhart’s Law objections. Simple rules, and rule of law, require high enough trust and willingness to uphold those rules, or they become another source of complexity and a tool of power. If you don’t have the trust, it must be established slowly over time, or you need to structure things such that this trust emerges.

The word trust is overloaded, so there are things often called ‘trust’ that are directly opposed to the kind of trust that enables simple rules. It would be better if those things used a different word. Words need to mean things and this is one place where the lack of that makes it hard to communicate or even think clearly.

Side note: This is mostly a topic for another day that I really hope isn’t what the comments talk about, but feels worth motioning at here: One answer to the question “What is blockchain?” is that blockchain enables trustless systems. Another perspective is that blockchain is rule of law. You have faith that the system will follow the rules of the system, because humans don’t have a reasonable way to prevent that. The weirdness comes from many of the actors in the system being even less trustworthy than usual because you created a system that lets you work around lack of trust and lacks the  commonly used tools to punish bad behaviors. You’re relying on the rule of law, which depends on choosing good rules and correctly and securely implementing them. Where people choose and correctly implement the right incentives and structures, great things can happen. Where they have other goals or choose poorly… not so much. So you simultaneously get things you can trust a lot, and a lot of lying and fraud along side that, often built directly on top of and relying directly on the trustworthy things.

2.) Ignorance or lack of intelligence: the idea of fair, impartial rules is a bit abstract, and has to be taught, and not everybody gets taught and not everybody copes well with abstraction.

I have a five year old son. Based on my observations of him and other children, I do not think that the idea of fair, impartial rules has to be taught. I think it is a strong instinct that things work this way. Perhaps that’s because my kid is my kid, but from what I can tell younger kids (e.g. ages 5-8) are very big on what the rule is and what is fair. Monkey see, monkey do, and monkey enforce local norms. The kids will still lie and they’ll still break the rules when it suits them, but they totally get it. Then, as they get older, they learn more subtle ways to break and twist the rules for advantage, and play more complex games.

I do think that the idea that following simple rules and having rule of law can get better overall results needs to be taught, trust needs to be learned and maintained and built, and that there’s subtle stuff going on that’s hard to grasp. I mostly want to blame this on bad culture and failure to teach, rather than on lack of intelligence. Most people couldn’t figure this stuff out on their own, which makes it our collective responsibility to give them the tools to get there. Abstraction is by default hard, and we need concrete examples and stories and traditions that resonate – we need culture. The problem is that the powerful, and powerful natural forces, are actively fighting against us, as discussed in the previous post. And as Sarah notes next.

(I sometimes think that if civics isn’t taught in schools people will eventually grow up without actually grokking the idea of “checks on power” being a good thing *independent* of who’s in power.)

Quite so. And of course, most schools do not effectively teach civics. I don’t think most people get that checks on power are good, only checks on particular powerful people and things. The authoritarian instinct runs deep. Those with power would prefer people not learn that.

3.) Power. Often we have a discretionary rather than rule-based system not because *most* people like it that way, but because the *powerful* people like it that way. (as @TheZvi also said.) It’s TurboTax lobbyists, not regular people, who prevent automatic tax filing.

Full agreement here. There are many things most people don’t support. Because most things are not up to most people.

4.) Price discrimination. Often, you can get a better deal if you ask for a favor (or bargain) face to face than if you follow procedure. The average person isn’t overconfidently estimating their charm: they’re *correct* that askers do better than nonaskers on average.

Askers do better than non-askers unless asking is explicitly costly or punished. Otherwise, you sometimes get a yes and profit, and other times you get a no, and break even. And asking is usually much cheaper than it looks. We instinctively get nervous about asking, think it is risky, when usually it isn’t.

Asking more often is a proven winning life strategy.

Price discrimination maximizes profits so everyone would like to engage in price discrimination, but no one wants to be subject to it. Again, we get stories of theft and power. We also get stories of forbidden considerations, and of letting ourselves consider all the data and avoid Goodhart’s Law issues. Asking allows us to say no without explaining our reasons. Rules-based price discrimination seems universally hated, whereas discretionary price discrimination is mostly seen as good. I think this plays strongly into baselines and rewards versus punishments, which we’ll get to at #6.

Asking is more of a method of complexity than it is an explanation for it.

Most people, from what I can tell, strongly dislike having to ask for things and having to haggle and navigate uncertainty. A few people like the advantage they get from being better at it, again a story largely about power and theft.

5.) “Copenhagen interpretation of ethics” = condemnation of intentional but not unintentional harm. This makes some sense as a legal standard, but it’s crazy when you expand it to policy, as many do.

Most people prefer policies with large, harmful unintended consequences over policies which explicitly admit to causing some, smaller harms. This seems like a result of confusing the question of “would this be a good world to live in?” with “should these people be punished?”

The Copenhagen Interpretation of Ethics runs even deeper than that, and was one of my five core stories. Even in this less deep form, it’s still key and highly toxic. Somehow, “spend money that will require higher taxes” or “tax everyone” or the general “start with a lower baseline to reserve resources for special treatment” don’t seem to trigger people’s notions of intentional harm. They should. It’s intentional harm, much more directly than many things that are objected to as intentional harm.

6.) There’s a weird thing where justice/rationality/impersonal principle is coded as “mean” while making exceptions is coded as “nice.” A “judgmental” person is one who makes *harsh* judgments — even though judgments can be good as well as bad.

This may just be loss aversion or pessimistic bias: the fear of being punished for our failings is more salient than the hope of being rewarded for our merits.

Emphasis mine. I don’t think this is loss aversion.

When people say someone is judgmental, or is judging, they’re usually talking only about negative judgment. Rarely about positive judgment. They are the same and imply each other, but people don’t see it that way. The mean thing is bad. The nice thing is good.

Which makes is super important to code actions as nice rather than mean. Actions are already on thin ice. Any system of action, whether by rule or by discretion, needs to do its best to be seen as taking nice-coded action and avoiding mean-coded action. There’s also the practical problem that confiscating things is not typically something one can simply do, whereas bestowing them is allowed.

If you’re dividing resources, that means you want to set the inaction baseline distribution as low as possible. Then you can make ‘exceptions’ to the baseline, and pay out rewards, be seen as nice, and enjoy the power from selecting the distribution of the resources you withheld or confiscated.

Screwing over the baseline scenario is not merely an incidental effect. It is a goal. It is necessary.

That (give or take a comment thread) should wrap things up. Ideally these thoughts can then be distilled into posts that are easier to make evergreen so we can build upon them better.

 

 

1 comments

Comments sorted by top scores.