Guest Post: The Limits Of Engineering

Tyler Durden's picture

Submitted by Charles Hugh Smith from Of Two Minds

The Limits Of Engineering

The notion that better engineering can solve all our problems is rooted in an ignorance of non-linear systems and selective/wishful thinking.

I recently received an email from a Chastizing Retired Engineer (CRE) that reflected both a ubiquitous spiritual disease and a profound ignorance of the limits of engineering. What fired the CRE's undimmed rage was my long analysis which explained why the economy isn't going to generate 20 million jobs, and that as a result, we need a Plan B, which I also outlined: Getting 20 Million Unemployed Back to Productive Work: Here's How (August 16, 2011).

The CRE began by criticising the length of the entry, and by expressing the heartfelt wish that I was poorly paid for my writing. What does it say about the state of the nation's spiritual and intellectual health when retired engineers start their critique by hoping those they disagreee with are subjected to poverty? The spiritual illness at the heart of this sentiment is a topic for another time.

We might also wonder what's to become of the nation if retired engineers only want to read analyses of 250 words or less--perhaps because that is enough for them to decide the writer should be banished to poverty for the sin of disagreeing with their views.

Though he obviously held his own intelligence and powers of observation in high regard, Mr. CRE apparently failed to notice that this blog is free, and thus I earn nothing directly from writing it. This further suggests a rush to judgment so spritely that the basic facts of the matter are skipped over as irrelevant.

Mr. CRE recommended that I turn my attention to telling General Motors how to engineer the Car of the Future (COTF) and similar projects, presumably as a way of generating employment.

In other words, the problem is a dearth of bloggers telling corporations how to design products of the future. That the solution to our collective problems is for bloggers to tell GM et al. how to design the Car of the Future is certainly a novel idea, but it overlooks several key points.

Is GM interested in the views of bloggers? Perhaps in terms of marketing their vehicles, but I doubt they are seeking engineering advice from bloggers. GM undoubtedly has many experienced, talented engineers, some of whom designed the Chevy Volt as The Car of the Future, or at least as a car aimed at the future.

The design parameters of the Volt were dictated by GM management, which is probably even less interested in the advice of bloggers than GM engineers.

There are a number of other assumptions of interest in Mr. CRE's recommendation. One is that the Car of the Future needs to be designed, and another is that the Car of the Future would go a long way toward solving our unemployment problems.

Some in GM reckoned they've already designed the Car of the Future with the Volt. Some at Nissan reckon the all-electric Leaf is the Car of the Future. (Both manufacturers sold about 2,200 of these models in the first 5 months of 2011.) Those working at Tesla clearly believe their next production car is the Car of the Future.

The engineers at Volkswagen have been working on their Car of the Future for years, the so-called 1-litre car that travels 100 kilometers on 1 litre of fuel --about 235 miles on a gallon of gasoline. The vehicle weighs about 1,600 pounds and is quite compact--severe trade-offs had to be made to reach this level of efficiency. (Side question: how many adult Americans could even get into the car?)

The latest model is the XL1 which achieves 260 miles per gallon and a top speed of 99 miles per hour with a two-cylinder turbo-diesel engine (47 horsepower) and electric-drive assist.

By comparison, my 1998 Honda Civic weighs about 2,400 pounds, has a 1.6 liter ICE (internal combustion engine) that generates 106 horsepower. On our recent camping trip, we drove 2,400 miles and despite climbing above 4,000 feet on five occasions and using the air conditioner, the car achieved 39.91 miles per gallon (total gasoline consumption 60.5 gallons). This is roughly comparable to the mileage of the far more costly and complex Toyota Prius, and about double the mileage of the U.S. fleet average of 19.8 MPG.

The point here is that behavior and common-sense engineering can double the efficiency of the existing fleet without any new fancy (and costly) technologies.

All of which is to say that there are already many competing designs for the Car of the Future, and it is the marketplace and external factors such as fuel cost and availability that will decide which wins, not retired engineers, executives at GM or bloggers. Clearly, the ICE (internal combustion engine) is intrinsically inefficient compared to an electric motor, as a huge percentage of the energy consumed by an ICE is lost to heat and friction of its many moving parts. Just as clearly, a Car of the Future that relies on composite materials and complex technologies will not be cheap to manufacture.

There are a few other assumptions implicit in Mr. CRE's view that are questionable. Will building the Car of the Future generate millions of jobs? If Mr. CRE has toured any advanced auto factories, he would note that most of the work is performed by software-controlled robotic machines. At some plants in Japan, the total workforce is about 500 people.

While there are certainly many jobs provided down the supply chain, the idea that building the Car of the Future will generate millions of jobs is simply fantasy. Software and robots are replacing human workers everywhere, even in low-wage nations like China.

Indeed, we can guess that Mr. CRE was not a software engineer, as he doesn't seem to grasp that software is munching through industry after industry, hollowing out entire swaths of advanced economies while creating far fewer jobs than it has eradicated. If this concept is new to you, I recommend reading Why Software Is Eating The World.

The larger question is how big a part the Car of the Future will play in the U.S. economy. Vehicle sales have plummeted from a record 17.4 million in 2000 to 11.5 million in 2010. Certainly the recession was a factor, but there are much deeper forces at work than recession: people are driving less, and more people are opting out of car ownership: The Road... Less Traveled: An Analysis of Vehicle Miles Traveled Trends in the U.S..

This is a global phenomenon. In Japan, the younger generation is not enamoured by auto ownership; rather, they view it as a costly burden.

In analytic terms, we can chart this topping out and decline of miles driven and vehicle sales on a S-curve, as described by Cesare Marchetti in his seminal paper, A Simple Model for Complex Systems.

Simply put, the entire vehicle industry is in decline for structural reasons beyond fuel costs. These include demographics, financial considerations, cultural values and other factors which cannot be adequately quantified by engineering methodologies.

If the structural dynamics I described in the entry which so annoyed Mr. CRE do in fact play out, then we can also ask who will be able to afford to buy the Car of the Future other than the top 10% of households.

If imported oil drops precipitously for a combination of non-linear and linear systemic reasons-- geoplitical upheavals, aging super-giant field production, rising domestic consumption in exporting states, etc.--then we might also profitably ask where the fuel will come from to power the 254 million passenger vehicles registered in the U.S.

In conclusion, we can conclude that Mr. CRE is remarkably ill-informed about the auto industry and the structural factors influencing employment, design, sales, fuel consumption and availability, etc. Mr. CRE's tirade reflects several other common afflictions:

1. A deep, pervasive nostalgia for 1965

2. Know-It-All Syndrome (KIAS), which is unfortunately endemic to a subset of the engineering community.

The nostalgia for the mid-1960s is understandable. Back in those heady days, all the government had to do was throw unlimited sums of money at big engineering projects like putting a human being on the moon, and the engineers went to work and fashioned miracles.

This same mindset spread throughout the government, including the Pentagon, where it led to the engineering analysis of how to win the war in Vietnam, which can be summarized thusly:

1. Calculate number of bombs needed to kill X number of bad guys

2. Estimate number of bad guys

3. Estimate number of U.S. troops needed to locate/corral bad guys

4. Station X number of U.S. troops in Vietnam and drop X tonnage of munitions on bad guys

5. War will be won when number of bad guys killed rises above replacement level

This is not an isolated example of engineering idiocy or KIAS. The entire premise of the engineering mindset is that problems can be broken down to a small set of quantifiable inputs, processes and outputs. This works fine when measuring and controlling water flow, flow of electrons, and other linear systems, but it is catastrophically mis-applied when Know-It-Alls besotted by their success in extremely limited linear systems attempt to "solve" non-linear problem-sets with linear "solutions."

Case in point: war is highly non-linear. The "Whiz Kids" at the Pentagon did not even understand the problem-set, or the nature of war; how could their simplistic, Know-It-All "solutions" possibly work in the real world?

Most of our problem-sets are non-linear, and are thus inaccessible to engineered solutions. Even something as straightforward as designing the Car of the Future turns out to be largely controlled by non-linear, impossible to quantify factors such as cultural values, future availablility of fuel, etc.

The very success of engineering in highly limited fields of endeavor feeds a hubris which is largely lacking in scientists schooled in life sciences. I wonder how much Mr. CRE knows about the human immune system, which is extremely complex, interactive, dynamic and non-linear in nature. I once pondered a chart of the basic immune system on the wall of a PhD immunologist friend; the number of "moving parts" and the interactions of those parts is mind-bogglingly large.

The human immune system cannot be usefully mapped as an engineering project. This may explain why life scientists and others who work in non-linear systems tend to be circumspect about the applicability of their expertise in other fields. Meanwhile, quants in economics brim with confidence in their cargo-cult models of non-linear human behavior.

No wonder the economists are so successful in "solving" our structural economic problem-sets.

If we consider the stock market a problem-set, then shouldn't it be possible to engineer 11 good trades in a row? After all, the data is all there for the taking. If a whiz kid could engineer 11 trades that doubled the capital invested--not that impossible when trading futures contracts or options--then in 11 iterations a mere $500 blossoms into $1 million.

So go ahead and engineer a "solution" to the stock market "problem" which yields 11 good trades in a row.

The problem is that the market--and most of life--is non-linear, and "solutions" cannot be conjured out of simplistic linear models and inputs which cannot be quantified except with a highly illusory accuracy.

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.
BaBaBouy's picture


Hopium Works Better ...

egdeh orez's picture

Market turns positive on no positive news.

Is this market broken or what?

topcallingtroll's picture

a market rising on bad news or no news is sometimes a sign of a nascent bull market.

Not that I want to see one now.  I am all out of stocks again. 

 I drank too much zero hedge koolaid I guess.   I hope I didn't pull a Leo by getting out at the wrong time.

eureka's picture

The trajectory of every single individual life - is, in terms of its inevitable outcome, namely death, 100%  linear and 100% predictable, regardless of the particular swirls and fluctuations each individual perceives in terms of quality, failure, success and happiness - and so, perhaps we should all prioritize the "happiness" that can arise in every moment from addressing it, each moment, to our maximum capacity for understanding, for applying ourselves, for sharing and caring - rather than to the perpetual, restless hunger for competitive advantage, which ultimately isolates us from the whole of life, from our fellowship and from community.

For all the chips we may win, none translate into eternity, but shared moments do - and allow us to say "I have lived and loved and my life was full and I am grateful for all those moments, which made my life bigger than I could ever have engineered it myself.

Be the competitive warrior, gambler and count your chips every day - or - be the lover, giver and receive the grace of all things beyond yourself and your constructs. It is your choice - and a choice for eternity.

RockyRacoon's picture

 I drank too much zero hedge koolaid I guess.   I hope I didn't pull a Leo by getting out at the wrong time.

Don't worry about it -- worry being the point.   Too little emphasis is placed on peace of mind and enjoying a great night's sleep.   That is a dividend, a return, if you will.   Count that into your technical analysis of your decisions.

Oh regional Indian's picture

There are no limits to engineering. Limits are imposed to make engineering "work" within the current dysfunction of it's own creation.


We are living the madness of Teutonic linearity. 

There are other ways, but they require a change so deep that most shrink at the thought. The destruction of one's dominant paradigm is too frightening for most.

Cycles, Waves, spirals and ellipses. 

Order in nature is implicit, not explicit. Ponder that if you will.


LawsofPhysics's picture

Come on ORI, you are asking people to understand exponential functions and metabolic as well as energy flux.  Just too tall an order I think.

Oh regional Indian's picture

Well LawsofP, if that is what we live, then it behooves us to understand it, ne?
Not too tall an order to study our own bodies without the overlay of science. We can heal without incision. Nature works best if we understand cause and effect at a higher degree than we currently do.

We are living a Newtonian dystopia in a decidedly non-newtonian world.


topcallingtroll's picture

exponential functions are linear functions, you alleged law of physics.

A non linear function has two or more simultaneous y values for each value of x.

And there are lots of linear functions in nature, or at least close enough approximations we can land on the moon.l

exiledbear's picture

Perhaps you're referring to linear differential equations? Exponential functions are NOT linear.

Linearity is defined very precisely as follows:

f(Ax) = Af(x)

f(x + y) = f(x) + f(y)

Neither of which actually hold up if you start using exponential functions (or even polynomials).

exp(Ax) does not equal Aexp(x)

exp(x + y) does not equal exp(x) + exp(y)

You can linearize exp(x) for small values of x, and this trick is done all the time in the sciences.

exp(x) = 1 + x, for small values of x

This is why using a system that relies on exponential growth is so dangerous in the real world, because at the start, everything LOOKS linear, but it's not. It's only when you get to the stage we're in that the higher order polynomials start kicking in and then it all blows up.


DCFusor's picture

Yup, finally someone who knows what they are talking about.  Non linear isn't the same as single valued, nor is it necessarily non-monotonic.

Non linear problems are not insoluble, it's just that there is no P-complete rote solution to most of them.  Much more often the situation is utterly NP-complete and stochastic methodology has to be used to approximate an answer, and which can't guarantee the best anser in a multi-dimensional multi-zero error space is found.

Back in the day, while writing a book about digital signal processing, a real smart college kid and I devised an algo that would find best-fit coefficients to IIR filters to match an arbitrary frequency response (think speech bandwidth compression codecs for a use -- our code might be in your phone).  It was one heck of a job, but we did it and used everything from simulated annealing to genetic algs along the way, before we found a trapdoor for our case using MLFN neural networks, themselves rather computationally intensive to train.  Helps to know what you're trying to accomplish up front though, and you won't get that by trying to fudge incorrect linear models onto reality.  No good map there.

I'd have to believe that such tech would wipe the floor with most of the semi-linear plus ruleset algos out there, but haven't had time and energy to build up a robot of my own.  There's a heck o a lot of plain grunt work involved to hit the API of most of the brokers that makes my head hurt.  Funny how the simple part seems the hardest to a certain mindset.

RockyRacoon's picture

Haven't the foggiest what you said, but I did get one good piece of advice out the comment:

Helps to know what you're trying to accomplish up front though...

Good advice in any endeavor, and so seldom seen.

Flakmeister's picture

It is a little more subtle than that... Semantics is at play here.

Classical Electrodynamics is a linear theory, i.e. principle of superposition, as is Newtonian gravity.

Einsteinian gravity is a non-linear theory....

Non-linear functions as you describe (multivalued) are different from non-linear systems...

Anyway... another damn fine article from CHS

Boop's picture

A "function" which has two or more simultaneous y values for, lets say, a value of x is not a function. I'd say you were referring to continuous functions, but your description is of something which is not a function.

Engineering deals with non-linearity all the time, despite the author's imaginings.

Chaotic systems, discontinuous systems, and particularly adaptive systems (by which I mean systems which change with time) like the stock market are much harder to control or predict.

Hell, trying to calculate the motion of 3 bodies under Newton's law of gravity seems likely to be simple enough - but in general it's not possible.

Flakmeister's picture

The three body problem is a bitch isn't it....

AnAnonymous's picture

Engineering deals with non-linearity all the time, despite the author's imaginings.


No. Engineering deals with linear approximations of non linear problems. The key has been to linearize 'everything' as solutions to linear problems are more understood and known.

AnAnonymous's picture

exponential functions are linear functions,


I have a strong heart, thank you. Nice try to give people a heart attack though.

US citizens know no limit. Exponential functions linear functions?

We need somebody to be able to monetize efficiently US citizens natural trend to cheap propaganda. It will solve every debt issue.

Exponential functions, linear functions, good one. Typical to the US world order.

Flakmeister's picture

Be careful, linear means many different things....

Non-linear usually means functions of the form  dy/dt = f(x,y)

The exponential function is non-linear from additivity and homogeneity, but it is a solution to many linear Diffy Qs..

Dr. Acula's picture

The mathematical fetishists here miss the point. The problem in economics is more fundamental:

"Here we are faced with one of the main differences between physics and chemistry on the one hand and the sciences of human action on the other. In the realm of physical and chemical events there exist (or, at least, it is generally assumed that there exist) constant relations between magnitudes, and man is capable of discovering these constants with a reasonable degree of precision by means of laboratory experiments. No such constant relations exist in the field of human action outside of physical and chemical technology and therapeutics." - Ludwig von Mises, Human Action p. 55


"Deluded by the idea that the sciences of human action must ape the technique of the natural sciences, hosts of authors are intent upon a quantification of economics. They think that economics ought to imitate chemistry, which progressed from a qualitative to a quantitative state.1 Their motto is the positivistic maxim: Science is measurement. Supported by rich funds, they are busy reprinting and rearranging statistical data provided by governments, by trade associations, and by corporations and other enterprises. They try to compute the arithmetical relations among various of these data and thus to determine what they call, by analogy with the natural sciences, correlations and functions. They fail to realize that in the field of human action statistics is always history and that the alleged "correlations" and "functions" do not describe anything else than what happened at a definite instant of time in a definite geographical area as the outcome of the actions of a definite number of people.2 As a method of economic analysis econometrics is a childish play with figures that does not contribute anything to the elucidation of the problems of economic reality." - Ludwig von Mises, The Ultimate Foundation of Economic Science p.63


Flakmeister's picture

Given that Smith's "invisible hand" is an unknown non-linear DE, as was Keynes "animal spirits", is it any bloody suprise that Economics is basically a light-weight social science?

Dr. Acula's picture

>Given that Smith's "invisible hand" is an unknown non-linear DE

If it's admittedly unknown, then why do you think one exists?

Note that your belief is unfalsifiable - and unscientific.

>is it any bloody suprise that Economics is basically a light-weight social science?

I am not sure what you mean by "light-weight". Praxeology is currently the best tool available for understanding economics.

I could say that physics is light-weight: although the human brain is an element of the natural world, physics is utterly incapable of coping with teleological human action. Physics is powerless to explain why a human being walks into a McDonalds. 

Flakmeister's picture

Well... the goal of physics is not to understand the human brain, but you can be damn sure that whatever theory is found that the law of physics will be respected...

You might find some work by Penrose to be of interest....

And Mises was a sharp guy, but he wasn't the second coming....

RockyRacoon's picture

How does the ghost in the machine figger into them thar cakalashuns?

Flakmeister's picture

New avatar?? Or is your brother-in-law filling in?

RockyRacoon's picture

This is my sterner, more determined visage.

Flakmeister's picture


    Physics is the modelling of the universe via mathematics while being guided by experiment....Many deep principles may be expressed in words, but their power lies in the constraints on the Mathematics..... 

I'll bet my left nut that a quantitative model of the "Invisible Hand" will be a non-linear DE....  

Dr. Acula's picture

>I'll bet my left nut that a quantitative model of the "Invisible Hand" will be a non-linear DE....  

Your belief is unfalsifiable through experimentation. After all, if you come up with any equation, when it fails catastrophically, then you can just say it wasn't the right one.

And your belief is logically dubious. If people had such an equation and made predictions, they could intentionally violate the predictions.

But keep believing in chimeras. There's no law against stupid.


FeralSerf's picture

You seem to be suggesting that there is no cause and effect.

Just because the current practitioners of the physical sciences are unable to adequately describe these phenomena does not mean that it's impossible to do so.  Maybe you just need to wait for the fullness of time?  Got immortality?

StychoKiller's picture

Almost all of the time, MOST Humans do NOT know their own desires, much less how any market is supposed to fulfill them!

Roughly 7 Billion enigmatic equations with an unknown number of variables, AND an unknown number of solutions...

Truly, a "Fatal Conceit!"

Flakmeister's picture

You've taken this somewhere else....

No one is saying that a model predicts, in a deterministic sense, the behaviour of an individual.. Now can apply such an equation to an ensemble and predict average behavior? Yes, I think its possible. Useful, I doubt it.

Much like the behaviour of an individual gas molecule...

Dr. Acula's picture

>Much like the behaviour of an individual gas molecule...


Even when the problem is actually about gas molecules, the clowns on TV can't predict the weather 24 hours from now.

And I don't think Warren Buffet, Ben Bernanke, and Steve Jobs are each like an "individual gas molecule".

I think you might enjoy this video on marketing actually: 


AnAnonymous's picture

although the human brain is an element of the natural world, physics is utterly incapable of coping with teleological human action. Physics is powerless to explain why a human being walks into a McDonalds.


Dead bodies are also elements of the natural world.

The human brain might be an element of the natural world. It is not though when it commands the move to a shop. As it is not when it is used to comprehend why people walk into shops.

The natural world is not the cultural world. The distinction is only 2000 years old. So one could guess that US propagandists still need time to come with better propaganda about what the natural world.

AnAnonymous's picture

Non linear solution to linear problems exist. Being a solution to a linear problem (a lot of time linearized by the way) does not exclude being non linear.

snowball777's picture

All hail CHS and the know-nothing mindset!

So VW, a German company, has been working on a COTF for years? And what's their unemployment in Germany again? And their GDP for 2011? don't say.

FeralSerf's picture

Germany exports her unemployment.  This is the reason that the "PIIGS" have no money to pay their bills.  They've been underemployed for too long and have gone into debt to support German employment, i.e. German manufactured goods.  America is in the process of doing the same thing, first with Germany and Japan and now with China.  The end result will be the same.

We have met the PIIGS and they are us.

snowball777's picture

You really think Greece would be an economic powerhouse that competes on equal footing with Germany, if they were on the Drachma? Seriously?!

Think about how many VWs are assembled in Mexico and Tennessee for a second.


FeralSerf's picture

You have a reading disability.  I never said or implied that.  Why are so many Americans illiterate?

Your example of the Mexican and  Tennessean VW assembly plants is not an apt one.  Neither Mexico nor the US are in the EC or use the euro and both of them, in contrast to Greece, have import restrictions on German autos.

eureka's picture

In the final analysis and outcome, neither machines nor capital need human beings, except for another short while, as consumers. Capital is printed & leveraged out of thin air. Software/AI exponentially run all machines.

Consequently, all nationalist - or economical class - competition, is ultimately irrelevant.

The real question is: what will be the function of human beings in the future?

Or rather, what will we choose it to be - if we address the question before it is to late?

FeralSerf's picture

What is the function of human beings today?  Some claim their function is to satisfy the ego of a supernatural being.  Observation suggests their only function is to breed and make more human beings, to "go forth and multiply".  There appears to be limits to this function.

We have met the pond scum and he is us.

RockyRacoon's picture

Some talking head today was in Spain or Italy or one of those furren countries -- said that there wouldn't be riots 'cause it was too hot.    Too hot.   What has the unruly world come to.   Too hot?

FeralSerf's picture

Global warming has made revolution without airconditioning obsolete.

LawsofPhysics's picture

I'll be the first to admit that there is nothing worse than a technocrat on hopium that has a few facts, but limted experience in the application of said facts.  Knowledge versus wisdom.

NotApplicable's picture



Great article. Linear vs. non-linear systems do not get the attention they deserve, IMO. Of course, this might be due to all of the doers out there who insist upon marching, but knowing not where.

Azannoth's picture

There's only 1 equation every (smart) person is(should be) looking at, how to I maximize my personal gains  vs. everybody else, everything else is just logic noise

snowball777's picture

Your zero-sum vision of the world is a great strategy, for an insecure weakling.

topcallingtroll's picture

I thought we were all Austrians here.  Azannon is giving the formula for a successful robust society.

Economics and maximizing personal utilities is not a zero sum game by the way.  No economic philosophy believes that, not even Keynesians

Mark_BC's picture

No, we are all NOT Keynesians here. Common enemies make for strange bedfellows.

And economics is indeed a zero sum game. All the wealth we have realized over our history has been taken at the expense of the natural world. It's a pie chart.

TheAkashicRecord's picture

//All the wealth we have realized over our history has been taken at the expense of the natural world.//

Thank you for stating the obvious (I am being sincere, not sarcastic here), this point has to be reiterated and hammered into the heads of people; people think labor produces things, lol, nothing is produced, just transformed. Silly humans -forgetting about thermodynamics.