Know the Rules So You Can Break 'Em

    Ronald Knox is a name that pops up with some regularity in Japanese detective fiction. You've probably never heard of the early-20th-century author unless you're an avid mystery fan, and that's to be expected--his books don't exactly feature prominently in popular memory. Knox is mostly known for his "10 Commandments of Detective Fiction," a set of rules penned to assist aspiring authors in what was then an overly saturated genre to avoid writing stories that are either trite or frustratingly unsolvable. The strict text of his rules is somewhat dated now, but the spirit of the rules is alive and well in most mystery fiction--and, of course, more than a few authors have taken his guidelines as a challenge to write satisfying mystery narratives that explicitly break what Knox saw as necessary guiding principles.
    In light of this (and because I think Knox's Laws are pretty fascinating), I thought I'd go through his laws one by one, analyzing the possible reasoning behind them and in some cases providing examples of ways they've been broken.

1. The criminal must be someone mentioned in the early part of the story, but must not be anyone whose thoughts the reader has been allowed to follow.

    This first rule--perhaps the cardinal rule of detective fiction--has two key parts to it. The first half of the rule is fairly straightforward and is adhered to consistently and (usually) intentionally. There's sort of an unwritten promise on the part of the author that the criminal in a mystery is not going to come out of nowhere. The culprit will be introduced, or at least referenced, as early as possible, often being among the very first characters introduced. For a masterful example of this, look at Persona 4's Izanami, who is (counting generously) the sixth character to be introduced, preceding the vast majority of the game's significant characters, but who draws little attention to herself and does not appear again until her role is revealed in the game's final arc.
    The second half of the rule seems rather obvious: of course anyone narrating the story cannot be the culprit, as the reader would see the crime happen and the mystery would be purposeless. Generally speaking, if this half of the rule is violated the work moves from "mystery" to some variant of "crime drama," as with Death Note (or, to provide an older example, Macbeth), which involves an investigation of a serial murder case, but which is not a mystery because the viewer follows the killer's thoughts throughout the whole show. I only know of one example of a true mystery that successfully breaks this portion of the rule--Danganronpa V3: Killing Harmony--and that particular instance is crafted with extreme care such that the violation of the rule works well.
    As a whole, this first law is meant to establish a sense of fairness on the part of the reader. Introducing the culprit early precludes arbitrarily introduced villains and encourages the reader to be skeptical of all characters introduced. Agatha Christie's And Then There Were None would be a lot less satisfying if the killer were not one of the ten major characters introduced at the beginning of the novel, for example. The latter half of the rule just reaffirms a basic assumption readers are likely to hold, as a poorly-written the-narrator-did-it twist could easily leave a reader feeling frustrated or betrayed.

2. All supernatural or preternatural agencies are ruled out as a matter of course.

    In Tanigawa Nagaru's The Sigh of Haruhi Suzumiya, Koizumi Itsuki differentiates the fantasy, science-fiction, and mystery genres by analyzing the way they explain supernatural phenomena, using the example of talking animals. According to Koizumi, in fantasy, animals can speak as a matter of course--no explanation is needed. In science-fiction, there is (usually) a reason given for why animals can speak, whether that's scientific progress or aliens or something similar. In mysteries, though, if an animal seems to be able to speak, there is always some sort of trick to it, such as hidden speaker built into the animal's collar.
    Revealing rational explanations for seemingly supernatural occurrences is such a staple of detective fiction that Tanigawa used it to define the genre. There is an expectation in detective fiction that everything is based on the world as we know it. This provides an accepted ruleset through which we can view the mysteries with which we are presented, and it allows us to rule out impossible occurrences by relying on our knowledge of what we know to be impossible in the real world. Allowing for the supernatural would make these assumptions unsafe. We could not, for example, immediately accept that a room locked from the inside was locked by one of its inhabitants unless we know that supernatural tools (like teleportation of telekinesis) are impossible.
    In that respect, this rule could be re-framed as, "common-sense assumptions about the world hold true." There are examples of mysteries that successfully incorporate elements of the supernatural--Takumi Shu's Ghost Trick: Phantom Detective is one, as the game's detective is a poltergeist trying to solve the mystery of his own murder--but even in those cases the supernatural elements are usually introduced early, clearly defined, and carefully restricted to their initial rules. The key point of this law is to ensure the mystery's world behaves in predictable ways, which  means supernatural forces can be allowable so long as their rules are defined such that they fall within what the reader accepts as "common sense" with regards to the mystery's world.

3. Not more than one secret room or passage is allowable.

    This one's pretty self-explanatory, and I don't have any notable exceptions to the rule on-hand. This falls into the category of rules that are intended to avoid clichés and to keep from frustrating the reader, as characters popping out of secret passages left and right could quickly get old. Secret rooms and hidden passages are such an established component of detective fiction we really don't think much of them anymore.

4. No hitherto undiscovered poisons may be used, nor any appliance which will need a long scientific explanation at the end.

    This is related to the second law. Where the second law says, "Fantasy explanations cannot be used," this one says, "Science-fiction explanations cannot be used." Much of the same analysis applies, although the wording of this rule suggests a bit more of a focus on the resolution of the mystery specifically. This rule is partially designed to avoid lazy deus-ex-machina solutions to puzzles, which would consequently create mysteries that are not solvable by the reader. If the solution to the mystery requires a long scientific explanation, the reader is not going to be able to solve the mystery through the clues provided, and the mystery is unlikely to be satisfying.
   For a case-in-point example, look at Zero Escape: Zero Time Dilemma, which sets up an intriguing mystery and then solves it through a mixture of parallel worlds and time travel. The former mechanic is foreshadowed and explained well-enough (mostly in the prior game, Virtue's Last Reward), but physical time travel is explicitly stated to impossible, which makes its surprise usage to solve a major mystery--one that dated back to the game before--singularly irritating, as it violates the "common sense" of the game's world. In this is a lesson that Knox's Laws are still valuable guidelines and should only be violated with careful planning and good reason.

5. No Chinaman must figure in the story.

    This is the most dated of Knox's laws and it needs a bit of explanation as a result. The intention behind this one is that the culprit should not be an outsider, someone who exists outside of the primary social networks of the characters involved in the crime (as Knox was approaching this from a Western viewpoint and from his perspective Asian characters would be outsiders). This is, in part, to keep the solution to the mystery from being too obvious--the noticeably different character can be a suspect, but never the true culprit. The other reason for this is that the reasonings, motives, and social tensions for crimes are often just as interesting as the strict logistical components of the mystery, and the expectation that the culprit will have a personal stake in the crime both makes the criminal more believable and allows for tension between the story's characters leading up to the final reveal, as suspicions are lobbed back and forth and tempers run hot.
    Despite the dated wording, this law actually holds up quite well in modern detective fiction. It's pretty uncommon for the culprit to be unrelated to the primary group of people affected by the crime, and even in cases where the culprit is an outsider (such as the final trial in Phoenix Wright: Ace Attorney: Justice for All) there are generally extenuating circumstances that place at least some of the responsibility on members of the "in-group." This also makes intuitive sense, as people generally don't commit crimes for no reason, and random thefts and the like don't make for interesting mysteries.

6. No accident must ever help the detective, nor must he ever have an unaccountable intuition which proves to be right.

    This is an easy rule to accidentally break, and it's a quick way to lose reader investment in a mystery. It's important that the reader feel like the mystery is both sensible and solvable. The second and fourth rules cover the "sensible" concern, ensuring a mystery adheres to its own rules, but this rule focuses on the "solvable" part. If the detective has a sudden and unexplained insight into an aspect of the crime or the culprit's identity, it leaves the reader feeling somewhat left out, as that usually means no clues leading in that direction were presented and the reader is forced to rely on the detective's judgment. This eliminates the fun of solving a puzzle that comes with reading detective fiction, and makes for a much less enjoyable experience overall.
    Avoiding things like this also, somewhat paradoxically, makes the detective seem more intelligent and admirable, rather than less so. Being presented with a clear line of reasoning that leads to the detective's conclusions makes those conclusions more believable and also more impressive. The reader's reaction to the detective's reasoning should be "Wow, I totally missed that!" rather than "I never would have got that." It's a slight difference in nuance, but it has a significant effect on the reader's enjoyment of the story.
    Although I've already acknowledged Death Note is not a mystery and therefore should not be held to Knox's Laws, this is a principle the show's two main characters--particularly the lead detective, L--break with a great deal of regularity, and it contributes to my antipathy towards the show. In Death Note we see the thoughts of both the killer and the detective, and much of the tension and excitement comes from their attempts to outwit each other. This leads to some great moments of gambit chess, but it also makes the player less sensitive to the absurd logical leaps that are sometimes made. L makes several arbitrary guesses and is right almost every single time. The viewer doesn't tend to question these guesses, as the viewer already knows the guesses are right and they seem reasonable as a result, but with a bit of thought L's reasoning rapidly becomes difficult to believe. His unaccountable intuitions make for thrilling individual moments but an unbelievable overarching story, which weakens the work as a whole. To use a tabletop game term, L seems to be metagaming, or operating based off of information his writer knows but his character should not logically have access to. This is something of a cardinal sin in games like Dungeons & Dragons, and it's equally unsatisfying in more traditional fiction.

7. The detective must not himself commit the crime.

    Another seemingly obvious one, and sometimes redundant in context of the first law. The distinction here comes in stories where the narrator is not the detective, like in Sherlock Holmes, which is generally narrated by Holmes's assistant, Watson. Even if the detective is not the protagonist of the story, it is generally necessary for him to be trustworthy, as he is the one providing clues for the reader and unraveling the mystery itself. If the detective is unreliable, the mystery usually becomes unsolvable and consequently unsatisfying. Even when this rule is broken, it tends to be broken in half-measures, with a detective, but not the detective around whom the story centers, being the culprit (as happens in Ace Attorney on occasion), or with the culprit appearing to be the detective while another character has, in fact, been subtly filling the role (as in Danganronpa V3).
    If pressed, I would call this the closest to an inviolable rule of all of Knox's Laws. All of the others can be circumnavigated with clever and careful writing--even if it may be exceptionally difficult to do well--but this rule is one that often holds even in works where the author appears to be directly attempting to break it. The reason for this is the flexibility with which "the detective" is defined. The detective is not necessarily an actual detective--he may be a journalist on scene, or just a curious or concerned citizen. The detective need only be the character working actively to uncover the truth of the mystery. In a case where the ostensible detective is in fact the culprit, that character is working to hide the truth rather than to uncover it and consequently is not actually filling the role of the detective. Either another character steps into this role, or you're looking at something more akin to a crime drama or a heist than a mystery.

8. The detective must not light on any clues which are not instantly produced for the inspection of the reader.

    This is another rule that plays into the importance of a solvable mystery, but there's a little more wiggle room here than the wording implies. At the heart of this rule is the necessity that the reader be provided with enough clues to solve the mystery. The detective can (and often does, assuming he is not the protagonist) conceal clues or information so long as that information is not pivotal to solving the mystery. In this case, though, the reader must be made aware that the detective is concealing something, and that fact may itself be an important clue regarding the overall case.
    When Knox refers to a "clue" in this rule, I believe he means something slightly broader than the concrete evidence the detective might stumble upon. A footprint without context, for example, is not a clue. A footprint that the reader is informed matches the shape of a known character's shoe, on the other hand, might be. The reader must be provided with all necessary and relevant information, and all such information must be produced as soon as the detective finds it (or at least confirms it to be true). Extraneous clues that might influence the behavior of the detective or tie together loose threads can be safely concealed so long as the reader knows something is hidden and the broad strokes of the case can be solved without the hidden clues.

9. The stupid friend of the detective, the Watson, must not conceal any thoughts which pass through his mind; his intelligence must be slightly, but very slightly, below that of the average reader.

    This is the longest of the laws, but it's also pretty straightforward. Assuming "the Watson" is the narrator, he is not permitted to hide his reasoning from the reader and his thoughts must be clear and simple enough that the reader can understand them, without being frustratingly slow to catch on to what's happening. This serves two purposes. First, it ensures the reader has time to reason out the mystery before the narrator explains it directly, and second, it prevents the reader from missing out on key deductions or getting lost. The detective is often highly intelligent and poor at expressing himself, so "the Watson" serves as an interpreter of sorts, making the mystery easier to digest.

10. Twin brothers, and doubles generally, must not appear unless we have been duly prepared for them.

    The final rule is also concerned with the solvability issue. Twins and doubles are an easy way to, for example, cast suspicion on the wrong character or to allow someone to seemingly be in two places at once. Readers are generally going to assume that if a credible witness sees a character at a specific place and at a specific time, that character cannot have been anywhere else. In order to account for that perfectly reasonable assumption, twins must be introduced before they matter or not at all.

He's the Culprit! Grab Him!

    It's easy to look at the wording of Knox's Laws and think they're too strict or that they're easy to break well (and some of them certainly are), but if you look at bit deeper at the intent behind each guideline, you'll find that most contemporary detective fiction still adheres to these rules pretty closely. A well-crafted violation of these intuitive standards can be really fun--just look at Danganronpa V3 for an example--but more often than not breaking these commandments leaves the reader unsatisfied. Plenty of creative and fun mysteries exist within the framework Knox outlined, and while the exceptions are neat, there's nothing wrong conventional detective fiction.
    In any case, it's interesting to break down these conventions, going beyond the "what" and examining the "why." Knox's works may not have had the staying power of the likes of Sir Arthur Conan Doyle or Agatha Christie, but he clearly had a pretty good understanding of what makes mysteries work, and his thoughts still influence today's authors of detective fiction.