Chapter 37 7.3 The Flush Toilet: A Prototype of Confession Logic

In order to concoct more accurate shooting tables, a group of mathematicians from the Human Computing Laboratory were conscripted to the Aberdeen Proving Ground during World War I, and among these conscripted mathematicians, few were like Private Norbert Vickers. Nana has qualifications far beyond the standard.The former math prodigy had a gift for heresy. In the eyes of the ancients, genius was supposed to be something given rather than created.But the United States at the turn of the century was the place where conventional wisdom was successfully overturned.Norbert Wiener's father, Leo Wiener, came to America to start a vegetarian group.He turned out to be troubled by other unconventional problems, such as the reformation of God. In 1895, Leo Wiener, a professor of Slavic languages ​​at Harvard University, decided that his firstborn would be a genius.It is a deliberate genius, not a natural genius.

Thus, Norbert Wiener was born with high expectations.He was able to read at the age of 3 and received a doctorate from Harvard at the age of 18.At the age of 19, he began to study metamathematics with Russell. At 30, he was a math professor at MIT and a complete freak.Small, stocky, with splayed feet, a goatee, and a cigar in his mouth, he waddled around like a clever duck.He has a legendary ability to learn while asleep.More than one eyewitness has reported that Wiener fell asleep in the middle of a meeting, woke up suddenly when someone mentioned his name, commented on conversations he missed while dozing off, and Often presents some penetrating insights that leave others dumbfounded.

In 1948, he published a book on the philosophy and feasibility of machine learning for the layperson. (For various circumstantial reasons) This book was originally published by a French publishing house, and in the first 6 months it was printed in 4 editions in the United States, and 21,000 copies were sold in the first ten years— It was a bestseller at the time.Its success can be compared with the "Kinsey Report" released in the same year, which focused on sexual behavior. A BusinessWeek reporter wrote in 1949: "In one respect, Wiener's book is like The Kinsey Report: the public reaction to it is as significant as the book itself. .”

Although not many people can understand this book, Wiener's deafening ideas have entered the public mind.The reason is that he gave his idea and his book that wonderful, colorful name: cybernetics.As many writers have pointed out, the word cybernetics comes from the Greek word for "helmsman"—the one who controls the ship.Wiener studied servos during World War II and was struck by its mysterious ability to assist all types of steering gear.It is not usually mentioned, though, that in ancient Greek the word was also used to refer to the ruler of a state.According to Plato, Socrates once said, "The helmsman/governor can save our souls in great danger, save our bodies, and save our material possessions." different meanings.Governance (for the Greeks, self-government) was the creation of order through the struggle against chaos.Likewise, man needs to steer a ship to keep it from sinking.And after this Greek word was misused by Latin as kubernetes, it derived governor (governor, regulator), and Watt used it to mark his flying ball regulator that plays a controlling role.

For French speakers, this administrative word has an earlier antecedent.Unbeknownst to Wiener, he is not the first modern scientist to revive the word.Around 1830, the French physicist Ampere (the ampere, the unit we use to measure electricity, and the abbreviation "Ampere", followed his name) followed the tradition of the great French scientists and designed a system for human knowledge. Sophisticated classification system.Among them, Ampere defined a sub-discipline called "understanding science", and political science is a sub-discipline under this branch.In political science, under the subgenus of diplomacy, Ampere is included in the discipline of cybernetics, the doctrine of governance.

However, the definition in Weiner's mind is more specific.He stated this definition prominently in the title of that book: Cybernetics: The Science of Control and Communication in Animals and Machines.As Wiener's rough ideas on cybernetics were gradually concretized by later computers and supplemented and enriched by later theorists, cybernetics gradually had the meaning of governance that Ampere said, but the meaning of politics was removed. The effect of Wiener's book has been to permeate almost every aspect of technical culture with the idea of ​​feedback.Although this core idea is not only old but commonplace in some special cases, Wiener gave it legs and axiomaticized it: realistic self-control is a simple technical task.When the idea of ​​feedback control is perfectly combined with the flexibility of electronic circuits, they combine into a tool that anyone can use.Just one or two years after the publication of "Cybernetics", electronic control circuits set off a revolution in the industrial field.

The avalanche effects of the use of automatic control in commodity production are not always so obvious.On the shop floor, the automatic control lived up to expectations, with the aforementioned ability to tame high energy.At the same time, the overall speed of production is also increased due to the inherent continuity of automatic control.But these are relatively minor compared to the unexpected miracle of self-control circuits, their ability to pick the best from the rough. To illustrate how basic circuits can produce precision from inaccurate parts, I follow the example presented in French author Pierre Latier's 1956 book Thinking with a Machine.Before 1948, generations of technicians in the steel industry tried and failed to produce sheets of uniform thickness.They found that there are no less than six or seven factors that affect the thickness of the steel plate rolled out of the rolling mill-such as the speed of the roll, the temperature of the steel, and the traction force on the steel plate.They spent many years painstakingly adjusting one by one, and then spent more time in sync, and nothing worked.Controlling one factor can inadvertently affect other factors.Slowing down increases the temperature; decreasing the temperature increases the pull; increasing the pull decreases the speed, and so on and on.All factors are interacting.The entire control process is surrounded by an interdependent network.Therefore, when the rolled steel plate is too thick or too thin, it is a waste of energy to track down the culprit among the 6 interrelated suspects.Before Wiener's "Cybernetics" came up with his brilliant idea of ​​generalization, the problem was stuck there.Immediately after the book was published, engineers all over the world grasped the key ideas, and within a year or two they installed electronic feedback facilities in their factories.

In practice, a thickness gauge is used to measure the thickness of the newly rolled sheet metal (output), and this signal is then sent back to a servo motor that controls the variable tension force, which maintains its influence on the steel until it enters the rolls .With such a simple single loop, the whole process is straightened out.Because all factors are interrelated, as long as you control one of the factors that directly affects the thickness of the product, then you are equivalent to indirectly controlling all the factors.Whether the tendency to deviate is from uneven metal stock, worn rolls, or undue high temperatures is of little importance.It is important that this automatic loop is adjusted so that the last variable compensates for the other variables.If there is enough leeway (and there is) to adjust the tension to compensate for over-thick or poorly heat-treated raw metal and the deviations caused by the rolls being mixed with iron filings, the end result will be a plate of uniform thickness.Although each factor interferes with the others, the continuous and almost instantaneous nature of the loop leads the unfathomable web of relationships between these factors toward a stable target, a stable thickness. .

This cybernetic principle that the engineers discovered was a general one: If all the variables are closely related, and if you can really control one of them to the maximum, then you can indirectly control all of the others.This principle is based on the integrity of the system.As Lattier writes, "The regulator is not concerned with the cause; its job is to detect fluctuations and correct them. Errors may arise from factors whose effects are hitherto unknown, or from There has never been an element of doubt." How and when a system reaches consensus is beyond human knowledge, and more importantly, there is no need to know.

The irony of this breakthrough — this feedback loop — is technically quite simple, Latier said, and “it could have been approached with a more open mind. It was brought in 15 or 20 years in advance...".What is even more ironic is that the openness to adopt this idea was actually established in economics circles 20 years ago.Attempts to trace feedback paths in complex networks have been dissected by Friedrich Hayek and the influential Austrian school of economics, and concluded that such efforts are futile.Their argument was known at the time as the "computational argument".In a command economy, such as the embryonic top-down economic system established by Lenin in Russia, resources are allocated through calculation, trade-offs, and control of communication channels.And computing multiple feedback factors across distributed nodes in an economy, even with less-than-strong controls, is as unlikely to succeed as an engineer tracking down those cunning, interconnected factors in a steel mill.In a swinging economy, it is impossible to calculate the allocation of resources.In contrast, Hayek and other Austrian economists argued in the 1920s that a single variable—price—could be used to moderate all other resource allocation variables.According to this theory, people don't care how many bars of soap each person needs, and don't care whether they should cut down trees for houses or books.These calculations are performed in parallel, on the go, bottom-up, out of human control, and spontaneously by the interconnected network.Order will emerge spontaneously.

As a result of this automatic control (or lack of human control), engineers can finally relax their tense nerves and no longer worry about the uniform specifications of raw materials and the perfect adjustment of processes.So they can start working with imperfect materials and imprecise processes.Let the self-correcting nature of automated processes optimize to release only high-quality products.Or, input raw materials of uniform quality, set the feedback loop to a higher quality level, and provide the next process with higher-precision products.The same idea can be applied upstream to raw material suppliers, who can also use a similar automated loop to select higher quality products.If this concept runs through the upstream and downstream of the entire industrial chain, then the automated self will become a quality management machine overnight, and human beings who always manage to improve precision can effortlessly extract materials from materials. got. The introduction of Eli Whitney's interchangeable standard parts and Ford's assembly line concept have fundamentally changed the way production is done.However, these improvements require extensive equipment upgrades, capital investment, and are not applicable everywhere.On the other hand, domestic automatic circuits--a suspiciously cheap accessory--can be ported to almost any proprietary machine.It's like an ugly duckling, after printing, it becomes an elegant goose and lays golden eggs. Not every automatic circuit, though, can produce the sure-fire immediacy that Bill Bowles' barrel possesses.In a chain of loops, each additional loop increases the possibility that a signal roaming through the larger loop, returning to its starting point, finds that something happened earlier in it. Fundamental changes have taken place while still wandering in the loop.Especially for large networks with rapidly changing environments, the fraction of a second required to traverse the entire line may be greater than the time required for the environment to change.And in response, the last node tends to issue larger corrections to compensate.However, such a compensatory command will also be delayed due to too many nodes that need to be traversed, so it also misses the movement mark when it arrives, and another unreasonable correction occurs.This is the same reason that novices always drive in zigzags, because every time the direction is corrected, it will always be overcorrected, exceeding the previous overreaction.This goes on and on until the novice learns to tighten the entire feedback loop to make smaller, faster responses, otherwise he will inevitably (futilely) change direction on the highway looking for the center line.This is why simple automatic lines died.It tends to go into a "swing" or "twitch" state, that is, nervously oscillating from one overreaction to the next, trying to find stability.There are a thousand ways to counteract this tendency to overcompensate, each of which has been realized by a thousand more advanced circuits that have been invented.Over the past forty years, engineers with degrees in control theory have filled bookshelves after books of papers communicating the latest solutions to newly discovered oscillatory feedback problems.Fortunately, feedback loops can be incorporated into useful configurations. Let's take the control prototype prototype of a flush toilet as an example.By attaching a handle to it, we can adjust the height of the water line in the tank.The self-regulating mechanism in the water tank will then adjust the water to the height we set.Flip it down and the self-regulating mechanism will stay at a satisfactorily low level, and flip it up and it will let water in to reach a high level. (Modern toilets actually have this kind of handle.) Now let's go a little further and add a self-regulating circuit to pull the handle.In this way, we can even let go of this part of the work.The job of this second loop is to find the object for the first loop.Let's put it this way, the second mechanism will move the handle when it senses the water pressure of the water inlet pipe. If the water pressure is high, it will set a high water level for the water tank, and if the water pressure is low, it will set a low water level for it. The second loop controls the fluctuation range of the first loop, which controls the water.In an abstract sense, the second loop gives a second level of control: control over control, or meta-control.And with this meta control, our new secondary toilet behaves "on purpose".It can be adjusted in response to changes in goals.Although the second circuit that does the goal setting for the first is equally a mechanical thing, the fact that the whole mechanism itself does choose its own goals gives this metacircuit a certain biological feel. It is such a simple feedback loop that can be stitched together in an endless process of integration, working together forever, until it forms a tower of sub-goals of the most incredible complexity and intricacy.These loop towers will continue to surprise us, because the signals flowing along them will inevitably cross their paths with each other. A causes B, B causes C, and C causes A.In a straightforward paradoxical form: A is both a cause and an effect.Cybernetic expert Heinz von Foster called this elusive loop "circular causation."Early AI authority Warren McCloch called it "non-transitive prioritization," meaning that prioritization would be endlessly self-referential like a child's rock-paper-scissors game. The way intersects itself: the cloth can wrap the stone, the stone can collapse the scissors, and the scissors can cut the cloth, and the cycle is endless.Hackers call this situation a recursive loop.Whatever the name of this enigma, it dealt a blow to 3,000-year-old philosophy of logic.It shakes up everything traditional.If there is something that is both cause and effect, isn't so-called rationality within easy reach of anyone?