The Security Value of Inefficiency

For decades, we have prized efficiency in our economy. We strive for it. We reward it. In normal times, that’s a good thing. Running just at the margins is efficient. A single just-in-time global supply chain is efficient. Consolidation is efficient. And that’s all profitable. Inefficiency, on the other hand, is waste. Extra inventory is inefficient. Overcapacity is inefficient. Using many small suppliers is inefficient. Inefficiency is unprofitable.

But inefficiency is essential security, as the COVID-19 pandemic is teaching us. All of the overcapacity that has been squeezed out of our healthcare system; we now wish we had it. All of the redundancy in our food production that has been consolidated away; we want that, too. We need our old, local supply chains—not the single global ones that are so fragile in this crisis. And we want our local restaurants and businesses to survive, not just the national chains.

We have lost much inefficiency to the market in the past few decades. Investors have become very good at noticing any fat in every system and swooping down to monetize those redundant assets. The winner-take-all mentality that has permeated so many industries squeezes any inefficiencies out of the system.

This drive for efficiency leads to brittle systems that function properly when everything is normal but break under stress. And when they break, everyone suffers. The less fortunate suffer and die. The more fortunate are merely hurt, and perhaps lose their freedoms or their future. But even the extremely fortunate suffer—maybe not in the short term, but in the long term from the constriction of the rest of society.

Efficient systems have limited ability to deal with system-wide economic shocks. Those shocks are coming with increased frequency. They’re caused by global pandemics, yes, but also by climate change, by financial crises, by political crises. If we want to be secure against these crises and more, we need to add inefficiency back into our systems.

I don’t simply mean that we need to make our food production, or healthcare system, or supply chains sloppy and wasteful. We need a certain kind of inefficiency, and it depends on the system in question. Sometimes we need redundancy. Sometimes we need diversity. Sometimes we need overcapacity.

The market isn’t going to supply any of these things, least of all in a strategic capacity that will result in resilience. What’s necessary to make any of this work is regulation.

First, we need to enforce antitrust laws. Our meat supply chain is brittle because there are limited numbers of massive meatpacking plants—now disease factories—rather than lots of smaller slaughterhouses. Our retail supply chain is brittle because a few national companies and websites dominate. We need multiple companies offering alternatives to a single product or service. We need more competition, more niche players. We need more local companies, more domestic corporate players, and diversity in our international suppliers. Competition provides all of that, while monopolies suck that out of the system.

The second thing we need is specific regulations that require certain inefficiencies. This isn’t anything new. Every safety system we have is, to some extent, an inefficiency. This is true for fire escapes on buildings, lifeboats on cruise ships, and multiple ways to deploy the landing gear on aircraft. Not having any of those things would make the underlying systems more efficient, but also less safe. It’s also true for the internet itself, originally designed with extensive redundancy as a Cold War security measure.

With those two things in place, the market can work its magic to provide for these strategic inefficiencies as cheaply and as effectively as possible. As long as there are competitors who are vying with each other, and there aren’t competitors who can reduce the inefficiencies and undercut the competition, these inefficiencies just become part of the price of whatever we’re buying.

The government is the entity that steps in and enforces a level playing field instead of a race to the bottom. Smart regulation addresses the long-term need for security, and ensures it’s not continuously sacrificed to short-term considerations.

We have largely been content to ignore the long term and let Wall Street run our economy as efficiently as it can. That’s no longer sustainable. We need inefficiency—the right kind in the right way—to ensure our security. No, it’s not free. But it’s worth the cost.

This essay previously appeared in Quartz.

EDITED TO ADD (7/14): A related piece by Dan Geer.

Posted on July 2, 2020 at 9:26 AM52 Comments


Andy July 2, 2020 9:43 AM

efficiency is doing things right – effectiveness is doing the right things.
By focusing on eliminating inefficiencies we have lost sight of what those right things are. Rather than “blindly” bringing old inefficiencies back we have to have a very thorough look at why the existed in the first place to realize that – at least some of them – where not inefficiencies at all, but redundancies, fall backs and safety measures that are necessary for the system to be effective.

William July 2, 2020 9:51 AM

Government regulation generally makes it easier for larger businesses to dominate, since their scale allows them to afford compliance and monitoring costs. Regulations to “require inefficiency” might help medium-sized firms against giant firms, but could be harmful to small firms.

R. Brooks July 2, 2020 9:54 AM

Thank you very much for this perspective. My years as an office worker have shown me that “efficiency” is used as a weapon to push workers (and the existing system) to the breaking point so that they can be replaced either with massive software systems which do no better (and are sometimes worse) but are cheaper, or with outsourced workers who are paid less.

T452_TOM July 2, 2020 10:23 AM

well, that is a highly political and subjective point of view.
bottom line is the standard Progressive-Politics view of heavy government interventions in the general economy (e.g., antitrust & regulation) — such actions are actually extremely “inefficient” in theory and historical practice.
‘Government Inefficiency’ is obvious everywhere.

EFFICIENCY (output versus input) depends upon the “system” under analysis.

for example, research & development in any industry or economic endeavor is often inefficient in terms of dead ends and ‘wasted’ resources — but it’s critical to innovation and development of new more efficient processes and products.

Narrow views of “Efficiency” and disregard for unintended consequences can lead to great inefficiency in the broader systems involved.

Hawkins Dale July 2, 2020 10:41 AM

Consider voting. It’s much more efficient to use electronic voting machines. It’s even more efficient to allow online voting. And the vote counts are available instantly. Yay!

Compare to paper ballots, dropped into locked boxes, and counted in the presence of multiple humans. GROSSLY INEFFICIENT. It may take all night for vote counts to be reported. But the opportunities for mischief are greatly diminished. (And it’s usually by far the cheapest approach. No money for well-connected voting-machine vendors!)

Catherine July 2, 2020 10:43 AM

Amen: resilience is worth money. Lots of money. Let’s make sure we invest in it.

I recall reading and being convinced by many of these arguments in Joshua Cooper Ramo’s book The Age of the Unthinkable. From 2009, still very much worth a read.

Kurt Seifried July 2, 2020 11:06 AM

Fundamentally if I have $1 million dollars in extra capacity lying around (materials, extra machines, whatever) if I’m not using it, that money is literally being wasted. If I take that $1 million and invest it in growing my business I might end up with more (or pay it out to myself as a bonus, whatever).

Now I understand that someone might say “simple, you apply ALE (Annualized Loss Expectancy), a disruption in operations costs $11 million dollars and has a 10% chance of happening in a year so the cost is $1.1 Million, so spending $1 million to prevent it makes sense.”

How do we properly regulate or incentivize what is fundamentally at odds with how business owners want to operate (growth, paying out bonuses, etc.)?

The insurance industry have actuarial data and can cost risk, perhaps we should simply require companies to insure against this disruption, but again how do we define it/etc? In theory the premiums would be cheaper for companies that take corrective/preventative actions (ideally that are cheaper than paying higher insurance premiums).

The fundamental problems I see:

1) How do we measure or cost these risks/disruptions? What did the great toilet paper shortage of 2020 actually cost us? Do we rely upon the insurance industry to do it? This is more than one question I realize.

2) How do we assign limits/targets/goals? Costco must never run out of toilet paper? Never for more than 3 days? How do we define what is and is not critical (gasoline? Diesel? meat? toilet paper? which medicines?).

3) How do we then incentivize the behaviour we want and punish the behaviour we don’t want? Require insurance? Penalties for disruptions (so companies under strain get driven out of business, probably not what we want… right?).

Martin Ewing July 2, 2020 11:32 AM

Efficiency is generally a good thing if the objectives are right. Profitability is not enough. We need robustness against supply disruptions, political or medical. We need long-term reliability, etc.

We need these things, but we don’t have institutions to achieve them. Regulation is no use if the regulators are under resourced or if they don’t have political support.

uh, Mike July 2, 2020 11:37 AM

COVID has proven that the U.S. government has limited capacity for readiness.
It has less limited capacity to do harm.
It’s arguable that the U.S. government is doing more harm than good in the COVID crisis.
Everyone has to practice business continuity skills, and identify alternative resources that do not depend on government regulation.
In short, the best strategy is not to rely on the government.
Government regulations will always lag. Use foresight instead.

Tim P July 2, 2020 12:05 PM

A mechanism to combat the drive to monopoly would be to tax in proportion to market share. 100% market share results in 100% tax on profit.

mark July 2, 2020 12:10 PM


I’ve heard that word for decades, and I’m not sure I understand it. For example, I hear every year how more “efficient and productive” American workers are… and then the jobs are outsourced overseas, and/or there are massive cuts (but the execs still get their annual bonuses).

It seems to me that this “efficiency” is more in par with the downsizing of the nineties, which I have always referred to as “undersizing”, which was noted, in the mainstream media, as not actually necessary, but kept happening as long as Wall St. rewarded the CEO with increased value on his stock options… never mind how it damaged the company.

Competence specs things out based for peak surge, not for “average”. Could it be this “efficiency” is “spec it out for average, and we’ll take the leftovers as profit”?

It’s like JIT stocking… which is never just in time, and now is a disaster.

Kurt Seifried July 2, 2020 12:12 PM

A mechanism to combat the drive to monopoly would be to tax in proportion to market share. 100% market share results in 100% tax on profit.

How do you measure that? By sales $? By units sold? By usage? What about niche industries where you only have a single player? What happens if you have a duopoly and one goes out of business, you punish the surviving one for simply surviving? This proposal needs a LOT more thought put into it.

Sergey Babkin July 2, 2020 12:25 PM

This makes an assumption that inefficiency creates an overcapacity that can be easily extracted if needed, temporarily making the works more efficient. But that’s not how the inefficiency typically works. It does create the overcapacity but this overcapacity cannot be easily extracted because it gets wasted. If the process involves producing a large amount of waste, there is no magical way to recover that waste. This waste is very, very difficult to recover (if it were easy, it would have been already recovered and the process would become more efficient!).

Bob Paddock July 2, 2020 12:53 PM

“…What’s necessary to make any of this work is regulation. …
We need more local companies,…””

Sadly it is the small local companies that fail due to the cost
of far to many regulations and unfunded government mandates.

Regulations as a solution sound great on paper, sadly they have real world costs that can not be overlooked.

Are there enough honest regulators?

@Tim P

“100% market share results in 100% tax on profit.”

Where is the incentive for that remaining company to stay in business at that point,
or for any new business to start up a new market?

MarkH July 2, 2020 1:18 PM

A small, but relevant and useful example is the U.S. Civil Reserve Air Fleet (CRAF), a program almost 70 years old. Its purpose is to make civil airline capacity available for military transport purposes when needed.

U.S. airline companies voluntarily contract with the federal government to commit a selected subset of their aircraft to CRAF. The way CRAF worked when I first learned about it was that the planes would be configured with reinforced floors, and (if necessary) seat installation tailored for quick removal, to make them suitable for cargo roles.

When called upon, participating airlines must dedicate as many of their committed aircraft as required to military transport, flown by the airline’s own crews.

In return for this commitment, the government pays a sort of charter rate when the planes are used by the military, and an annual fee for participation.

The airlines carry the costs of somewhat heavier aircraft, perhaps a larger or costlier crew roster than they would otherwise carry (they’re required to provide 4 flight crews per plane, who might be called away from their homes for an extended period), and either (a) losing passenger market share when called up, or (b) making their fleet extra large to mitigate that.

The contract payments compensate the airlines for any such compromises of economic inefficiency.

Todd Carmichael July 2, 2020 1:47 PM

Critical societal processes such as water, fuel (truck, rail, air), Food chain, electricity, and communications (internet, phone), medical system and care require full system probabilistic risk assessment using fault analysis to identify weak links and faults and blast radius when a link fails. This kind of a system assessment is already a requirement for electric utilities operating nuclear power plants which have the potential for an immense blast radius. The analysis is submitted to a government regulatory panel for review (Risk assessment for nuclear power plants go to the NRC). Remediation of weak links is required for the process to continue operating. This fault analysis is probably only needed for the large providers as to the point Inefficiency, that small provider’s blast radius is small. The term during the financial crisis of 2008: too big to fail. Unlike money, the government cannot print workers. These large delivery systems have too many players. How and who is making sure that all players are fault tolerant and that their redundancy systems are orchestrated.

Scott July 2, 2020 2:04 PM

There’s a well demonstrated trade-off in evolutionary biology between productivity and adaptability- which applies here as well. You can see it in everything from viral replication to whole ecosystems.

The underlying economic problem isn’t efficiency per. se., it is that commonly used metrics for “efficiency” don’t include resilience.

Clive Robinson July 2, 2020 2:20 PM

@ Bruce, ALL,

We need a certain kind of inefficiency, and it depends on the system in question. Sometimes we need redundancy. Sometimes we need diversity. Sometimes we need overcapacity.

The word you need is “robustness” not “inefficiency”.

I’ve been going on about this and the inadvisability of outsoursing for years, and I’ve been mentioning it all through from the start of the “novel corona virus”.

Likewise I’ve been warning about food supplies especially that of protien via meat supply and the various reasons it’s going to go very badly. Oh by the way there is a new form of “swine flu” starting up killing off one of the major supplys of meat protein and fat “pork”… Which is why I’ve been advising,

    If you have a freezer ‘fill it’, if you know how to can then ‘can it’ and if you know how to cure ‘cure it’, if all else fails then go buy ‘canned ham, chicken, beef’.

If nothing else it will save you money in the near future as the prices go up…

@ Bruce,

Another thing to think about is what economists “assume” but otherwise ignore,

    Distance costs

Not just in time, but ALL resources as well.

The economists notion of a “free market” is axiomatic on the fact that a startup in an existing market has an advantage by being more local to it’s customers than a large “National” or “International” company, thus has an initial cost saving to offset it’s startup costs.

The Internet where the cost of distance is effectivly a “socialist one” in that every one pays for their own piece but graciously carries others traffic in return for carrying their traffic, had a major disruptive effect on the “information economy”. Because information has no mass or energy implications in of it’s self it is the “carrier” it is “modulated onto” where those costs arise, it’s why we prefere massless photons to tablets of stone.

Thus much as I don’t like to say it the easiest way to build robustness back into the system is by making “distance cost” one heck of a lot more. That is you make the cost of transporting items logrithmically more expensive with distance. This would make being “local” cost considerably less than “distant” and the rest would follow on.

But the cost of things would rise to more in line with the local cost of labour, which would be more than somewhat unpopular with those living in the US where most “innovation” is based on things only now available in the Far East.

@ ALL,

As I’ve mentioned before, if you go take a look at nature and evolution you will learn some very important life lessons. If you evolve into a niche then just a small change in the environment will wipe you off the face of the earth, mostly without even a fossil record to morn your passing.

Due to neo-liberal mantra such as “don’t leave money on the table” economists and MBA and other business students believe that to maximize profit is the only important business measure. I urge people to take a look back in history to the likes of “Business Process Reengineering” (BPR) it killed more businesses than it saved. That was back in the 1980’s/90’s and was pushed by the likes of the big accountancy firms and big business consultants with the likes of IBM and co leaping in to grab their pound of flesh.

In essence BPR was the notion of,

    Shop floor workers bring in the money, directors drive this with strategy and middle managment cost more than 50% of the fixed costs.

Thus the obvious but wrong decision was to,

    Replace middle managment with computers, thus “enabling” the shopfloor and “informing” the directors.

By getting rid of middle managment you in effect lobotomized the organisation as well as giving it total “amnesia”. You also took out all the resilience and robustness in the organisation…

Not exactly a bright thing to do but it would have made the company books look good in the short term if not for the consultancy fees and IT systems that failed to deliver… More often than not the company ended up hemorrhaging money the moment there was a tiny problem because it had to “hire back” those middle managers who were the companies memory and brains when anything not normal happened.

JC Checco July 2, 2020 2:44 PM

There is a great allegory taken from Eric Saylors 2017 article “Fire departments are response models, not production models” …
Imagine your family doctor suggests the eradication of your white blood cells since you have not been sick for a year. The logic is reasonable when the doctor compares the production of your white blood cells to your red blood cells, which are at 98% capacity when measured by oxygen saturation. However, your white blood cells have done minimal work over the last year when measured by time spent fighting disease. The conclusion is your red blood cells are at maximum efficiency while your white blood cells are a waste of “excess capacity.” Prudence suggests the elimination of your white blood cells for more efficient red blood cells.

Matt July 2, 2020 2:46 PM

Bruce, the term you’re looking for is “slack.” Systems, like ropes, need slack, or they snap at the first sign of tension.

Markets are extremely bad at maintaining sufficient slack, because as you point out, they ruthlessly eliminate what they see as inefficiencies. But a more holistic view realizes that a system with no inefficiency also can’t deal with disruption. And of course the only way to enforce slack is if the people demand it, and enforce it… via the government.

There’s a lot of ways to implement that, but the single-minded pursuit of efficiency does more damage, overall, than making sure the system has some slack in it.

Monica July 2, 2020 2:54 PM

Tagging onto Matt’s comment, the concept of “slack” and its importance in adapting to change is discussed in Sendil Mullainathan and Eldar Shafir’s book “Scarcity: Why Having Too Little Means So Much.” They come at it from the Behavioral Economics perspective, showing that scarcity of any resource (not necessarily money; they even demonstrated the effects in a purpose-built Angry-Birds-style video game) increases our ability to manage that particular scarce resource, but reduces our overall congnitive capacity and executive control.

Slack is crucial to robust systems (including individual humans). It’s the capacity to absorb unexpected demands on our resources. If you book yourself back-to-back all day, and one appointment runs long, or one big rig jackknifes on the highway you need, you’re going to be late to everything. We know this, intuitively, and we attempt to build slack into our schedules… an extra half-hour here, 15 minutes there, “just in case.” But large organizations don’t have intuition. They have MBAs identifying that slack as fat, as you say.

vas pup July 2, 2020 3:02 PM

@Bruce: very many good points.
My nickel: in order to implement some of your suggestions it is important to make government functional on daily basis and in the time of crisis in particular, not dysfunctional.
In a past on this respected blog I assumed that genuine dislike of the government by many Americans (as President Reagan put like government is always a problem, not solution)coming out of their experience with the government. By the way, Madison stated that if people were angels, we do not need government. Take a look at government in Singapore, South Korea, Japan to have the idea what is functional government.

So, when we have profit as just one dimensional measurement of anything and utilize only one driving force of human behavior as greed (here and now), then we put people as far as possible from angels, and as result need more government. Until we change paradigm itself, we could go nowhere or’looping’ as programmers used to say.

We should consider that requirements to the government itself and President are not the same during normal time and during crisis.

I doubt that liberal agenda fit the modus operandi during times of crisis when it required sometimes fast and swift actions, not blah-blah to bring society out of the crisis with minimum loses (life, property, economy – you name it).

Pete July 2, 2020 3:42 PM

Thank you!
I had come to this set of conclusions myself the day after reading about why there was no toilet paper on the store shelves back in March. The usual business platitudes about efficiency made it clear that the market was indeed making brittle systems, and some government intervention would be needed to require slack in the systems.

V July 2, 2020 3:43 PM

An older example: the Titanic. Because it was unsinkable, Titanic had no need for a lifeboat seat per person. A few surplus lifeboats were included as a public relations gesture. That worked out well for the first part of the first voyage.

Billbo July 2, 2020 4:06 PM

Re: uh, Mike: US government failure

What we have seen is that this particular incarnation of the US (federal and some state) governments seem unable to deal with the current situation. That says nothing about what other incarnations might accomplish. We are also seeing many other governments around the world doing both better and worse than we are. Some of those governments are dictatorial and others are well functioning democracies. The suggestion that we should just give up entirely on government involvement seems unjustified to me.

tfb July 2, 2020 6:38 PM

Related properties hold for computing systems. If you want to make a very large infrastructure – googlebook’s infrastructure, say – very cheap to run, you do so by reducing the number of points of control, ultimately to one. You also reduce the bureaucracy involved in making changes from that single point so you can make changes very easily and wuickly. And when you’ve done that, anyone who can compromise that point of control owns the infrastructure and has no bureaucratic hurdles to jump through to turn it to their purposes. Since the infrastructure is huge and influential, people with very significant resources and very few scruples become interested in doing this.

This has mot happened yet, probably. It certainly will.

Lawrence D’Oliveiro July 2, 2020 7:26 PM

In a famine, it is the ones with excess fat who are best placed to survive.

Also, what about the perverse incentives in a privatized healthcare system? That during a time of national emergency and maximum need, the reduction in profits leads to the layoffs of large numbers of staff?

David Leppik July 2, 2020 8:37 PM

One obvious example that hasn’t been mentioned: the financial system. Banks and insurance companies are required to maintain a certain amount of cash on hand in order to weather downturns or bad luck.

In computer science, memory efficiency often comes at a cost of speed efficiency. Same thing in nature. An animal that is a messy eater may be efficient when measured against something else, such as avoiding predators or managing more valuable resources such as territory.

This is true in the economy as well. A company which is efficient at managing one resource on one time scale is likely inefficient by different measures. For example, a company which is always ready to sell what the customer needs may have an inefficient supply chain but be very efficient at preserving customers.

The American health care system is particularly notable for having a tight supply chain yet high prices for middling outcomes. The inefficiencies are well protected from scrutiny. For example, every visit to the doctor’s office involves that office spending a lot of time and money to make sure that the insurance company doesn’t underpay them, with an equal amount of time and money spent by the insurance company to make sure they don’t overpay. Meanwhile the drug companies are often in a position to charge any price less than the cost of an alternative, such as a long hospital stay or surgery. All of which shows up in higher insurance premiums, well insulated from any particular patient’s activities. Attempts to fix this, such as requiring electronic record keeping, have so far lead mainly to higher clerical costs, which force industry consolidation, leading to inefficient local monopolies.

Steve July 2, 2020 9:22 PM


And we want our local restaurants and businesses to survive, not just the national chains.

Just like we want local retailers and grocers, not Walcostget.

Oh, wait. This just in: There are no local retailers or grocers any longer.

Rj July 2, 2020 11:49 PM

I have done a lot of safety critical software work in my career. Whether it is for an implantable medical device, and commercial avionics system, an industrial control system, or a complex military system, if the potential for serious loss of life or property is significant, safety critical designs usually have to follow strict regulatory requirements from design, development, testing, manufacturing, and deployment.

In very high criticality situations, you might find 2 identical systems running in lockstep and continuously comparing every move that they make. In even more critical situations, functionally equivalent systems with very different implementations are run side by side doing similar comparisons. Sometimes, a design will use 3 identical, or even functionally equivalent systems, and generate a warning fault if all 3 do not agree, but continue to operate as long as 2 out of 3 agree. If all 3 disagree, then an uncorrectable fault has occurred.

It is not efficiancy that it desired here; a single properly operating system would produce the same result at much lower cost. It is redundancy and resiliance. Redundancy allows failures to be detected. Resiliance allows the system to continue to function even in the event of most failures.

Compare this to the science of error detecting and correcting codes. Redundancy must be added to the data stream to detect errors. Even more redundancy must be added to correct errors. At least a small amount more of redundancy must be added to detect when an uncorrectable error has occurred.

Claude Shannon became famous for developing a theory that allows us to determine what the maximum capacity of a communications channel could be based on speeding up the bits and allowing the errors that thereby occur randomly to be corrected.

We need to develop similar results as Shannon’s theory to deal with these “efficiencies” that Bruce is talking about. I think he use the word “efficiency” to allow him to create a catchy title and get people to actually read his article.

SpaceLifeForm July 3, 2020 1:03 AM

@ Rj

“Sometimes, a design will use 3 identical, or even functionally equivalent systems, and generate a warning fault if all 3 do not agree, but continue to operate as long as 2 out of 3 agree. If all 3 disagree, then an uncorrectable fault has occurred.”

I prefer 5 minimum. Detect double faults but continue.

What if triple fault? Well, go to minimum 7.

Now, before you say “that’s expensive”, consider that the cost of 5 is not a doubling of 3, but you get way more assurance, security, trustability.

One may call it slack.

SpaceLifeForm July 3, 2020 1:53 AM

@ Rj

Forgot a point.

If (your words), all 3 disagree, then how do you know it was a fault?

Need clarification.

Distinguish disagree and agree in the 3 machine case.

If it was a triple fault, how will you know?

If three identical turing machines (same start state) get to the same future state via X steps, how can they all be wrong at the same time?

Tom July 3, 2020 4:30 AM

A lot would still work if it were more inefficient. Where I do not see a working alternative is software:

Messengers, social networks, clouds and cryptography.

I would be very interested in Bruce’s view on “inefficient software”.

Petre Peter July 3, 2020 6:45 AM

Inefficiency is also a luxury and in the quest for it, the rich waste space, and the poor waste time.

Clive Robinson July 3, 2020 7:59 AM

@ SpaceLifeForm, Rj,

Need clarification.

Goodness that takes me back, not as far as New Yourk Telephone or even NASA, but yeah it takes me back to explaining one of the small but important parts of C-v-P.

@ ALL,

The history of voting circuits or what we now tend to call “modular redundancy” goes back quite a ways, the most common of which is called “Tripple Modular Redundancy” (TMR). The idea of using three of a mechanical device for error correction goes back to before Harington and his clocks. Back to a time when a pendulum clock even with care could gain or loose five minutes a day at sea, which could give you position errors of over 250kM or 120 nautical miles. In fact there is an old maratime navigation adage which has saved more lives than we can know which is,

    Never go to sea with two chronometers, take one or three

But as far as electrical systems voting systems go back certainly before the transistor and also before the thermionic valve/tube became sufficiently well known that most people had seen the glow in their homes. Infact it was around in the days of relays being used as primative electrical control circuits more than a century ago[1] and caused the spread of the telephone system.

The idea for voting circuits in part goes back to the time of “Alice in Wonderland” and Greek mythology as applied to the works of George Bool and his logic.

If looked at from a distance you actually get two signals from a voting circuit. The first is a parity signal the second is a direction. Thus with three inputs you are looking for an error signal and a majority vote. Thus the error signal can be generated by a three input OR gate for positive logic or a three input AND gate for negative logic.

The majority vote is often achived in positive logic by using three two input AND gates with the three outputs feeding a three input AND gate. The output of each system S is fed to two of the two input logic gates so S1 goes to G1(A) and G2(B), S2 to G1(B) and G3(A), and S3 to G2(A) and G3(B).

Thus whilst the systems are tripply redundnt unfoetunatly the voting circuit is not. Frequently this does not matter because the voting circuit logic is so simple. However with a simple modification you can use two sets or more sets of voting circuits that are effectively in parallel giving extra,reliabiliry which is what NASA decided to do on the Saturn V rocket systems logic used for the Apollo moon landings in the late 1960’s and early 1970’s.

NASA’s use of voting circuits traces it’s history back to the New York Telephone company. They had a problem which was whilst relay systems were reliable individually they became less reliable the more relays that were used. Also fixing relay issues were slow and ment circuits were down for a lot longer than desired. The problem was the solution to speeding the repair up which was to use standardized plugable units, due to the use of connectors reduced the in use reliability. It was known that putting two units in parallel could with a fast switch over solve the problem. However they wanted to use valve amplifier circuits and they wanted these to fail over in use as not just a revenue but reputation saving measure. Their engineers came up with the voting circuit idea which also had the benifit for managment that less expensive exchange people could be used as well as quite a few less of them…

So it is possible to up your availability as well as reduce your maintenance costs. Which should be a “win win” but something managment have kind of forgoton these days. I guess because the prize of big bonuses predicated on reducing costs especially of maintenance in the short term pays better than long term reliability and availability. Thus managment gamble that short term they are going to get away with destroying the company, because they can “skip out” early to abother better rewarding job that hopefully has a large golden parachute to tide them over if the gamble does go wrong on their watch. Kind of like Russian Roulette but without the loosing your brains, just sitting out a turn or two… And people wonder why we had not just Financial Crisis One, and Two but now also have a stimulus package where the only winners are the stock market vampires and everyone else is a looser for generations to come.

[1] The electromechanical relay was a late starter in the electromechabical world following on long after motors were demonstrated and telegraph systems had started spreading out across the globe. In fact it’s original use which gave it it’s name was to “Reley Telegraph signals” and was in effect a current amplifier, in that the weak current of one telegraph circuit could be amplified in a second telegraph circuit, all be it with certain issues. In a way it was a primitive logic circuit that alone could be a buffer or an inverter but pairs of relays could give you not just two input AND, OR, NAND, NOR but XOR XNOR and latch functions. Something Conrad Zuse made use of in the first electro mechabical computer around a century after the relay had been invented.

Kaleberg July 3, 2020 4:57 PM

Computer people have this drummed into them. That’s why computer storage media store a lot more bits than it says on the package. Some bits are going to go bad. Meanwhile, you can compress stuff, but god help you if all you have left is a damaged compressed file.

Sancho_P July 3, 2020 6:04 PM

@SpaceLifeForm, Rj (Clive Robinson)

”I prefer 5 minimum. Detect double faults but continue.
What if triple fault? Well, go to minimum 7.”

Um, there is a problem when you think of each system consisting of sensors, network, processing unit and actors with all the power supply and cabling on different cable trays.
And we don’t need only 5x that “blower motor is on” but that it delivers enough air, not just air pressure, from different sources, also to check each sensor’s plausibility within each system.

”… all 3 disagree, then how do you know it was a fault?”
Per definition disagree is only possible between 2, the first out of bound system is immediately taken out, only two remain. If they disagree a special routine will take place to decide which will remain and usually runs the shutdown, if possible.
Plausibility is the point before it comes to (the safety bottleneck of) voters.
Each safety system has to permanently check it’s sensors and actors as much and often as possible, to find a plausibility value for it’s decisions. It would cut out ambiguous signals (or itself) before the (mostly binary) voter has to decide between systems.

As an example take the 737 MAX: One sensor used, and the sensor failure was not detected (or likely never tested for plausibility?), but would have been easy to detect before takeoff.
A very sad example of failed system engineering.

Btw. as others have mentioned here: Likely the topic’s term “Inefficiency” is inadequate, I’d prefer resilience.

Mr. Peed Off July 3, 2020 7:35 PM

Along the same stream of thought, I read this article:

Former U.S. Treasury Secretary Lawrence Summers has recently conceded: “In general, economic thinking has privileged efficiency over resilience, and it has been insufficiently concerned with the big downsides of efficiency.” Policy across the globe is therefore moving in a more overtly nationalistic direction to rectify this shortcoming.

What we are experiencing is the realization by state planners of developed countries that new technologies enable a rapid ability to expand or initiate new and profitable production capacity closer to or inside their own markets. The cost savings in transport, packaging and security and benefits to regional neighbors and these countries’ domestic workforces will increasingly compete with the price of goods produced through the current internationalized trade system. U.S. national politicians from President Trump to Senator Elizabeth Warren will be joined by a growing chorus who see the long-term domestic political benefit of supporting this transition.

Clive Robinson July 4, 2020 3:41 AM

@ Mr. Peed Off,

What we are experiencing is the realization by state planners of developed countries that new technologies enable a rapid ability to expand or initiate new and profitable production capacity closer to or inside their own markets.

You need to follow the logic down further still and change “inside their own markets” to “inside their own homes”.

Thus the 3D printer mostly only requires “feedstock” and “lubricants and cleaners” as bought in items along with the design files.

Thus the only monetizable thing on which significant profit can be made to sustain the economy is the IP in the design files…

But as with high definition colour printers and “printing money” crime becomes an issue. Currently we only hear about “printed guns” but that is actually an extream corner case currently and for various reasons remain that way for some time.

As I pointed out on this blog some years ago, traditional keys hanging from a waist key ring can be photographed with a telephoto lense. This image gives the key profiles on which the lock security actually rests. Thus cutting up a key can be done by hand or less laboriously by 3D printer. It was not too long after that before people published design files for “handcuff keys”…

How we manage 3D printers in the future will be an interesting study. Colour laser printers made printing illicit copies of books very inexpensive, and would have had a more noticable effect if it had not been for the fact high resolution screens eliminated the need for paper. Instead it was the duplication of audio on magnetic tape and later videos onto DVD that caught peoples attention before it all became “on-line”.

We will however as physical beings continue to need physical items to interact with our physical world. Thus how 3D printers will play out on the world economy and on crime is yet to be seen. But as the ancient Chinese curse has it,

    May you live in interesting times…

Wallace July 4, 2020 3:59 PM

In the same way we don’t expect the market to implement strategic inefficiency because it makes sense, we should also not expect Congress to enact regulation because it makes sense. Instead of thinking about policies that make sense, shouldn’t we focus instead on legislation that can actually be implemented or changing the dynamics that control which legislation gets passed?

Kirk July 6, 2020 10:13 AM

A relevant related concept is “Antifragility”, coined by Nassim Nicholas Taleb. The folks commenting here about robustness or resilience should find the concept interesting.

His home page is here

Kari Lantto July 12, 2020 1:52 PM

About “I don’t simply mean that we need to make our food production, or healthcare system, or supply chains sloppy and wasteful. We need a certain kind of inefficiency, and it depends on the system in question. Sometimes we need redundancy. Sometimes we need diversity. Sometimes we need overcapacity.”

No, I think that’s not the way to think about it! We instead need optimal (or efficient) inventories. Economists (e.g. Arrow, with some associates) formulated the problem around the examples of 1. Keeping optimal amount of cash. 2. Keeping the optimal water level in the dam of a power station. There is a whole anthology about this.

About “The market isn’t going to supply any of these things, least of all in a strategic capacity that will result in resilience. What’s necessary to make any of this work is regulation.”

Well, there seems to be no great externalities here, so an economist would worry less here than about many other considerations.

Still, inventories were clearly inadequate this winter, no doubt about that. Here in Sweden, the responsibilities of public bodies were very unclear, not on paper but in fact.

Kari Lantto July 12, 2020 1:59 PM

The obvious market failure here is the lack of future markets.
A market solution without markets will not e great!

Billikin July 18, 2020 11:08 AM

Not to add much to the discussion, but

The term, efficiency, in economics, is a funny term, and does not have a lot to do with efficiency in the ordinary sense. Certainly its opposite is not waste. And operating on the margins is not generally wise.

As young gambler I struggled with the conundrum of how much to bet, given advantageous odds. Operating on the margins would mean that I should bet as much as possible for even a slight advantage, since that would maximize my expected return. But to do so all the time means that my expectation of losing all my money approached certainty over time. OC, that problem had already been solved. For betting, the solution is called the Kelley criterion. For a sure thing bet you bet your entire stake, while for a bet with no advantage you bet nothing. For a bet with some advantage you bet a corresponding proportion of your stake. The key is not to operate on the margins and strive for maximum “efficiency” all the time, but to take your resources (stake, wealth) into account, as well. As a result, you not only avoid bankruptcy but maximize your return on investment. 🙂 Maximizing return on investment is a good thing. It makes me wonder why marginalism still survives.

BTW, profits rely upon inefficiency, in the economic sense. An efficient market drives profits to zero over time. That is why the principal business model today is to aim for monopoly or monopsony, because monopolies and monopsonies make for inefficient markets, and, hence, profits.

Put simply, risk requires redundanc y. How much and what kind of redundancy is best is another question. I have heard that for long distance running you want to run at 80%. Human language is about 50% redundant. Encryption approaches 0 redundancy, because any redundancy makes the message easier to read (decipher).

Rob van Kranenburg August 2, 2020 2:12 AM

This is a bit too simple and easy. Your very root of thinking is drenched inevitable in this optimizing. and efficiency paradigm, now you want your cake and eat it? Come on!

Leave a comment


Allowed HTML <a href="URL"> • <em> <cite> <i> • <strong> <b> • <sub> <sup> • <ul> <ol> <li> • <blockquote> <pre> Markdown Extra syntax via

Sidebar photo of Bruce Schneier by Joe MacInnis.