Jun 19, 2015

You should read this

Nate Soares has been writing an interesting series of posts, which begins by arguing against feelings of listless, nihilistic guilt that are frequently associated with young people in the West. The series then shifts towards arguing against feelings of specific, pointed guilt that are associated with a certain subset of young people in the West, a similar subset to the one that is attracted to the Effective Altruist (EA) community.

This argument against feelings of specific, pointed guilt concerns me.
A while back, Luke Muehlhauser identified a distinction between EAs who are motivated by an opportunity to good and EAs who are motivated by an obligation to help those who need it. I suspect that the topic of this post touches heavily on this distinction, and that my concerns only really apply to people who are motivated primarily by obligation. I don't personally identify as an effective altruist, but I spend enough time thinking about this stuff to serve as a functional proxy here.
I have had discussions about the topic with a couple of people in the EA community, and have come away troubled each time.

I parse the argument as positing that the concept of normativity is harmful. A plainer way of stating this would be: "should's and ought to's are bad, and we would be better off without them."

My purpose here is to lay out why I find this concerning, then propose a couple of hypotheses about what I think is going on with the argument. I'll start with a little groundwork, for framing and context.

Groundwork

I draw a strong distinction between things I want to do and things I ought to do. I usually call the things I want to do "desires", and I call the things I ought to do "obligations." I don't think these terms can be easily defined precisely, so in lieu of precise definitions, here are lists of things which I put into each category:

Things I class as "desires":

  • Getting hungry and wanting to eat a filling meal
  • Wanting to eat ice cream, regardless of how hungry I am
  • Becoming aroused and wanting to have sex
  • Wanting to exercise
  • Wanting to watch funny sitcoms
  • Wanting to read interesting books
  • Wanting to have a successful career
  • Wanting to have a family
  • Wanting to travel
  • Wanting to be well-traveled, or worldly
  • Wanting to be the sort of person that helps other people

Things I class as "obligations":

  • Buying presents for friends and family during the holidays
  • Staying in touch with extended family
  • Refraining from assaulting, abusing, or murdering other people
  • Paying taxes, and following laws
  • Helping people who are suffering and close to me (emotionally, genetically, spatially, temporally)
  • Helping people who are suffering and far from me (same qualifiers)

All of the "desires" I listed are self-centered, focused on the agent. All the "obligations" I listed are contractual, focused on satisfying the requirements of agreements the agent holds with others. I'm not going to speculate further on these observations.

The takeaway is that I when I talk about my desires, I'm roughly speaking about things I want; and when I talk about my obligations, I'm roughly speaking about things I have to do.

Why a lack of "should's" is concerning

In this post, the "no-should's" argument is stated plainly:

...the pattern is the same: the subject thinks there's something they should be doing (or some way they should be), and they're not doing it (or aren't being it), and so they feel really guilty.

I claim that the word "should" is causing damage here.

In fact, as far as I can tell, the way that most people use the word "should," most of the time, is harmful. People seem to use it to put themselves in direct and unnecessary conflict with themselves.

In most of these posts, the argument against "should" is presented in a specific, context-dependent way. The given advice is to unpack the "should" statement into a more meaningful "if-then" statement. For example:

"I should call my father this week" might cash out to "if I don't call my father this week, he'll feel disappointed and lonely."

I don't have an issue with this advice. Making "should" statements into "if-then" statements has the happy result of making the consequences of actions explicit. This is probably useful for most decision-makers, and very useful for decision-making in a consequentialist framework. And it's not problematic, because once you've unpacked all your "should" statements, you will be better positioned to judged which of the "if-then" statements is best, and then you can do the thing that is best.

But Nate proceeds to make a stronger claim – that this procedure of unpacking your statements, then weighing your options, can be applied to more significant decisions. From this post:

I've seen many people use the word "should" to highlight a conflict between what they perceive as desires and what they perceive as moral obligations. For example, they might say "well I want to buy this ice cream, but I should donate the money to the Against Malaria Foundation instead."

I say, this is a false conflict. Imagine this person precommiting to never doing anything just because they "should." How might they feel?

They might feel relieved, because they actually didn't care about helping others, not even a little bit. So they discharge their guilt, buy their ice cream, and go on their merry way.

But more likely (in someone who thought they "should" give to AMF), that would feel a little bad, and a little hollow. This person, when committing to never do things because they "should," might feel a bit of fear. They might worry that if they didn't keep themselves in check then they'd never do anything to help those less fortunate than themselves. That might seem bad, to them.

Which lets them actually see the true problem, for the first time: they both want to buy the ice cream and help those who are worse off than them. Now they can actually weigh both desires on the scales, or search for clever third options that fulfill both desires, and so on.

Deciding whether to buy ice cream or bednets is certainly a specific, context-dependent decision. But it has a lot more philosophical heft behind it, so the implications of following the "unpack-your-should's" procedure grow more severe.

My basic issue with this framework is that it sounds relativist and egoist overtones, two concepts that rub me the wrong way.

One way of stating my objection is that doing away with "should's" gives desires primacy over behaviors. Thinking hard about what you want, then doing what you most want is probably a great procedure for desire-fulfillment maximization, but not necessarily a good procedure for ethical conduct. It does seem empirically true that some people have a strong desire to help other people, thus fulfilling this desire results in ethical conduct and good outcomes. However, not everyone have desires so happily aligned. In the extreme case, it seems like someone who followed this procedure, and on reflection had a strong desire to murder innocents, would then go on killing spree to satisfy their blood thirst. If desires direct behavior, there is not an obvious way in which this behavior is "wrong."

Extreme cases like this are non-controversial – no one argues that it is okay for mass murderers to mass murder, even if that is what they genuinely desire to do. Cases like this highlight a premise implicit in the "get rid of the should's" argument. The premise looks something like this:

If you introspect carefully, you will find that your strongest desires align with what most people call "ethical conduct."

This is probably true for most people, most of the time. However, for some people, most of their desires seem to point in very unethical directions. And for most people, some desires occasionally point in unethical directions. I see the useful work of ethics to be constraining these edge cases – the sociopaths, and the sociopathic tendencies within all of us. Normativity – telling us what we ought to do, and what we ought not do – seems to be the best tool in the toolbox for this work. So I'm reluctant to define away the best tool we have; without normativity, ethics becomes a method of desire clarification and maximization. That is not a pursuit I'm excited about.

Now, onto some educated guesses at what is going on with this argument and my disagreement with it.

Hypothesis 1: Language Games

Most people I've talked to about this with in the EA community don't seem to be asserting a strong relativist position, nor do they seem eager to endorse the actions of serial killers. However, they are arguing against the use of "should" language. Because everyone having the conversation is some sort of objectivist who dislikes serial killers, I'm guessing that the argument has become confused somewhere.

My first guess about this confusion is that our conversations have been language games played poorly, wherein we talk past each other a fair bit. Here are some guesses about what that could look like:

Blown out of proportion

It's possible that I've overgeneralized the "get rid of should's" argument. I'm interpreting it as a general philosophical statement – the argument's proponents might be saying something more specific, like: "if you introspect about what you want, and like most people, discover that your desires align with what we call 'ethical conduct', then you would be better off not using an obligation framework. Instead, just do what you most want to do! But if upon introspection, you discover that your desires are unsavory, follow a different rule system. Don't act on bad desires."

This position seems plausible to me, though I'm not sure it's entirely consistent. If the "no-should's" argument is actually saying something this specific (i.e. "if you are the sort of person who has a strong desire to help people and also feels guilty about not helping enough, then stop feeling guilty!"), I would have to think about it more.

However, based on the conversations I've had, I'm pretty sure that a stronger claim is being made (i.e. "the concept of normativity is harmful and we'd be better off without it"). It's this stronger claim that I'm reacting against, and I've spilled a lot of pixels in vain if nobody is asserting that.

Positives and negatives

I might have overgeneralized the "no-should's" argument in another way: perhaps only "action-inducing should's" (positives) are to be done away with, whereas "action-preventing should's" (negatives) can be kept in place. For example, there could still be normative, deontological constraints like "you should not kill people" and "you should not steal things," but these constraints would be all negative – things you couldn't do. Positive deontological mandates, like "you should give most of your money to effective charities" would be disallowed. Desires would dictate positive actions.

I'm pretty convinced that this is not what is being proposed, though I'm not entirely sure. However, it's not a consistent position – the division between positivity and negativity is too fluid (for example, the negative constraint "you should not neglect your family" easily morphs into "you should spend time with your family," a positive mandate, and it's not clear which of these is the true form of the concept).

Redefinition

The categories of "desires" and "obligations" are fluid as well – the argument's proponents and I could be talking past each other by reclassifying the concepts proposed by the other party. For example, I say "I feel a strong obligation to help those in need", which they then redefine as "One of your strong desires is to help those in need," both of which basically point towards the same concept. This could be happening repeatedly, allowing for a persistent, insoluble argument.

This might have been happening in the conversations I've had about this issue. If so, we've been caught in some pretty impressive loops.

Now, on to my second hypothesis...

Hypothesis 2: Running straight up a mountain isn't the best way to scale it

My hypothesis-1 speculations are all about the nature of the discussion around this topic. I have a second hypothesis, which deals with the substance of the topic:

It could be the case that there are some things which we are obligated to do, and that the best way to accomplish these things is by not feeling any obligation to do them. This would mean that obligation is a meaningful class of concepts, yet feeling obligated is not an effective way to fulfill one's obligations.

This might not be true in the general case, but it is seems very likely to be true for some subset of people. With this restriction, the statement would look like:

There are some things that we are obligated to do, and for a certain type of person, the best way to accomplish these things is by not feeling any obligation to do them.

When we apply this hypothesis to the case in question, we get something like:

There are some things that we are obligated to do. It's plausible that one of these obligations is to help improve the lives of others. If helping others is one of our obligations, we can say that we should help others.

However, for some people, feeling as though they should help others is not an effective way to help others. For these people, feeling this obligation will likely result in demotivation, guilt, and self-loathing. Therefore, it would be better for these people to find alternative motivators, like desire, to explain their actions.

I like this hypothesis, because it address my concern, and I think it addresses most of what the "no-should's" argument is reacting against as well. It's a little subtle though – in simple language, I'm saying something like "hey, you don't like feeling like you have to do things? But you still feel compelled to help people? Well then, try not to feel that way! I bet you'll still want to help people."

This seems quite close to the "no-should's" argument, which in simple language says "hey, you don't like feeling like you have to do things? Well, I've got good news: you don't have to do anything! Just think about what you really want, then do that!" But that goes a step too far – as I discussed above, it legitimizes too many desires that seem clearly unethical. The step too far is taken with the implicit assumption that, for the most part, people's desires align with what we would consider "ethical behavior." Empirically, this is probably true most of the time, but it fails frequently enough to be disqualified as a building block for a moral theory.

So, closing restatement – "obligations" exist as a meaningful class of concept (at least as meaningful as the class of "desires"). This means that there exist things that we ought to do. If the existence of obligations bothers you, yet you still find yourself drawn to doing good actions, don't worry over it too much. You're probably better off not feeling obligated.

[rereads: 3, edits: fixed typos, reworked some phrases, tightened up language, rewrote the second half of the "blown out of proportion" section]