The fifth essai in a series, following ‘Teaching Pegasus to crawl’. The original appeared on LiveJournal in May, 2006.
Fiction is, among many other things, a game between writer and reader, a kind of mental strip-tease in which the writer slowly reveals the details of the story, and the reader tries to guess at their significance. Mystery stories exhibit the game in its purest form, of course; but the element of guessing ‘whodunit’ turns up in every kind of fiction.
As stories have grown more complex, and the telling more elliptical and compressed, guessing out the storyteller’s meaning has become a difficult and demanding skill. Usually we don’t think of it in those terms, because as readers, we began to develop that skill early in childhood; it was fun to do, and after all, children can take delight in the most fiendishly elaborate games. Generally speaking, we don’t notice the skill involved until it stops working — that is, until the writer breaks the rules of the game. Then our involvement in the story goes up in a puff of what is sometimes known as ‘fridge logic’. All the inconsistencies, the factual errors, the implausible connections and narrative slip-ups, which we ignored as long as we were inside the story, come back to us in a rush, and we can never enter imaginatively into that story again.
To prevent this, writers have developed a complex vocabulary of cues and signals, to let the reader know when she is receiving the straight goods and when the writer (or a character) is deliberately misleading her. The largest part of this technique falls under the heading of point of view. This simplifies the game and reduces most of it to one question: Who is the narrator, and what does he know?
A first-person narrator is expected to be generally truthful, unless he ‘accidentally’ betrays, early in the story, that he is a liar and his words need to be sifted. Even if he is lying, we expect the lies to be self-serving — attempts to colour and distort facts that we might find out from some other source, rather than to invent a narrative from whole cloth unrelated to the truth. If the narrator is making up a story that is fictitious within the context of another made-up story, and that framing story is not even told, the reader is going to laugh bitterly at the whole pointless exercise and put down the book. Not even Borges cared to risk that kind of recursive knot.
If the story is written in ‘close third’, so that we follow one character’s thoughts and perceptions at a time, we know that the narrative is subject to the limitations and biases of that character’s knowledge; but we also have a corrective, for we trust the omniscient author’s selection of the viewpoint character. The writer may not, at any given moment, have chosen the character who gives us the most truthful or accurate account of the facts; but we have faith that all the essential facts will be there in the end, and discoverable, so that we will get all the elements we need to make sense of the story.
The extreme points of view, which are also the simplest, are always assumed to be reliable, because they allow us no other imaginative angle from which to correct their unreliability. If the story is told in a camera-eye viewpoint, we expect the camera to function properly and to record faithfully everything that it sees. If the story is told by an omniscient narrator, we expect the narrator to remain true to the facts he has already laid down. In Chapter One of an omniscient tale, anything goes; but Chapter Two is limited, for the things already said in Chapter One are as the law of the Medes and the Persians, which altereth not. And if Chapter Twenty-five sees the author painted into a corner, so that no consistent interpretation of events is possible, then we expect him to go back to Chapter Four and cleverly put a door in that part of the room before releasing his work to the public.
These are the general laws of the game — we may call them the laws of narrative protocol —which we as writers break at our peril. But there are other laws, the laws of character and conflict, which sometimes clash with them; it is our job to make sure that the clash never occurs — and if it does, we must edit it out. Repainting will not do; we must go back and put in that extra door in Chapter Four, or make still larger alterations to the structure, until the rules of viewpoint and the nature of the story run smoothly together. Everything that exists for a reason outside the story (to convey information to the reader) must also have a reason inside the story; at the very least, there should be no reason inside the story why it should not happen.
In particular, we have to make sure that the story is being told by someone who could plausibly tell it. Someone identified only as ‘rysmiel’ left this comment when Sherwood Smith discussed the question:
I have a strong dislike for first-person narratives where there is no inherent reason for the narrator to be writing this stuff down, nor context in which they could reasonably have done so. I particularly don’t like this when the narrative gives away terrible secrets, or casually mentions things that would be liable to send the writer/protagonist to jail. It’s not a book-killer, but it will reliably make me fume. First person narratives should have a reason for existing, and unless they are explicitly diaries there should be an assumed person they are written for.
This is exactly the reaction we can expect our readers to have if we violate narrative protocol. When it happens, we have no excuse or defence: if we saw in the first draft that this problem was going to arise, we should have chosen a different protocol and rewritten as required. We should have put that door in Chapter Four.
In the drama, when an actor directly addresses the audience (in character, not in the polite exordium that used to be common in plays) or otherwise interacts with it, they call it ‘breaking the fourth wall’. It is permissible as a form of comic relief, and in certain kinds of ‘experimental’ theatre, but in general it’s a no-no, and has been ever since they discontinued the iniquitous practice of putting high-priced seats right on the stage. The novel is much more forgiving about the fourth wall, because the author is permitted, indeed expected, to make up his own rules about it. But once he has made the rules, we expect him to play fair and stick by them. If he doesn’t, any but the least attentive reader is likely to find herself bounced right out of the story, ‘looking at the little abortive Secondary World from outside’, as Tolkien put it. Storytelling is very like a mild form of hypnotism, and if the hypnotist violates his subject’s trust in any obvious or startling way, he will almost certainly break the trance.
At one time, when stories were recited and not read, it was usual to exalt the storyteller’s vanity over the reader’s intelligence, by including all kinds of dramatic ‘turns’ that would allow the skald or scop to show off his histrionic skills, at no matter what cost in narrative probability. The most famous ‘turn’ is, of course, the high-flown speech made by a heroic character on the point of death. If he is slain in battle, then the battle itself must stop to listen while he makes his declamation; nay more, even his blood must stop flowing, and if his lungs have been punctured by arrows, they must somehow do their office long enough to give the ham his moment in the limelight. It is the same impulse that forced composers and librettists to write beautiful dying-swan arias for their expiring divas.
Now, when a story is recited, that is, when it is treated as a dramatic performance, this kind of trumpery is permissible, if ill-advised. But when the story is written to be read, the matter is more difficult. Instead of putting our trust in the bard, we put it in an invisible narrator, a harder thing for most of us to do. We are always tempted to ask such questions as: ‘Is this true?’ ‘Does this make sense?’ And above all, ‘But how do you know?’ And the wilder and more improbable the story, the more careful the writer has to be to give, or at least solidly imply, plausible answers to these questions. It is dangerous to let the reader think too much about the narrative assumptions of the story, but it cannot always be prevented. It is fatal to let her reach the conclusion that you are not playing by your own rules.
I have written about one of the most annoying ways in which an author can violate his own protocols. In Report on Probability A, Brian Aldiss sets up an elaborate game with point of view, based on the rather precious idea that none of his characters are willing to make even the most obvious assumptions about each other’s psychology — that they are all human, for instance, or that the articulate noises coming out of their mouths are in fact English words. But he is grossly inconsistent, for his characters are continually slipping and making assumptions about each other’s thoughts and motives. Worse yet, they know things that they could not possibly have observed. One character thinks that his broken clock might still work, but is afraid to wind it up and prove himself wrong. We find this out from the supposedly objective, camera-eye report written by another character watching him from afar. The best camera in the world could never observe any such thing, and if it could, we would have no grounds at all for doubting one another’s psychological makeup. Aldiss bursts the bubble of Secondary Belief, not once but with dreary and deliberate regularity, by this same elementary blunder. And since the game is the entire plot and point of the story, once the rules are broken, the remainder of the tale equals nothing.
You can buy entire how-to-write books that deal with nothing but point of view: first person, second person (rare and inadvisable), limited third, omniscient third, camera eye, deep penetration, and the fabled ‘M’ structure. The list reads like a Kama Sutra of voyeurism. Many readers are obsessively concerned with these details; getting them wrong is a quick and sure way to lose them.
It was not always so. The omniscient point of view was good enough for almost every writer of fiction until the advent of the novel in the 18th century. The degree of ‘penetration’ varied, of course. In Beowulf, I am reliably informed, there is only one line in which we are made privy to a character’s interior mental processes: ‘his breast boiled with dark thoughts, as was not his custom’. But we are not told any details about the dark thoughts, and Beowulf goes right on with his preparations to fight the dragon. As long as narratives were so little concerned with psychology, and so much with action, point of view was not important enough to take trouble about. The question is not even raised, for instance, in Aristotle’s Poetics. In the Greek epic, it is taken for granted that the narrator knows the whole story and tells it reliably; in drama, of course, the question does not even arise.
Then Defoe gave us first-person narration, and Richardson followed up with the epistolary novel, the first form that really encouraged the exploration of multiple characters’ thoughts. The psychological novel burst on the scene, and nobody since has been able to tell a story in quite the naive, old-fashioned way. We have to worry about these things now.
A century later, Edgar Allan Poe perfected the unreliable narrator, and so we must also question at every moment whether our storyteller is telling lies. We are playing a constant and intricate detective game with our texts, and this has become so integral to our reading of novels that we almost feel cheated if a narrator proves too reliable.
But in fantasy, and to a lesser extent in SF and other imaginative literatures, it is vital that the narrative be reliable and controlled. We must always be able to distinguish between impossible events in ‘reality’ (the characters’ reality) and mere illusions or hallucinations, even if the characters themselves cannot. The default assumption must be that any event, however bizarre, is essentially what it appears to be; and if we violate that assumption without giving clear cues, our readers will be furious because we have violated their trust. I have made this error myself, and reaped the consequences in sufficient abundance.
Nobody knew this better than J. R. R. Tolkien, who expended enormous effort to give his fantasies a solid air of vera historia. The Appendices, prologues and footnotes, the maps and other impedimenta of The Lord of the Rings, all serve to build up a massive and unquestionable assurance of the fundamental consistency and honesty of the text. One could almost believe that there really was a Red Book of Westmarch, painstakingly translated by modern scholarship, and that one is getting a glimpse of the history and culture of a former world. Nobody else has taken the same degree of trouble, which is one reason why so much other fantasy seems arbitrary and frivolous by comparison.
It is amazing how consistently Tolkien accounts for everything said to be written in the Red Book. There are no windy death-speeches, no reports on events that did not come from a surviving eyewitness. We never meet Sauron, not because Tolkien wants to keep his Ultimate Evil vague and depersonalized (a frequent and foolish accusation), but simply because ‘those who pass the gates of Barad-dûr do not return’. We are made privy to Gollum’s thoughts only because of his habit of ceaselessly talking to himself, and we never glimpse the soul of an Orc. Indeed, when we come closest, in the talk between Shagrat and Gorbag at Cirith Ungol, Tolkien takes care to explain how Sam was able to understand Orkish (by the power of the Ring, of course, greatly increased by proximity to Mordor). In some of the outtakes from The Return of the King, he goes still further, specifying the periods after the Field of Cormallen when Frodo shut himself up in a room in Minas Tirith to write the copious notes that he brought back to Bilbo at Rivendell. And at the very end we see the Red Book itself, eighty chapters long, bringing the story down to Frodo’s departure for the Grey Havens. I cannot think of anything so scrupulously buttressed in all of literature before it.
But even Tolkien nods. The most beautiful face may be marred by a tiny flaw that would pass unnoticed on a person of average looks, and the most meticulous argument can be spoilt by a venial error. Only once does Tolkien slip and tell us the thoughts of someone who could not possibly have been interviewed by the authors of the Red Book:
A fox passing through the wood on business of his own stopped several minutes and sniffed.
‘Hobbits!’ he thought. ‘Well, what next? I have heard of strange doings in this land, but I have seldom heard of a hobbit sleeping out of doors under a tree. Three of them! There’s something mighty queer behind this.’ He was quite right, but he never found out any more about it.
The temptation is strong to make excuses for this ghastly lapse. It could have been put there to signalize (for this is quite early in Fellowship, when the hobbits are still in the Shire) that this part of the Red Book was written by Bilbo, a less truthful and more florid narrator than Frodo. Other pleas could doubtless be devised. But in fact it is pure fiction, even within the characters’ own frame of reference, because all three hobbits were fast asleep when this is supposed to have happened. Nobody ever saw the fox again, or asked him his opinion of the matter; nobody even had any way of knowing he had been there.
Really, this is nothing more than the last faint twitch of the twee, talking-down, children’s-bedtime-story reflex that so damaged The Hobbit (in Tolkien’s own opinion). There were considerably many such touches in the first drafts of ‘the new Hobbit’, but Tolkien rooted them out with great thoroughness and industry in the revision. Only this one remains, like a lone outcrop of rock in a level plain where once great mountains stood. I always stumble when I read this passage, and marvel that Tolkien let the book go to press in two successive editions without excising it.
It is this kind of thing that prevents me from taking A Song of Ice and Fire as seriously as Tolkien’s work. Martin, like the scops of old, plays the ham with Death. Addicted to visceral thrills, he cannot resist the temptation to show us a character’s dying thoughts and feelings, even when the character dies alone. I cannot now put my finger on a good example, for the books are very ill organized, but the body count is enormous and suitable examples should not be difficult to find. Every time he resorts to this device, I am reminded that I am only reading a novel, after all, and I am distanced from the characters and their plight. The extreme violence of the situations combines with the careless abuse of narrative integrity like a binary poison, so that far from intensifying my experience of the story, it vitiates the impact and leaves me regarding the text coldly and critically from outside.
This is not a fatal fault, but it is a grievous one. Like the tiny flaw in the otherwise perfect face, it is more disfiguring to a good book than to a bad one. Martin is a brilliant writer, but he falls just short of being a great one, partly because he continually undermines himself by these small bouts of self-indulgence. ‘Show, don’t tell,’ like all rules, should not be followed off a cliff. Martin not only follows it off cliffs, he makes Death keep his camcorder running right to the moment when the body is smashed to jelly on the jagged rocks below.