David R. MacIver's Blog

Some thoughts on ethics and narrative

This post was originally published at https://drmaciver.substack.com/p/some-thoughts-on-ethics-and-narrative.

Hi everyone,

As a result of Wednesday’s newsletter about thought experiments, I’ve been thinking about the role of narrative in our lives and our sense of ethics. I don’t have a single cohesive piece for you, but I’m going to share some thoughts on the link.

Interpreting the world

One of the things I found myself thinking about in response to Wednesday’s newsletter is that one way to look at this is that you construct weird thought experiments by fuzzing your claimed universal ethical principles by throwing random examples at them and then doing a simplifying process to produce a clear and comprehensible demonstration of the problem. Of course, just throwing random examples at it won’t necessarily work, so you need to do some sort of guidance of the process by paying attention to what parts of the ethical principles are exercised and guide your construction of the thought experiment onto novel paths and…

Look, sorry, I’ve thought about software testing a lot over the last few years, and it’s very much gotten into my brain. Sometimes this is even explicit in my life advice. e.g. Life as an anytime algorithm. Sometimes the analogies are very good, but they’re mostly very good if you’re the sort of person who has thought an awful lot about both feelings and software testing (i.e. me).

This is normal though. Everything we do is a rich source of analogies and comparisons, and the things we think about shape our ability to think about other things.

And a lot of what we think about is fiction and narrative. We learn about the world from stories, and we analogise our lives to the things that happen in stories. As a result, the way we make sense of our lives is often very story shaped.

This is, of course, a very Ravenclaw way of looking at it.

Fiction as ethical training

At its best, Fiction serves as the kind of grounded hypothetical history I talked about as the good kind of thought experiment on Wednesday. It provides us with ethical scenarios that we can examine and learn from, to a much greater degree than real life can - we experience fiction at an accelerated rate compared to the events it depicts, and we have enough distance from the events that we can learn from them.

This does require us to form our own judgement about the scenarios presented rather than taking the authors’ viewpoint on the subject as canon of course. Also, how well it works depends greatly on how close the ethics we need to get better at is to the scenarios in question, and in what ways. Sometimes we need fiction with characters who are ethically better than us (in realistic ways), sometimes we need fiction with characters whose failings we find relatable.

I suspect very rarely do we need fiction where characters are grappling with life and death ethical decisions, which tend to be quite far removed from our day to day ethical considerations.

Different distributions of events

One of the reasons why I was thinking about the analogy between fiction and software testing is that it is actually legitimately good and describes a real problem that people experience: Efforts to automatically find bugs often produce bugs that do not matter.

One example of this problem that I ran into recently is this discussion about bug finding efforts on the Z3 Prover. There are a variety of things going on here and if you’re not interested in the specifics of software testing I wouldn’t worry about it too much, but the core of the complaint is that some people were bombarding the issue tracker with bugs that their software was finding in Z3, and these bugs were typically of the form “If you do these very specific weird things that nobody would do, something a bit bad happens”.

The complaint is that these bugs slow down development on Z3 significantly without improving it in real world situations, because they force attention to events that do not really come up in practice, diverting effort to making the software better in real cases.

But one thing that happens here is that this notion of “real cases” is very fragile. I often think about John Regehr’s “Operant Conditioning by Software Bugs”. The sort of real cases you get with an experienced user is very different from the sort of real cases you get with a novice, because the latter hasn’t learned to create the same cases as the former. Similarly, when you employ software with very nonstandard use cases you often run into bugs that nobody has ever seen before.

I think our ethics runs into a very similar problem: When you try to use a set of ethics in the context for which it is developed, it will work very well. When you try to use it outside that context, you will run into a really surprisingly large number of problems.

Fiction as ethics porn

A concept that has thoroughly established itself in my brain as soon as I encountered it is the idea of “generic porn” from Nguyen and Williams’s paper Moral Outrage Porn. They suggest that:

A representation is used as generic porn when it is engaged with for the sake of a gratifying reaction, freed from the usual costs and consequences of engaging with the represented content.

e.g. actual porn is a representation of sex, engaged with for the sake of an ahem gratifying reaction, freed from the usual costs and consequences of sex.

Since encountering this definition a lot of ways of engaging with the world seem like porn to me, and I think a lot of our fiction in particular is a sort of ethics porn. A representation of people trying to live a good, ethical, life, engaged with for gratification, freed from the usual costs and consequences of engaging in a good, ethical, life.

I suspect often the lessons we learn from this are about as reliable as those we can learn from actual porn.

When the world feels off kilter

I think one thing we might experience as a result of this ethics learned through narrative is that the world perpetually feels subtly wrong. There’s an experience I’ve talked about before (in On feeling blocked) about switching between two very similar games, Celeste and Hollow Knight. The games were close enough to feel very similar, but different enough that it was impossible to play one like the other. It’s almost worse to be highly competent at the wrong thing than it is to be incompetent - it’s much easier to get better starting from incompetence.

The worst part of this is that everything feels subtly wrong. You know how the world should work, and it doesn’t work that way.

I suspect trying to navigate life with an ethics learned on fictional narratives is like this, because nothing quite works as it “should”. The difference between real life and narrative is that narrative is supposed to make sense. Events get satisfying resolutions, plot threads tie together, there are consistent themes. Unfortunately, reality gives you none of these things.

You can try to construct narratives to give life a sense of meaning, and this is a thing I have seen various people try to do, but I think ultimately we have to let go of the idea that you can run your life on narrative. It’s a powerful tool, one we should not give up on, but I don’t think it can be the foundation of an ethics that fits the world as it is.