On rationalist fiction

(Originally posted here)

Let’s ignore for a moment the association between ‘rationalist fiction’ and HPMOR, LessWrong, etc. Let’s ignore the cultural associations here. Let’s ignore that most existing rationalist fiction by word count is HPMOR and most existing rationalist fiction by title count is MLP fanfic. 

Because the techniques and guidelines given for rationalist fiction are a toolset not uniquely tied to the cultural and political atmosphere that gave rise to them, and they can be used in a different context.

One problem with popular fiction – a problem that negatively impacts representation and a problem that Storming the Ivory Tower & others have mentioned in a critical sense – is that characters make dumb or out of character decisions in order to avoid interfering with existing ideological constraints in the plot. This is a killer of meaningful representation – because non-white-male characters are the first to be compromised for dramatic effect, being typically supporting characters – and it’s a killer of immersion.

Let’s look at the toolset, and look at how these tools can be used to prevent flaws in characterization.

1) Level-1 intelligent characters consistently behave in ways that are realistic for people with interiority. In other words, they act in their own self-interest unless they have abiding reasons not to. (This can be taken too far, in the sense that you can easily drop realistic character attributes that might cause real people to act against their own self-interest, like infatuation or various forms of mental illness. I would argue against this strong interpretation, in favor of clearly establishing character traits that might cause deviation from self interest before plot points hinge upon them – someone with anger control problems should be shown being enraged in a context that establishes their tendency for anger to overwhelm their other traits, for instance. Furthermore, particular circumstances can definitely make poor self-control or other self-defeating behavior reasonable. However, the general rule should be: err on the side of the characters behaving in a reasonable way, and make the justification for any self-defeating behavior clear.) This particular rule prevents us from grossly misrepresenting the decisions of minor characters in order to move the plot along – it does away with the ‘magical negro’ trope, ‘fridging’, ‘badass’ female characters who are inexplicably kidnapped, and every other case wherein a character sacrifices themselves to the plot; it can furthermore act against inaccurate and stigmatic representations of mental illness, since even characters with established mental illness will act rationally within their mental framework (we cannot excuse a character’s behavior with “well, they’re crazy”, but we can point to manic or depressive behavior and explain it if the character is established to have those patterns of behavior).

2) Problem-solving tools are introduced prior to their use, and the setting and world rules are established prior to those elements being important to the plot. In other words, we cannot introduce a previously-unknown force during a dramatic moment that kills off half the characters. (It is possible for the characters to be unaware of these things and act accordingly, but the reader should not be.)

3) Stereotypical shorthand is not used to code particular types of characters; instead, the beliefs of the characters are extended out to produce their behavior. (In the original form, this was limited to intelligence – we eschew technobabble, bowties, and other nerd stereotypes in favor of spending more mental effort planning the behavior and thought processes of level-2-intelligent characters than level-1-intelligent characters. I think we can reasonably benefit from extending this to all characters.)

4) Eschew a good-versus-evil conflict in favor of a good-versus-good conflict. (In other words, avoid straw villains. If two intelligent, knowledgeable, and reasonable people cannot disagree on a subject, it does not make for a good conflict. However, to the extent that it is reasonable, two characters on opposite sides of the conflict can have access to different information.) As a corollary, avoid conflicts without an intellectual basis: while conflicts over resources are realistic, the intellectual justifications of imaginary conflicts over imaginary resources are relatively uninteresting.

5) When a primary character makes an important and well-reasoned decision, step through their reasoning. (This should probably be done for both protagonists and antagonists; furthermore, I would do this for supporting characters, for any important decisions they make.) Showing your work prevent you from accidentally causing a character to make a poor decision out of laziness, and establishes the interiority of the character to the readers, exposing beliefs that may not have been made as explicit from their actions as the author thinks. When performed with antagonists, it is a useful tool for the author to avoid straw-man antagonists and avoid giving readers the impression that an antagonist is a straw-man.

As for problems I have with the original formulation of rationalist fiction, I have a few:

1) Assuming all characters behave rationally all the time is not only unrealistic but also problematic. A clear head is a privileged position provided in part by resources like time, energy, food, and lack of mental and physiological stress. Behaving in one’s own interest in a complex and forward-thinking way when any of these resources is removed is a skill that must be learned from practice, and a skill that is rare. Furthermore, the occasional unconsidered decision – so long as it is not *transparently* stupid – injects some entropy into a story that would otherwise become a relatively predictable game of minimax. The general rule of thumb should be that poor decisions are rare and irrational decisions are both even rarer and justified, and that furthermore those decisions should lead to a more interesting world-state than could have been trivially predicted (but one that could with some effort be predicted).

2) While we all at some times learn a great deal from the though processes of others, it’s presumptuous to hold up the internal monologues of rationalist fiction as a model of rational thought. Certainly some readers will find certain ideas and strategies clever; however, an honest author will learn just as much from her own attempt to construct these trains of thought. Rather than considering these as models of rationality, consider them as interesting exercises in critical thinking and rhetoric – to benefit the author as much as the reader.

3) Trying to teach science in these stories is kind of stupid. Even if there’s a plot-justification, it’s liable to seem out of place; furthermore, you risk painting some of your characters into a corner in terms of justifying their familiarity with a topic and their ability to illustrate it – since teaching is a skill, and many characters are completely justified in not having it (particularly if it interferes with their other character traits). While Yudkowsky recommends avoiding it without sufficient plot justification, I would recommend avoiding it altogether unless it cannot be avoided.

4) Claiming that the only piece of media to do this before HPMOR is The World of Null-A betrays a lack of awareness of whole popular genres. Yudkowsky has seen Naruto, but despite his interest in rationalism and intelligent protagonists, is totally unaware of Spiral, Umineko, and every other anime that fits his ruleset to a tee? Having a highly rational protagonist engage in a battle of wits against highly rational antagonists is a pretty common pattern, and in many of these examples even my extensions of the rules are kept. I will admit that Spiral arguably fails in its attempts to keep itself as rational fiction in its second half; however, Yudkowsky described a set of rules defining the best of a genre that is alive and well across the Pacific and is claiming to have invented it.