Book 1 of seven in the Harvest Moon Week: small books written the week the Harvest Moon rose full over the ridge, each one set on one of the nights that follow it. This one is set on Saturday, September 26.
Frost on the Gauge
The rain gauge stands on the corner post of the garden fence, a clear tube in a bracket, marked in hundredths of an inch. This morning, the Saturday after the Harvest Moon, it held nothing at all. A thin rime had settled on the outside of the tube in the night, and the steward, walking out at first light with a mug in one hand and the coop key in the other, stopped to look at it the way you look at a clock you already know the time on.
Empty. No rain since it was last tipped out. That is an answer. It is a small answer, and a dull one, but it is a true one, and it was written down in the house book in pencil: gauge empty, frost on the tube.
The moon was going down in the west, a day past full and just starting to lose its edge. The hens were grumbling behind the door the way they do when the light comes later than they would like. The grass along the fence had gone stiff and silver where the cold air pooled, and green again where it had not. The dark half of the year is here, and it announces itself this way: a little later each morning, a little colder at the fence line, a little more quiet.
Then the steward went down the basement stairs to where a little computer sits on a shelf between the canning jars and the water heater, and looked at the other gauges. The ones on the screen, that are supposed to tell the household about the sky and the air and the radios without anyone walking out to a fence post.
This week, a good many of those turned out to be liars.
Not all of them, and not maliciously. But when the steward went through the whole system, card by card and table by table, he found numbers that had never been measured by anything. Placeholder numbers that had crept in when something failed and then sat there looking exactly like the real thing. Some had been there for months.
This is the story of that week, and of what the old world already knew about it.
THE REPORT THAT LIED: THE MORNING MEASURE
+----------------------------------------------------------+
| Sunrise / Sunset ..... about 07:00 / 18:58 |
| Photoperiod .......... about 11h 58m, losing ~2.5 min |
| Moon ................. Waning gibbous, a day past full |
| Ridge temperature .... -- (not read this morning) |
| Season ............... Dark half, first frost nights |
+----------------------------+-----------------------------+
|
+----------------+----------------+
v v
[ TIWAZ: THE TRUE MEASURE ] [ ANSUZ: THE TRUE WORD ]
Every number is measured, Say what you know, say
or it is a blank. what you do not. Both.
Look at the fourth line of that box. The temperature was not read this morning, so the box does not say “38 F” because that is about what it has been lately. It says two dashes. You will see a lot of those two dashes before this is done.
Chapter 1: The Number That Never Changed
It started with the morning summary.
Every morning the little computer writes a short report and sends it upstairs: what the weather did overnight, what the day may bring, and how the garden’s warmth account stands. That last is a growing degree number, one of the oldest and most useful tricks a grower has.
Plants and insects do not keep a calendar. They keep a thermometer. A seed of corn does not know it is the tenth of May; it knows how much warmth it has soaked up since it went into the ground. So growers keep a running tally of heat. Each day you average the high and the low and subtract a base temperature below which the crop does nothing much, commonly fifty degrees Fahrenheit. A day that ran from forty-eight to seventy-two averages sixty and banks ten growing degrees. A cold day banks nothing. Add them up and you know, far better than the calendar can tell you, when the beans will flower and whether the late squash has any hope before frost.
It is a beautiful little piece of arithmetic. It turns weather into a ledger. And like any ledger, it is only as good as the entries.
The steward had read that summary for a long time the way you read the thermometer on the porch post: a glance, a nod, on with the day. This week, because he happened to be paying attention, he noticed the growing degree number was the same as yesterday’s. And the day before. And the day before that.
Not close. The same. To the decimal.
In late September on a high ridge the heat account barely grows, and some days bank nothing. But a number that does not move at all, through warm afternoons and cold ones, through the equinox and past it, is not a measurement. It is a painting of a measurement.
So he went looking, and found that when the real calculation could not be done, a fixed number had been written in to fill the space. Probably a stand-in from when the report was first built, the way a carpenter tacks up a scrap board to hold a gap until the real piece is cut. The real piece never got cut. The scrap board got painted to match the wall, and every morning since, the report had printed it with a straight face.
Here is the part that stings. It was a reasonable number, the right size for the season. If it had said nine thousand, or a string of question marks, he would have caught it the first morning. It was caught only because it was too steady, and because someone happened to look twice.
That is the first lesson, and it is the one the rest of the week kept teaching over and over: the dangerous number is not the absurd one. It is the plausible one.
A crash announces itself. When a machine falls over, the screen goes dark or a card turns red, and someone goes and looks. A crash is a man shouting that the barn is on fire. A plausible false number is quiet. It sits where the truth should be, wearing the truth’s clothes, and it gets read, and nodded at, and believed. Every day it is believed, it spends a little of the household’s trust and gives nothing back.
The fix for the report was straightforward. The number now comes from the day’s real measured high and low, calculated fresh each morning. If either is missing, the report does not guess. It prints two dashes and says plainly that the day could not be counted.
The fix for the steward was less straightforward. Because once you have found one painted scrap board in a wall, you have to wonder about all the others. So the week became a walk through the whole house, knocking on every board to hear which ones were hollow.
Chapter 2: The King’s Standard
People have been lying with numbers for as long as there have been numbers, and for almost as long, people have been making laws against it.
The oldest form of the lie is the false scale. A merchant keeps a heavy set of weights for buying and a light set for selling, and profits in the difference. Or he shaves the rim of his measuring bowl, or thickens its base, so that his bushel holds a little less than a bushel. The customer sees the scale balance, sees the bowl filled level, and goes home content. The ceremony of measuring was performed in full view. And it was a lie.
The old writers were not gentle about it. The Book of Proverbs says plainly that a false balance is an abomination, and a just weight is a delight. That was written for people who bought grain and oil by weight every week of their lives, and knew exactly what it meant to be cheated by a stone a little too light.
In England the cure was the king’s standard. Magna Carta, the great charter the barons forced on King John in 1215, is remembered mostly for its lines about justice and the law of the land, but tucked into it is a clause about measures: there shall be one measure of wine throughout the realm, and one measure of ale, and one measure of corn, and one width of cloth. It sits right there among the great liberties, and it belongs there. A kingdom where every town has its own bushel is a kingdom where no one can trust a bargain made outside his own village. One measure for everyone is a kind of freedom.
Over the following centuries the crown kept official standard weights and measures, and towns checked their own against them. Officers of the market tested the scales and bowls of traders. A merchant caught with false weights could have them seized and broken, and could be fined or set in the pillory before the very neighbors he had cheated. He had lied in the marketplace, so he answered for it in the marketplace.
The bakers had their own law. In the thirteenth century the Assize of Bread and Ale tied the weight of a loaf to the price of grain, so that anyone could know what a fair loaf should weigh, and bakers who sold short weight could be fined or shamed. The usual story of the baker’s dozen grows out of this: a baker selling by the dozen would throw in a thirteenth loaf, so that if one came out of the oven light, the lot could not be called short. Whether that is the whole origin of the phrase, nobody can swear to now. But the instinct in it is sound. When the law says you must not give short measure, the careful man does not aim for the line. He leaves a margin on the honest side.
The Norse world had its own version. In Iceland, for a long time, a common measure of value was homespun woolen cloth, wadmal, reckoned by the ell. A length of cloth was money, so the length of an ell was the value of every debt and fine in the land. And the law itself was held in a man’s head. The lawspeaker stood at the Althing each summer and recited part of the law aloud, so that over his term the whole of it was spoken before the assembly. The law was a standard kept in public and checked in public.
There is one more old device worth remembering: the tally stick. For centuries the English treasury recorded debts by cutting notches in a stick of hazel or willow and then splitting it lengthwise. One half went to the payer, one stayed with the treasury. When the debt came due, the halves were laid together. If someone had shaved a notch or cut a new one, they would not match, and everyone could see it. The tally was a record that carried its own check.
Every society that has depended on measurement has learned that measurement must be honest, that honesty must be checkable, and that the cheat most to be feared is the one who performs the ceremony of measuring correctly and puts a false number in the scale anyway.
A dashboard is a marketplace scale. It is where the household comes to weigh the day. When a card shows a temperature or a signal strength or a count of aircraft overhead, it is performing the ceremony of measurement in full view: a label, a number, a unit, a little graph. The eye believes it the way the customer believed the balance.
And this week, the steward found filed-down weights all over the shop.
Chapter 3: A Zero Is a Claim
The weather cards were next.
The dashboard pulls weather from the household’s own sensors and from outside sources that report conditions, forecasts, and air quality for the area. Most of the time this works. But a fetch fails, a service is slow, the ridge’s connection has a bad minute. That happens to every system that talks to anything beyond its own walls.
What matters is what the card does when it happens. And what these cards did, it turned out, was fill in the gap with a default.
The air quality card, when its fetch failed, showed a zero.
Think about what a zero means on an air quality scale. It does not mean “unknown.” It means the cleanest air there is, mountain air after a thunderstorm. On a day when smoke from a distant fire had drifted over the hills, that card could have been cheerfully announcing perfect air because a web request had timed out.
The temperature card did something similar. When it could not get a reading, it fell back on a default, a sensible looking number of the sort a programmer types in so the card has something to draw. On the days it failed, that was the temperature, as far as the screen was concerned.
Now go back out to the fence post.
The rain gauge this morning was empty. That emptiness is a reading. It says: since the gauge was last tipped, no rain fell here. A farmer can write that down, and put it next to last week’s readings, and know something true about his dry spell. An empty gauge is data.
But suppose the frost had cracked the tube in the night, and whatever fell had leaked out the bottom. The gauge would look exactly the same from the porch. And now the emptiness means something false. A farmer who writes “no rain” from a cracked gauge has entered a lie in his book, and he will decide about watering and hay on the strength of it.
There is a difference between a gauge that reads zero and a gauge that is broken. The first is the most honest thing an instrument can say. The second is the most dangerous, because from a distance it says the same thing.
Every old farmer who kept a rain book knew this in his hands. You look at the gauge, not just into it. You check that it is plumb in its bracket, that the tube is not cracked, that the funnel is not clogged with maple seeds. You read the instrument and the state of the instrument together, because a reading from a broken instrument is not a small reading. It is not a reading at all.
A machine has to be taught to do what the farmer does without thinking. It has to know the difference between “I measured, and the answer was zero” and “I could not measure,” and say the second in a way that cannot be mistaken for the first. That is why the dashes. Two dashes cannot be mistaken for clean air or a mild afternoon. They are an honest hole in the page.
Here is a way to think about it. Every number on a screen is a claim. A zero is a claim. It is a strong claim, in fact, because zero is a specific, confident, falsifiable statement about the world: there is none of this here. When the air quality card said zero, it was not staying quiet. It was testifying. And its testimony was invented.
The rule the household settled on, written down this week, has three parts:
A number on the screen must be one of three things. It is measured, by an instrument the household owns and trusts. Or it is pulled from a real outside source and stored, so that there is a record of where it came from and when. Or it is calculated from those, by a formula anyone could check. There is no fourth kind. There are no defaults. There are no stand-ins, no placeholders, no numbers typed in so the card has something to draw.
If a number is missing, the card shows two dashes. Never a zero that means “unknown.” Never a sensible guess. Never yesterday’s value dressed up as today’s.
Chapter 4: The Steward Who Reported Full Barns
This is a story. It is not from the sagas. It is the kind of story that could have been told at a hearth in any cold country, and perhaps something like it was.
There was once a jarl who held land along a fjord, and beyond the mountains, a long hard journey inland, he had an outlying farm with great barns for the grain of the upland valleys. The pass was closed half the year. So he set a steward over it, and every autumn the steward sent a rider down to the hall with word of the harvest.
The first year the harvest was good, and the rider came down the mountain and stood before the jarl’s high seat and said: The barns are full. The jarl was glad, and he gave the rider a ring, and he planned his winter on it. He knew that if the coast farms had a lean year, he could send for grain from the upland barns, and his people would eat.
The second year, the harvest was fair, and the rider said again: The barns are full. They were nearly full. The steward reasoned that nearly full was near enough, and that his lord liked good news, and that there was no sense in troubling a great man over a few empty bays.
The third year, a wet summer rotted the grain in the fields, and the steward got in less than half. He thought long about what to send. He thought of the jarl’s face when the rider had said full, and the ring, and the praise. He thought of how the jarl might wonder whether a better steward would have saved more of the crop. He thought that next year would surely be better, and that one bad year could be made up. And he sent the rider down the mountain with the same words: The barns are full.
The fourth year was worse.
Down at the coast, the jarl was a careful man. Each autumn he planned his winter on what his stewards told him, and the word from the upland had been full, full, full, full. So when the coast harvest failed that fourth year, when the herring did not come and the barley lay flat in the rain, he was not afraid. He told his people there was grain beyond the mountains. And when a poorer farmer across the fjord came begging, the jarl, believing himself rich, gave freely from his coastal stores, knowing he could refill them from the upland barns.
In the deep of that winter, when the pass could just be crossed on skis, he sent men and sledges up the mountain for the grain.
They found the barns nearly empty.
What happened after that, the story need not tell in full. People went hungry in that hall who need not have. Not because the harvest failed. Harvests fail, and every wise lord plans for it. They went hungry because the jarl had made his decisions on a number that was not true. Had the steward sent word in the third year, less than half, the jarl would have kept his stores close, bought grain in the south while the ships could still sail, and told his neighbor he was sorry. A hard winter, but one everyone lived through.
The lie did not cause the bad harvest. The lie caused the hunger.
Now there is a second steward in this story, and he should get his turn too.
The jarl, having learned something, sent a new man over the mountains. And the first autumn, the new steward’s rider came down and stood before the high seat and said something no rider had ever said there before.
The east barns are full. The west barns hold perhaps a third. The north barn, I do not know. Its roof fell in under the snow last spring and no one has been able to get inside to count it. My lord should reckon it as nothing until we can.
The men along the benches laughed a little. It sounded like a weak report from a weak steward. What kind of man does not know what is in his own barn?
But the jarl did not laugh. He had learned what a strong report costs when it is false. He heard, for the first time in years, a report that told him where the edges of the knowledge were. East: known, full. West: known, poor. North: unknown, and said so, and a sensible instruction for what to do with the unknown. Treat it as nothing until it can be counted.
He could plan on that. Every word of it could bear weight. And the jarl gave that rider a finer ring than the first one had ever got.
That second steward’s report is the whole of the week’s work in one speech. East is full. West is a third. North, I do not know. Measured, measured, blank. Nothing painted. Nothing guessed to spare a lord’s feelings. And a clear word about how to treat the blank.
The first steward’s sin was not incompetence. Harvests fail no matter who keeps the barns. His sin was that he filled a blank with a comfortable number, and then kept doing it because the number had been accepted before. That is exactly what a default value in a machine does. It fills the blank with the comfortable number, every time, forever, and it never feels a moment’s shame.
Chapter 5: Ghosts in the Ledger
The worst of it was in the radio history.
Out around the ridge there is a small mesh of long-range radios, the little mesh radios, each about the size of a deck of cards. They pass short messages across the hills, and each reports how strongly it hears its neighbors. That signal strength goes into a history table on the basement computer, so the household can see how the links behave through wet weather and dry, leafy summer and bare winter. A long, honest log of how radio moves over this particular ground is not something you can buy. You can only keep it.
When the steward went through the table this week, he found that a great many of the signal strength entries were not measurements at all. Somewhere in the code that gathered them, when a real value was not available, a fixed fake number had been written into the permanent record instead, row after row, looking exactly like the real ones on either side.
There were thousands of them.
This is a different kind of lie from the weather card, which lies in the moment and is gone when the fetch comes back. A lie written into history sits in the ledger and waits. Any graph drawn from that table next spring would have those ghosts in it. Any average would be pulled toward the fake value. Any decision about where to hang a new relay would be partly a decision made by a ghost.
This is where the tally stick comes back. The radio history had no second half to lay the fake rows against. They had been accepted into the record simply because they were in the record. The only check was a person going through and asking of each kind of entry, where did this come from? When the answer was nowhere, the entry was a ghost.
So the steward cleared them out. But not carelessly, and this matters.
Before a single row was touched, the whole table was copied and set aside, so that if the cleaning went wrong, or some value that looked fake turned out to be real, nothing would be lost. Then the fake values were blanked. Not deleted, row and all, but blanked: the row stays, the time stays, and where the fake number stood there is now nothing. An honest hole where the ghost used to be.
That is the right way to clean a ledger. You do not burn the old book. You make a fair copy, strike the false entries where anyone can see they were struck, and keep both. A future steward who has forgotten this week entirely can still tell “no reading was taken here” from “a reading was taken and it was this.”
There was a smaller ghost in the same room, and it is almost funny.
The gateway, the main radio at the house that listens to the others, was reporting that it could hear one of the relays, when the relay it claimed to hear was, in effect, itself. It was like a man shouting from a hilltop who writes in his log that a stranger answered from across the valley, when all he heard was his own echo.
It sounds trivial. But it makes the network look bigger than it is, and on the day a real relay goes quiet, the map will still show the house hearing something, because it is hearing itself. It is a comfort that is not earned. The gateway no longer counts its own echo as a neighbor.
Then there was the battery.
One device reported its battery at one hundred and one percent. That is at least an honest kind of absurd; nobody believes it. But traced back, the number was not a battery reading at all. It was a code meaning this device is plugged into power. Someone had decided to say “on wall power” by putting a number one past full into the battery field.
The information in it was true: the device was on power. But it was dressed as a measurement it was not. Any graph of battery health would show a line sitting just above full forever, and anyone who did not know the secret code might believe it. A true fact wearing a false uniform is still a false report.
The fix was to give the fact its own name. The battery field now says what the battery holds, or two dashes if the device does not report it, and a separate plain word says whether the device is on wall power. Two different truths, each in its own place.
Chapter 6: Tyr’s Hand and the Names of Things
Of all the gods, the one who belongs to this week is Tyr.
The story is told in the Prose Edda. The gods had raised the wolf Fenrir among them, and he grew so fast and so strong that they became afraid. Only Tyr had the courage to go near him and feed him. Twice the gods brought great chains and called it a test of the wolf’s strength, and twice he snapped them. Then the dwarves made a fetter called Gleipnir out of things that do not exist: the sound of a cat’s footfall, the beard of a woman, the roots of a mountain, the breath of a fish, and the like. It was soft as a silk ribbon.
Fenrir was suspicious of it. A ribbon that thin, offered as a great test, smelled of trickery. He agreed to be bound only if one of the gods would put a hand in his mouth as a pledge that there was no deceit in it. The gods looked at one another. Every one of them knew there was deceit in it. And no one moved, until Tyr stepped forward and laid his right hand between the wolf’s jaws.
The ribbon held. The wolf could not break it. And when Fenrir understood that he had been tricked, he closed his jaws, and Tyr lost his hand at the wrist.
People read that story many ways, most often as courage and sacrifice, and that is true. But there is something harder in it. The gods made a pledge that was a lie. The hand in the wolf’s mouth said there is no trick here, and there was a trick. Tyr knew the pledge was false when he gave it. He gave it anyway, for the good of all, and paid the full price, and never pretended he had not.
Tyr is the god of the oath, of law and the lawful assembly. His rune, Tiwaz, is drawn like an arrow pointing straight up, and the Old English rune poem says, roughly, that Tir is a token that keeps faith well, ever on its course over the mists of night, and never failing. It was a navigator’s star. You can steer by a thing like that only because it does not lie about where it is.
What the Tyr story teaches a builder is this: a false pledge always costs somebody a hand. Maybe not today, and maybe not the one who made it. The hungry hall in the parable paid. The customer who carried home a short bushel paid. And a household that makes decisions on a dashboard full of painted numbers will pay too, some cold night, because every card on the screen said all was well.
So the steward went looking for the false pledges.
The buttons were the plainest ones. On the dashboard there are controls: restart this, send that. When you press one, a small message says how it went. Several of them said “success” whether or not the command had worked. The button fired the request into the machinery and, without waiting to hear back, cheerfully reported that all was done.
That is a hand in the wolf’s mouth: a pledge that there is no problem, given by something that has not looked. It is the rider saying the barns are full without anyone having walked into the barns. The buttons now wait for the answer. If the command worked, they say so. If it failed, they say so and show what came back. If they cannot tell, they say they cannot tell. That last is the hardest thing to teach a machine and the most important.
Then there was the ping.
In the language of networks, a ping is a small, specific thing: you send a tiny packet to another machine and time how long it takes to come back, like timing an echo across a valley. The dashboard had a card that said “ping,” with milliseconds under it. But the number was not a ping. It was the time it took to fetch a whole web page from somewhere out in the world. A real measurement, honestly taken. But a page fetch includes the road there and back and also the far machine thinking, building the page, and sending it. It is a bigger, noisier thing.
Nobody was lying about the number. The name was wrong, and a wrong name is its own kind of false weight. Someone who knows what a ping is will see a slow number, blame the road, and fix the wrong thing, when it was the far machine that was slow. The label is part of the measurement. A basket with “peck” painted on the side is a lie even if every grain in it is honest.
So the card was renamed to say what it actually measures, the fetch time for a page, and nothing more. It is less impressive. It is true.
There is a kind of courage in renaming a thing down. The honest label is plainer. It is also a thing you can steer by.
Chapter 7: The Keeper’s Log
Along every dangerous coast, for a couple of centuries, there were lighthouse keepers, and they kept logs.
The work was mostly tending: trimming wicks, polishing the great lens, carrying oil up the tower. But the log was the heart of the job. The keeper wrote down when the lamp was lit and put out, what the weather did, what ships passed, how much oil was burned. Inspectors came and read the logs and looked over the light, and a keeper whose log did not match his station had to answer for it. The log could be laid against the oil in the tanks like one half of a tally against the other.
Every one of those entries had a time on it. The lamp was not simply “lit.” It was lit at a certain hour on a certain night. The fog came in at a certain hour and was noted again when it cleared. A keeper’s log is a list of facts with their ages attached. The dashboard had forgotten that.
Some of the cards on the screen, when their source went quiet, simply kept showing the last value they had received. The temperature an hour ago is probably close to the temperature now. But nothing on the card said it was an hour old. Or a day old. Or, in one case, much older. The number sat there as fresh as the morning, long after the instrument had stopped reporting.
That is the keeper who writes “lamp lit” and forgets to say it was lit last Tuesday.
A stale number is a particular kind of lie. It was true once. Everything about it was honest at the moment it was taken. But a measurement is a statement about a moment, and when the moment is not given, the reader assumes it is now. And “now” is the one thing a stale number is not.
So the rule grew a second clause. Stale data carries its age, and a STALE flag. Every card now knows when its number was taken. If the number is older than it ought to be for that kind of reading, the card says how old it is and marks itself STALE in capital letters, so no one can glance at it and mistake it for the present.
The threshold differs, as it should. A temperature sensor that reports every few minutes goes stale in a quarter of an hour. A radio that checks in a few times an hour goes stale in an afternoon. The rain gauge by the fence is fresh at sunrise and old by next Saturday. The keeper knew each of his instruments and how often it spoke. A good dashboard has to know the same.
The rule does not throw the old reading away. Knowing the last measured temperature was forty-one at midnight is worth something at six in the morning, as long as you know that is what it is. The STALE flag keeps the past from pretending to be the present. That is all the keeper’s timestamps ever did.
The old lighthouse logs survive in archives, and part of why they are worth reading is that the keepers wrote down the dull nights too. A night where nothing happened, written down as a night where nothing happened. That is the logbook form of the empty rain gauge: a record that someone was watching, and saw nothing, and said so.
The dashboard had no way to say that. When a source went quiet, it either held the old number or filled in a default. It had no word for I was watching and nothing came, and no word for I was not able to watch. Those are two different nights, and a keeper would never confuse them. Now the dashboard has words for both.
Chapter 8: The Empty Sky
The last liar was the quietest one, and it was up on the roof.
The household keeps a small radio that listens to airplanes. Most aircraft broadcast where they are, how high, how fast, and which way they are headed, and a modest antenna and receiver can hear them from a good many miles off. An airspace panel on the dashboard shows a count of aircraft in range and who they are. On a clear evening you can step out on the porch and put a name to the flashing lights high overhead.
This week the panel said zero aircraft.
Zero can be a true count. There are quiet hours when nothing flies within range. An empty sky is a real thing, like an empty rain gauge.
But the panel had said zero for a while. And when the steward went to look, the reason was plain: the antenna was disconnected. The receiver was not hearing an empty sky. It was not hearing anything at all. It was deaf, and it was reporting its deafness as silence.
Of all the false readings in the house, this one draws the line most clearly. A gauge that reads zero and a gauge that is broken look identical from across the room. The panel used the same way to say “I hear no aircraft” and “I cannot hear.” A watchman asleep at his post and a watchman awake who saw no one will both tell you in the morning that nobody came. Only one of them is telling you anything.
Heimdall is the watchman of the gods. He sits at the end of the rainbow bridge, and the lore says he needs less sleep than a bird, can see a hundred leagues by night as well as by day, and can hear the grass growing on the earth and the wool growing on the sheep. The whole worth of Heimdall is that when he is silent, his silence means he watched and saw nothing cross. If he could sleep at his post and still be believed, the bridge would be no safer than an open field.
The panel now knows the difference. When the receiver hears the sky and the sky is empty, it says zero, and zero means zero. When the receiver gets no signal at all, it says NO RX, no reception, so nobody looks at the screen and thinks the sky is clear when the house has gone deaf. The antenna still needs mending. But even before it is fixed, the panel is honest.
That is the dignity of “I don’t know.”
We live in a time that does not much honor that phrase. Machines are built to always show something, because a blank screen looks broken and a number looks competent. The whole pressure of the world pushes toward filling the blank: with a guess, a default, yesterday’s value, a comfortable number nobody will question.
The old poem the Norse called the Havamal, the sayings of the High One, does not flatter anyone. Much of it is advice about speech, and much of that is some version of: do not claim more than you know. It notes that the unwise man thinks he knows everything while he sits safe in his corner, but has no answer when he is questioned. It says the foolish man does better to keep quiet among people, since nobody will learn how little he knows unless he talks too much. The wisdom it prizes knows where its own edges are.
That is Ansuz, the rune of Odin and of the spoken word; the Old English rune poem calls it the mouth, the source of all speech. And Odin, who hung on the windy tree nine nights to win the runes, who gave an eye at Mimir’s well for a drink of wisdom, is not a god who pretends to know things. He pays dearly to find things out, because he knows exactly how much he does not know. A god who would give an eye to learn the truth has no use for a painted number.
“I don’t know” is not a failure. It is a measurement of the edge of what you know. It is the most honest reading an instrument can give when it cannot give any other, and it is the most useful one, because it tells the one who reads it exactly where to go and look. A blank on a dashboard is a finger pointing at the fence post: go out there, walk around it, see whether the gauge is cracked. A painted number is a hand over the eyes.
Chapter 9: The Decision on the Fake Number
It would be easy to hear all this as fussiness. It is a hobby, a basement computer and some radios on fence posts.
But nobody builds a dashboard to admire it. You build it to make decisions. You look at the air quality card to decide whether to open the windows and work in the garden, or keep the house shut. You look at the temperature card to decide whether tonight is the night to cover the last tender things and drain the hose. You look at the radio history to decide where to hang the next relay before the ground freezes and ladder work gets dangerous.
Every one of those decisions was being made, some of the time, on a number that was invented.
That is what a plausible lie does. It does not merely fail to help. It pushes you, confidently, in whatever direction the invented number happens to point. A blank would have made you go look. A crash would have made you go fix. The lie made you do nothing, or do the wrong thing, and feel sure about it.
A plausible lie is worse than a blank. That is the sentence the week came down to. A blank is honest ignorance. It knows it is ignorant and it says so, and the one who reads it knows to be careful. A plausible lie is dishonest knowledge. It claims a certainty it has not earned, and the one who reads it has no reason to be careful at all.
And a crash announces itself. It is the most honest failure there is. Everyone in the old trades understood this about their tools. A cracked axe handle that shatters on the first swing is annoying. An axe handle that has a hidden crack and holds just fine until the day you are up on a ladder is how people lose fingers. The loud failure is a gift. The quiet one is the one that hurts you.
So when in doubt, fail loud rather than quiet. Show the blank, not the guess. Say NO RX, not zero. Say “command failed,” not “success.” Mark STALE rather than pretend to be fresh. None of that makes a system look better. All of it makes a system worth trusting. Catch a system in one lie and you can no longer believe any of it, and you are back to walking out to the fence post for everything.
Which, it should be said, is not the worst place to end up. The fence post never lies. You only have to look at it properly.
The Troth of the Measure
What the week wrote down, in the end, came to seven plain rules.
- Every number is measured, or pulled from a real source and stored, or calculated from those. There is no fourth kind.
- If it is missing, show two dashes. Never a zero that means unknown. Never a default. Never a guess.
- A zero is a claim. Only an instrument that is working may say zero. A broken one says it is broken.
- Stale data carries its age, and a STALE flag. The past may speak, but it may not pretend to be the present.
- Name a thing for what it truly measures. A page fetch is not a ping. Power is not a battery level. Rename down if you must.
- A button says success only when it has checked. Otherwise it says failed, or it says it does not know.
- Clean the ledger with a fair copy first. Back up the whole book, then blank the false entries where anyone can see they were struck.
And under all seven, the sentence that holds them up: a plausible lie is worse than a blank.
What Still Sits on the Scale
Not every board in the house has been knocked on yet. The worst of the painted numbers are gone, but a system that grew up over years has corners that were not visited this week. Some will turn out to be fine. Some will not.
The antenna on the roof needs going up to, on a dry day before the ladder gets icy, and until then the airspace panel will go on saying NO RX, which is exactly what it should say.
The fair copies of the old ledger are kept, and a day will come to decide how long to keep them. Not this week.
And there is a longer question, the one that will not be settled with a few lines of code: how to build every new thing on the ridge from the start so that it knows the difference between measuring nothing and being unable to measure. That is a habit, not a fix. Habits are built in the dark half of the year, slowly, at the bench, while the snow comes down.
The Blessing
Hail Tyr, who put his hand in the wolf’s mouth and paid in full for a pledge he knew was false, and never pretended otherwise. Keep our measures true, and let us pay our own debts.
Hail Odin, who gave an eye to know and hung nine nights to learn, and who has no use for a comfortable answer. Teach us the edge of what we know.
Hail Heimdall, the watchman who does not sleep, whose silence means that nothing crossed. Let our silences mean something too.
Hail to the land-wights of this ridge, who do not care what the screen says, and who were keeping the true measure of rain and frost and wind long before any of us came to count it.
And hail to the keepers, the lighthouse men and the rain gauge farmers and the market officers with their standard weights, and every plain person who ever wrote down nothing happened tonight because nothing did.
May every number on the ridge be measured, or honestly fetched, or honestly reckoned. May every blank be shown as a blank. May every zero come from an instrument that was awake. May the old readings carry their age, and no stale thing pass itself off as the present. May the buttons tell us when they failed. May the antenna come down from its silence, and the sky get its names back. May the frost be gentle on the last of the garden, and the gauge be empty for the right reason. May we have the courage to say I don’t know, and the wisdom to go and look.
The barns are what they are. Count them true. East is full. West is a third. North, we do not know yet.
Wes hal.
ᛏ ᚨ