A project is given six months, so it takes six months.
A committee approves a multimillion-dollar strategy in ten minutes, then spends an hour arguing over the wording of a minor email.
A company creates a performance metric. Employees learn how to maximize the metric. Soon, the number looks excellent while the actual performance gets worse.
Someone posts a false claim in seconds. Correcting it requires several hours of research, careful explanation, and citations that most readers will never open.
These situations feel irrational when considered individually. Taken together, however, they reveal recurring patterns in how people think, work, organize, measure, and respond to incentives.
Over time, scholars, engineers, psychologists, economists, managers, philosophers, and satirists have given names to many of these patterns. We call them laws, principles, effects, biases, paradoxes, and razors.
They are often described online as “laws of the universe.” That is memorable, but not entirely accurate.
Parkinson’s Law is a satirical observation about work and bureaucracy. Hick’s Law emerged from controlled reaction-time experiments. The Dunning–Kruger effect is a psychological finding whose interpretation remains debated. Occam’s Razor is a philosophical principle. Chesterton’s Fence is an argument about institutional reform. Brandolini’s Law is an internet-age aphorism about misinformation.
They do not all carry the same scientific weight.
What they share is more useful: each compresses a complicated pattern into a form that is easy to remember. They are mental models—tools that help us recognize what kind of problem we may be facing before we rush toward a solution.
A mental model does not tell you exactly what is happening. It gives you a better question to ask.
What are mental models?
Mental models are simplified representations of how some part of reality tends to work. They help us interpret situations, identify likely failure modes, and make better decisions without rebuilding our understanding from the ground up every time.
A good mental model is not a substitute for evidence. It is a lens through which evidence can be examined.
The danger begins when someone learns the name of a principle and starts treating it as a universal explanation. Seeing one possible pattern does not prove that the pattern caused the event. Occam’s Razor does not prove the easiest answer is correct. Hanlon’s Razor does not prove that nobody acted maliciously. The Pareto Principle does not prove that every outcome divides neatly into 80 and 20 percent.
The purpose of these models is not to eliminate thinking. It is to make thinking more disciplined.
I. Time, Productivity, and Why Projects Go Wrong
1. Parkinson’s Law
The principle: Work tends to expand to fill the time available for its completion.
C. Northcote Parkinson introduced the idea in a satirical essay published in 1955 and later expanded it in a book about administration and bureaucratic behavior. His original argument concerned more than procrastination. He also observed that administrative institutions can continue growing even when the amount of useful work they perform does not. (Internet Archive)
Imagine two people receiving the same task. One is given an afternoon. The other is given three weeks.
The first person identifies what is necessary, completes it, and moves on. The second may spend days researching minor alternatives, redesigning the format, reconsidering the structure, scheduling meetings, and perfecting details that nobody requested.
The additional time does not merely provide more breathing room. It can change the perceived size of the assignment.
Parkinson’s Law helps explain why deadlines can improve performance. A real constraint forces priorities to become visible. Without one, optional work begins disguising itself as necessary work.
But the law should not be interpreted as an argument for impossible deadlines. Some work genuinely takes time. Research, testing, training, reflection, and quality control cannot always be compressed without damage.
The lesson is not to allow too little time. It is to avoid confusing the amount of time available with the amount of time actually required.
2. Hofstadter’s Law
The principle: Complex projects take longer than expected, even when you account for the fact that they will take longer than expected.
Douglas Hofstadter coined this deliberately self-referential observation in his 1979 book Gödel, Escher, Bach. The joke is that recognizing our tendency to underestimate a project does not automatically correct the estimate. We add extra time, then discover that we underestimated how much extra time to add. (Wikipedia)
Hofstadter’s Law becomes most visible when a project contains unknown unknowns.
You can estimate how long it will take to write 5,000 words. It is harder to estimate how long it will take to discover that your central source is unreliable, restructure the argument, locate a replacement source, correct an earlier assumption, and resolve a technical problem you did not know existed.
Simple tasks can often be estimated from their visible components. Complex tasks generate new components while they are being performed.
This is why experience does not completely solve scheduling. Experienced people are often better at recognizing known obstacles, but no amount of experience makes every future dependency visible.
Hofstadter’s Law should not become an excuse for permanent lateness. Its practical value is that it encourages buffers, staged delivery, early testing, and honest distinctions between predictable work and exploratory work.
The correct response to uncertainty is not to pretend that schedules are meaningless. It is to make uncertainty part of the schedule.
3. The Pareto Principle
The principle: A relatively small share of causes often produces a disproportionately large share of outcomes.
The Pareto Principle is usually expressed as the 80/20 rule: roughly 80 percent of consequences come from 20 percent of causes. It grew from economist Vilfredo Pareto’s work on unequal distributions and was later popularized in management and quality control by Joseph Juran. The exact numbers are not a natural constant; the underlying point is that outcomes are frequently distributed unevenly. (Investopedia)
A small number of customers may generate most complaints. A few product defects may account for most returns. Several pages of a website may receive most of its traffic. A limited set of habits may create most of a person’s progress—or most of their problems.
The model is valuable because people instinctively distribute attention too evenly. We treat ten problems as though each deserves 10 percent of our effort, even when two problems are causing nearly all the damage.
The Pareto Principle asks a better question:
Which small number of inputs is producing most of the result?
However, 80/20 should not be treated as a mystical ratio. The real distribution may be 60/10, 70/30, or 95/5. In some systems, outcomes are distributed relatively evenly.
The principle is an invitation to measure concentration, not permission to invent it.
4. Brooks’s Law
The principle: Adding people to a late software project can make it later.
Computer scientist Fred Brooks developed this principle from his experience managing complex software projects. His point was that labor and time are not freely interchangeable. Some tasks can be divided among additional workers with little coordination. Software development often cannot. New team members must be trained, existing work must be explained, communication channels multiply, and tightly connected tasks still have to occur in sequence. (Internet Archive)
Nine people cannot produce a baby in one month.
That extreme example captures the larger issue: some processes contain unavoidable sequential dependencies. Adding labor helps only when work can be divided into independent pieces and the benefit exceeds the cost of coordination.
Suppose a project is already late because its architecture is poorly understood. Adding six developers may require the original team to stop working and explain the architecture to six new people. Those new developers may then make conflicting changes, create integration problems, and expand the number of conversations required to make each decision.
Brooks himself acknowledged that the law is an oversimplification. Adding the right specialists early enough can help. A project that is six months from completion may benefit from additional staff even when managers incorrectly believe it is only two weeks late.
The real lesson is not “never add people.” It is that coordination is work, and any plan that counts labor without counting coordination is incomplete.
5. The Planning Fallacy
The principle: People tend to underestimate how long their own future tasks will take, even when they know similar tasks have taken longer in the past.
Research by Roger Buehler, Dale Griffin, and Michael Ross found that people often build predictions around an idealized scenario of how a project will unfold. They focus on the steps they intend to take while discounting their own history of delays, interruptions, mistakes, and competing obligations. (Massachusetts Institute of Technology)
This creates an “inside view.”
From inside the project, the plan seems reasonable:
- Start research Monday.
- Finish the outline Wednesday.
- Write Thursday.
- Revise Friday.
The plan may contain no obviously impossible step. What it omits is the probability that Monday’s research exposes a factual problem, Tuesday disappears into an unrelated emergency, and Thursday’s writing reveals that the outline does not work.
One defense is the outside view: instead of asking only how the current project should unfold, ask how long comparable projects have actually taken.
If your last five reports took between eight and twelve days, the fact that this report could theoretically be finished in four days is not the most useful piece of evidence.
The planning fallacy differs from Parkinson’s and Hofstadter’s laws. Parkinson concerns work expanding into available time. Hofstadter concerns the stubborn unpredictability of complex work. The planning fallacy concerns how people construct unrealistic forecasts by privileging imagined plans over relevant history.
II. Judgment, Confidence, and Bad Decisions
6. Hanlon’s Razor
The principle: Do not attribute to malice what can be adequately explained by incompetence, confusion, neglect, or ordinary error.
The modern formulation became popular after appearing under the name “Hanlon’s Razor” in a 1980 collection of Murphy-style sayings, although similar ideas appeared much earlier. Its exact authorship is less important than its function: it warns against jumping immediately to the most hostile interpretation of another person’s conduct. (Wikipedia)
Someone misses a deadline. Perhaps they are sabotaging you. Perhaps they misunderstood the assignment.
A business sends an inaccurate invoice. Perhaps it is fraud. Perhaps an employee copied the wrong account number.
A government program produces a terrible result. Perhaps officials deliberately engineered that outcome. Perhaps the institution is fragmented, poorly managed, and incapable of doing what it promised.
Hanlon’s Razor is useful because malice creates a satisfying story. It transforms confusion into intention and gives disorder a villain.
But it must not become enforced naivety.
People do lie. Institutions do conceal information. Powerful actors sometimes benefit from outcomes they claim were accidental. Repeated “mistakes” that always advantage the same party deserve closer examination.
Hanlon’s Razor does not say malice is impossible. It says malice should not be your automatic starting assumption when a simpler human failure explains the evidence.
7. Occam’s Razor
The principle: When competing explanations account for the evidence equally well, prefer the explanation that requires fewer unnecessary assumptions.
Occam’s Razor is associated with the medieval philosopher William of Ockham, although the familiar modern wording was not a direct quotation from him. Philosophers have formulated the principle in several ways, usually involving parsimony: explanations should not multiply entities, causes, or assumptions beyond necessity. (archive.ph)
Suppose your internet connection stops working.
One explanation is that the router lost power. Another is that a foreign intelligence agency has targeted your house through a covert operation involving the cable company, your neighbor, and a modified satellite.
The second explanation is not impossible. It is simply carrying far more unsupported machinery.
Occam’s Razor tells us to begin with the explanation that accounts for the evidence with the fewest additional assumptions.
The crucial phrase is accounts for the evidence.
The simplest explanation is not automatically true. “Nothing happened” is simpler than a complicated criminal conspiracy, but it is useless if physical evidence shows that something did happen. A complicated explanation supported by strong evidence is better than a simple explanation that ignores half the facts.
Occam’s Razor does not reward simplicity for its own sake. It penalizes unnecessary complexity.
8. The Dunning–Kruger Effect
The principle: Poor performance in a particular domain can be accompanied by poor awareness of that performance.
In their original 1999 studies, Justin Kruger and David Dunning found that participants who performed poorly on tests involving grammar, logic, and humor substantially overestimated their relative performance. The researchers argued that some of the skills needed to perform well were also needed to recognize what good performance looked like. (LSA Technology Services)
The idea is frequently distorted online into something much cruder: stupid people believe they are geniuses, while experts doubt themselves.
That is not what the research established.
The effect was about calibration in specific domains. A person may be highly competent in engineering, poorly calibrated in politics, reasonably self-aware in cooking, and completely unaware of their limitations in finance.
It also should not be used as a sophisticated-sounding insult. Accusing another person of Dunning–Kruger does not prove they are wrong. In fact, confidently diagnosing the effect in everyone who disagrees with you may demonstrate the same failure of self-evaluation the concept is supposed to illuminate.
There is also an active methodological debate. Some researchers argue that the familiar Dunning–Kruger curve can be produced partly or largely by regression to the mean, measurement error, and bounded scoring systems rather than a unique psychological mechanism. (Frontiers)
The safest lesson is narrower and stronger: expertise and self-assessment are different abilities. Confidence should be tested against performance, feedback, and evidence.
9. The Sunk Cost Effect
The principle: People become more willing to continue an endeavor after investing money, time, effort, or identity in it—even when those past investments cannot improve the future outcome.
A classic study by Hal Arkes and Catherine Blumer described the sunk cost effect as a greater tendency to continue once an investment has already been made. Their experiments and field observations suggested that people are motivated partly by a desire not to appear wasteful. (Universität Klagenfurt)
You buy an expensive ticket to an outdoor event. On the day of the event, a storm arrives. You no longer expect to enjoy yourself, but you go because you “already paid for it.”
The payment is gone whether you attend or stay home. The rational question is not whether the ticket was expensive. It is whether attending now will make your life better than not attending.
The same trap appears in relationships, businesses, wars, educational paths, software projects, and political strategies.
“We have already spent too much to quit” can sound responsible while actually arguing for spending even more.
Past costs may still provide useful information. A company that has invested years developing a product may possess knowledge and infrastructure that improve its future prospects. That is not a sunk cost argument; it is a claim about present assets and future returns.
The fallacy begins when the past investment itself becomes the reason to continue.
10. Survivorship Bias
The principle: We draw distorted conclusions when we study visible successes while ignoring the people, businesses, strategies, or systems that failed and disappeared.
The classic example comes from statistician Abraham Wald’s wartime work on aircraft vulnerability. Damage could be inspected only on planes that survived and returned. The missing data—the aircraft that did not return—were precisely what made some apparently undamaged areas important. Wald’s surviving memoranda were later reprinted by the Center for Naval Analyses. (Praxis)
The broader pattern appears everywhere.
A billionaire says dropping out of college was the key to success. We see the billionaire. We do not see the thousands of people who dropped out, pursued similar ideas, and failed without becoming subjects of documentaries.
A business book studies ten companies that survived for 50 years and identifies their shared habits. But were those habits absent from companies that failed? Were the successful companies preserved by skill, luck, market conditions, government support, or selection after the fact?
Survivorship bias does not mean successful people have nothing to teach us. It means that studying only survivors cannot tell us which characteristics caused survival.
Whenever the evidence consists entirely of winners, ask what happened to the losers—and why their data are missing.
III. Organizations, Bureaucracy, and Complex Systems
11. The Peter Principle
The principle: In a hierarchy, people may be promoted based on competence in their current role until they reach a role requiring different skills they do not possess.
Laurence J. Peter and Raymond Hull presented the principle in their 1969 satirical management book. Despite its humorous framing, the idea describes a serious structural problem: strong performance in one job does not necessarily predict strong performance in the job above it. (Wikipedia)
A gifted salesperson may be promoted into management. Selling requires persuasion, customer knowledge, and personal initiative. Managing a sales team requires coaching, conflict resolution, delegation, hiring, and performance evaluation.
The person may have been promoted because of genuine excellence and still be unprepared for the new role.
Organizations often make matters worse by treating promotion as the primary reward for high performance. Employees must either leave the work they do well or accept that their careers have stopped advancing.
The Peter Principle is not inevitable. Organizations can reduce the risk through training, trial periods, technical career tracks, realistic job previews, and promotions based on the skills required in the next role rather than prestige earned in the current one.
The principle also does not prove that every ineffective leader was once excellent. Some people are promoted for loyalty, politics, seniority, or convenience.
Its central warning remains sound: success at one level is evidence of competence at that level, not automatic proof of competence at the next.
12. Chesterton’s Fence
The principle: Do not remove an institution, rule, or structure until you understand why it was created.
G.K. Chesterton developed the argument in his 1929 book The Thing. His hypothetical reformer encounters a fence across a road and proposes removing it because its purpose is unclear. Chesterton’s response was that not understanding the fence is a reason to investigate it, not proof that it serves no purpose. (Catholic Library)
The model is especially useful in old organizations.
A new executive discovers that two employees independently review the same transaction. It looks inefficient. Before eliminating one review, the executive should determine whether the duplication was created after a costly fraud, safety failure, or regulatory violation.
Old systems often contain institutional memory that no longer exists in written form.
But Chesterton’s Fence can also be abused. It is not a command to preserve every tradition forever. Some fences were built for reasons that were mistaken, discriminatory, obsolete, or openly harmful. Others continue existing only because nobody has taken responsibility for removing them.
Understanding the purpose of a rule is not the same as agreeing with it.
The model establishes a sequence:
- Find out why the fence exists.
- Determine whether that reason still applies.
- Examine what may happen if it is removed.
- Then decide whether to preserve, modify, or dismantle it.
Chesterton’s Fence is not a veto against reform. It is a demand for informed reform.
13. The Law of Triviality, or Bikeshedding
The principle: Groups may devote disproportionate attention to simple, low-stakes matters because those matters are easier for everyone to understand.
Parkinson illustrated this with a fictional committee considering three expenditures. The enormously expensive and technically complicated item passed quickly because few members understood it. A modest bicycle shed triggered extensive debate because everyone felt qualified to discuss it. The phenomenon became known as bikeshedding. (Wikipedia)
The same thing happens when a team spends five minutes examining a major security risk and 40 minutes choosing a logo color.
Complex questions create social discomfort. People fear exposing what they do not understand. Trivial questions create opportunity. Everyone can offer an opinion, demonstrate participation, and leave a visible mark on the decision.
This creates a dangerous inversion: the amount of discussion becomes negatively related to the importance of the subject.
The solution is not to ban discussion of details. Small details can matter. The solution is to structure attention deliberately.
Important agenda items should receive protected time. Technical decisions should be explained in accessible language. Decision-makers should be expected to identify what they do not understand. Minor matters should have proportional limits.
A group that speaks confidently for an hour may still be avoiding the only question that matters.
14. Conway’s Law
The principle: Organizations tend to design systems that reflect their own communication structures.
Computer programmer Melvin Conway developed the idea in his 1968 paper “How Do Committees Invent?” His argument was that the way design work is divided and communicated constrains the kinds of systems an organization can produce. (Mel Conway)
Imagine a company with four departments that rarely communicate. Each department builds one part of a software platform. The final product may contain four disconnected experiences, duplicated databases, conflicting terminology, and awkward boundaries that mirror the departments themselves.
The architecture is not merely technical. It has absorbed the organization chart.
Conway’s Law extends beyond software. A government divided among agencies may produce policies that require citizens to navigate the same institutional divisions. A hospital with poor communication between departments may create a patient experience fragmented along those exact lines.
This is why a redesign cannot always be solved by hiring better designers. If the organization remains structurally unable to communicate across boundaries, the product will continue reproducing those boundaries.
The inverse lesson is powerful: changing team structure can change what a team is capable of building.
Before fixing the system, inspect the organization that produced it.
15. Gall’s Law
The principle: A complex system that works is usually found to have evolved from a simpler system that worked.
John Gall presented the idea in Systemantics, a satirical but serious examination of why systems fail. Gall warned that a complex system designed from scratch is unlikely to function as intended and often cannot be repaired merely by adding more layers. (Internet Archive)
Consider the difference between two approaches to building an online marketplace.
One team spends two years designing a complete system containing payments, identity verification, dispute resolution, recommendations, international shipping, advertising, analytics, and 15 user roles.
Another team begins with one seller, one buyer, one product, and one reliable transaction. It then adds capabilities while preserving what already works.
The second system may eventually become more complex than the first. The difference is that its complexity has been tested through evolution.
Gall’s Law does not mean every successful system began as a toy, nor that planning is useless. Some systems require safety, scale, or regulatory features from the beginning.
Its deeper insight is that complexity conceals interactions. Building incrementally allows those interactions to be discovered while the system remains understandable.
Complexity is not the enemy. Unvalidated complexity is.
IV. Metrics, Incentives, Choice, and Design
16. Hick’s Law
The principle: As the number and uncertainty of possible choices increase, the time required to make a decision generally increases.
The principle grew from experiments by William Hick and Ray Hyman examining the relationship between information and reaction time. Their work showed that response time increased as participants faced more possible stimuli and responses, although the relationship depends on probability, organization, familiarity, and context—not simply the raw number of options. (SciSpace)
This has obvious implications for design.
A restaurant menu with 300 undifferentiated dishes is harder to navigate than a menu organized into several meaningful categories. A website with 20 equally prominent buttons creates more decision friction than one that makes the main action clear.
But Hick’s Law is often oversimplified into “fewer choices are always better.”
That is not true. Removing necessary options can make a system less useful. Experts may prefer more controls because they understand the categories. Familiar choices require less effort than unfamiliar ones. Ten well-organized options may be easier to navigate than five confusing ones.
The design lesson is not merely to reduce choice. It is to reduce unstructured uncertainty.
Good design helps people understand what choices mean, which choices matter, and where to begin.
17. Goodhart’s Law
The principle: When a measure becomes a target, it can stop functioning as a good measure.
The principle is associated with economist Charles Goodhart and originally emerged from monetary policy. It has since become a general warning about proxy optimization: once rewards or punishments depend heavily on a metric, people begin changing their behavior to improve the number rather than the underlying reality the number was meant to represent. Researchers have identified several distinct mechanisms through which this can happen. (arXiv)
A school is judged by test scores. Teaching gradually becomes test preparation. Subjects not included on the test receive less attention. Administrators may exclude weaker students, narrow the curriculum, or manipulate reporting.
The scores may rise while education deteriorates.
A customer-service department is rewarded for reducing call time. Employees begin ending calls quickly, transferring difficult customers, or providing incomplete answers. Average call time falls. Repeat calls rise.
The metric was useful when it was observed. It became dangerous when the organization treated maximizing it as equivalent to achieving the goal.
Goodhart’s Law does not mean measurement is futile. Organizations cannot operate without indicators.
It means every metric should be treated as a partial representation. Pair quantitative measures with audits, qualitative evidence, adverse-effect monitoring, and regular checks that the proxy still tracks the actual objective.
Never confuse the scoreboard with the game.
18. Campbell’s Law
The principle: The more a quantitative social indicator is used for high-stakes decision-making, the more pressure there will be to corrupt it—and the more likely it is to distort the activity being measured.
Social scientist Donald Campbell developed this warning while examining the evaluation of social programs. Campbell was concerned that indicators used to judge schools, public agencies, and policies could be manipulated and could transform the institutions they were intended to evaluate. (Human Learning Systems)
Campbell’s Law resembles Goodhart’s Law, but the emphasis is different.
Goodhart concerns the breakdown of a proxy when it is optimized. Campbell focuses more directly on social pressure, corruption, and institutional distortion when a number carries serious consequences.
Suppose police departments are judged by arrest totals. Officers may prioritize easy arrests rather than serious crimes. If hospitals are judged only by short-term mortality, they may become reluctant to accept the sickest patients. If researchers are judged primarily by publication counts, publishing more papers may become more important than producing reliable knowledge.
The indicator does not merely become inaccurate. It begins reorganizing behavior around itself.
The more consequential the metric, the more aggressively its failure modes should be examined.
Ask not only, “What does this number measure?” Ask, “What will people do once their careers depend on it?”
19. The Cobra Effect
The principle: An incentive intended to solve a problem can reward behavior that makes the problem worse.
The term is commonly linked to a story in which colonial authorities supposedly offered bounties for dead cobras, encouraging people to breed snakes for payment. When the program ended, breeders allegedly released the snakes. The historical documentation for the familiar story is disputed, so it is better understood as a parable than as firmly established history. The underlying phenomenon—perverse incentives—is nevertheless real and widely observed. (Friends of Snakes Society)
Pay employees for every bug they fix, and some may write more bugs.
Reward social-media platforms for maximizing time spent, and they may discover that outrage and compulsion hold attention better than truth or well-being.
Fine a company a small amount for violating a rule, and the company may begin treating the fine as a price rather than a prohibition.
The Cobra Effect appears when designers model the behavior they want but fail to model how people will adapt strategically to the reward.
The most important question in incentive design is not:
What behavior are we paying for?
It is:
What is the easiest way to collect the reward without producing the outcome we actually want?
A strong incentive system anticipates gaming, measures unintended effects, and can be changed before exploitation becomes institutionalized.
20. Tesler’s Law, or the Conservation of Complexity
The principle: Some complexity cannot be eliminated; it can only be moved to a different part of the system.
Computer scientist and user-interface pioneer Larry Tesler argued that applications contain a certain amount of irreducible complexity. The design question is who will have to manage it: the user, the application developer, or the platform. (Larry Tesler)
Consider tax preparation software.
The tax code is complex. A designer cannot make the underlying rules disappear. The software can, however, absorb part of that complexity through automatic calculations, guided questions, error checking, and imported records.
The interface becomes easier because the developer accepted more complexity behind it.
Poor systems often claim to be simple because they transfer work to someone less powerful. A company may simplify its internal process by forcing customers to enter the same information repeatedly. A manager may simplify oversight by requiring employees to produce elaborate weekly reports. One department’s efficiency can become another person’s administrative burden.
Tesler’s Law does not mean complexity is literally fixed forever. Better technology, clearer rules, and redesigned processes can reduce it.
The principle asks us to examine claims of simplicity more carefully.
When a process becomes “easier,” find out whose work disappeared—and whose work increased.
V. Technology, Information, and Unintended Consequences
21. Murphy’s Law
The principle: If a system permits a serious mistake, someone will eventually make it.
Murphy’s Law is commonly stated as “anything that can go wrong will go wrong.” Its precise origin is disputed, but it is associated with engineer Edward A. Murphy Jr. and late-1940s aerospace testing. In its most useful interpretation, it is not cosmic pessimism. It is an engineering argument against designs that rely on every person behaving perfectly forever. (Wikipedia)
A connector can be installed in two directions. One direction works. The other destroys the equipment.
A weak designer writes a warning label.
A strong designer changes the connector so it physically cannot be installed incorrectly.
Murphy’s Law asks us to treat foreseeable error as a design condition rather than a moral failure. People become tired. Instructions are misunderstood. Components fail. Rare events become likely when a system operates millions of times.
The law does not mean every possible disaster is inevitable. It means that plausible failure paths deserve attention proportional to the harm they could cause.
The practical response is redundancy, fail-safe design, testing, backups, monitoring, and graceful degradation.
Hope is not a control system.
22. The Law of the Instrument
The principle: Familiar tools shape how we define problems, causing us to force situations into forms our preferred tools can handle.
Abraham Kaplan described this as the “law of the instrument” in his 1964 work on research methodology. Abraham Maslow later popularized a related formulation: when the only tool available is a hammer, everything begins to look like a nail. (Wikipedia)
A lawyer sees a legal dispute. An engineer sees a technical failure. A marketer sees a messaging problem. A consultant sees a need for another consulting engagement.
Each may be partly correct. The danger is that expertise can create selective vision. People become highly skilled at applying one solution and gradually lose the ability to recognize problems that require a different approach.
The law also explains technological hype. Once an organization adopts a new platform, it begins searching for reasons to use it. Blockchain, artificial intelligence, surveillance technology, standardized testing, and data dashboards may be useful in some settings and absurd in others.
A tool should be selected after the problem is understood.
When the tool arrives first, the “problem” may be reverse-engineered to justify it.
23. Amara’s Law
The principle: We tend to overestimate the effects of new technology in the short term and underestimate them in the long term.
The observation is commonly attributed to futurist Roy Amara, who led the Institute for the Future. It describes a recurring mismatch between technological demonstrations, practical adoption, and eventual social transformation. (Wikipedia)
A new technology produces an impressive demonstration. Investors, journalists, and executives extrapolate rapidly. They predict immediate disruption across entire industries.
Then reality intervenes.
The technology is expensive. Infrastructure is missing. Regulations are unresolved. Workers must be trained. Consumers resist changing established habits. Early products fail to meet inflated expectations.
Disappointment follows, and observers swing too far in the other direction. Because the technology did not transform everything in two years, they assume it never will.
Meanwhile, costs decline. Infrastructure spreads. Interfaces improve. Institutions adapt. The technology becomes ordinary—and then quietly changes the world.
Amara’s Law is not a guarantee that every hyped technology will eventually succeed. Many disappear completely.
Its value is in separating two questions:
- Is the technology being overpromised right now?
- Could its mature form still have major long-term consequences?
Both answers can be yes.
24. The Streisand Effect
The principle: Attempting to hide, censor, or remove information can draw more attention to it and accelerate its spread.
The term emerged after Barbra Streisand pursued legal action concerning an aerial photograph of her California home. The dispute attracted far more attention to the image than it had previously received, and the incident became a shorthand for failed attempts at online suppression. (Verywell Mind)
Suppression creates a second story.
The original information may have been unremarkable. The attempt to remove it introduces conflict, secrecy, power, and curiosity. People begin sharing the material not because they care deeply about the original content, but because they object to the effort to control access.
Digital networks intensify the effect. Copies can be distributed instantly, mirrored across jurisdictions, archived, reframed as memes, and shared by people who view preservation as an act of resistance.
The Streisand Effect does not mean every takedown request will backfire. Illegal exploitation, private personal information, and harmful content may still justify intervention. Quiet platform moderation can succeed without attracting wider attention.
The principle warns that public, aggressive suppression can change the value of information by making the suppression itself newsworthy.
Before attempting to erase something, ask whether the removal effort will become its most effective advertisement.
25. Brandolini’s Law
The principle: Producing misinformation is far easier than carefully investigating and correcting it.
Italian programmer Alberto Brandolini formulated the idea in 2013. It is also called the “bullshit asymmetry principle”: the effort needed to refute a false or misleading claim can be dramatically greater than the effort needed to produce it. (Wikipedia)
A person can write:
“Scientists proved this common food causes cancer.”
The sentence takes seconds.
A responsible correction may require identifying the study, reading its methodology, determining whether it involved humans or animals, examining sample size and dosage, comparing the finding with the wider literature, checking conflicts of interest, and explaining the difference between correlation and causation.
By the time the correction is complete, the original claim may have reached millions of people.
This asymmetry rewards confidence, speed, and emotional simplicity. Truth is often conditional. Falsehood can be perfectly optimized for sharing.
Brandolini’s Law also explains why demanding that experts individually refute every false claim is impossible. A system capable of generating misinformation at industrial scale can exhaust any fact-checking system that operates one claim at a time.
The solution must include prevention: better information literacy, accountable platforms, credible institutions, transparent evidence, and communication that reaches people before the falsehood becomes part of their identity.
The truth does not automatically win because it is true. It must also survive the economics of attention.
How These Mental Models Connect
The real value of these principles appears when they are used together.
A project begins with the planning fallacy, as leaders underestimate how long it will take. It then encounters Hofstadter’s Law, because unexpected complexity creates new delays.
Management responds by adding employees, triggering Brooks’s Law. The team grows, communication becomes fragmented, and the product begins reflecting the organization’s internal divisions through Conway’s Law.
Executives create a performance target. Goodhart’s Law takes over as employees optimize the measurement. Because bonuses depend on it, Campbell’s Law increases pressure to manipulate the number.
A new incentive is introduced, producing a Cobra Effect that rewards the wrong behavior.
Nobody wants to cancel the project because of the sunk cost effect. The few organizations that survived similar projects publish triumphant case studies, creating survivorship bias.
When the system finally fails, everyone spends the review meeting arguing about a minor design choice, completing the cycle with the Law of Triviality.
This is why mental models are most powerful as a network rather than a collection of slogans. One model rarely explains an entire event. Several may describe different layers of the same failure.
How to Use Mental Models Without Becoming Overconfident
Knowing these names does not make anyone immune to the patterns.
A person who recognizes survivorship bias can still ignore missing evidence. Someone who understands Goodhart’s Law can still build destructive incentives. Someone familiar with the Dunning–Kruger effect can become more confident while becoming no more accurate.
Mental models should create questions, not conclusions.
When facing an important decision, ask:
- Am I relying on an idealized plan rather than historical evidence?
- Is a metric replacing the goal it was supposed to represent?
- What behavior will this incentive reward in practice?
- Which relevant failures are absent from the data?
- Am I preserving a system merely because it is old—or destroying it before understanding its function?
- Is the problem actually suited to my preferred solution?
- Is additional complexity solving the problem or hiding it?
- What would have to be true for my explanation to be wrong?
The strongest thinkers do not carry one perfect model. They carry many imperfect models and know when each one stops applying.
The Most Important Law: The Map Is Not the Territory
Every principle in this article is a map.
A map removes details so that a pattern becomes easier to see. That simplification is its strength. It is also its weakness.
Parkinson’s Law does not describe every deadline. The Peter Principle does not explain every bad manager. The Streisand Effect does not make every attempt at privacy self-defeating. Amara’s Law does not guarantee that today’s overhyped technology will become tomorrow’s revolution.
Reality remains more complicated than the model.
The purpose of learning these laws is not to win arguments by naming them. It is to notice recurring structures early enough to make better decisions.
A useful mental model should make you more curious, more precise, and less certain of your first explanation.
When it becomes a replacement for investigation, it has stopped being a thinking tool and become another way to avoid thinking.
Frequently Asked Questions
Are these really scientific laws?
No. The collection includes experimental findings, philosophical principles, management observations, engineering rules, cognitive biases, and memorable aphorisms. Hick’s Law has roots in controlled psychological research, while Parkinson’s Law began as satire. They should not be treated as equally proven or universal.
What is the difference between a law and a mental model?
A scientific law describes a consistently observed relationship under defined conditions. A mental model is broader: it is a simplified framework used to understand or predict a pattern. Many so-called “laws” in business and popular psychology are better understood as mental models.
What is the most useful mental model?
That depends on the problem. The planning fallacy is especially useful for scheduling. Goodhart’s Law is essential when designing metrics. Survivorship bias matters when studying success. The sunk cost effect is valuable when deciding whether to continue a failing commitment.
Can mental models make people more biased?
Yes. A person can force every situation into a familiar model, which is itself an example of the Law of the Instrument. Mental models improve judgment only when they remain open to evidence and competing explanations.
How many mental models should a person learn?
There is no ideal number. A smaller set understood deeply is more useful than hundreds memorized as slogans. The goal is to learn models from different fields so that no single framework dominates every decision.
References and Further Reading
Time, projects, and organizations
- C. Northcote Parkinson, Parkinson’s Law and Other Studies in Administration. (Internet Archive)
- Douglas Hofstadter, Gödel, Escher, Bach: An Eternal Golden Braid. (Internet Archive)
- Fred Brooks, The Mythical Man-Month. (Internet Archive)
- Roger Buehler, Dale Griffin, and Michael Ross, “Exploring the Planning Fallacy.” (Massachusetts Institute of Technology)
- Melvin Conway, “How Do Committees Invent?” (Mel Conway)
- John Gall, Systemantics: How Systems Work and Especially How They Fail. (Internet Archive)
Judgment and decision-making
- Justin Kruger and David Dunning, “Unskilled and Unaware of It.” (LSA Technology Services)
- Jan R. Magnus and Anatoly A. Peresetsky, “A Statistical Explanation of the Dunning–Kruger Effect.” (Frontiers)
- Hal R. Arkes and Catherine Blumer, “The Psychology of Sunk Cost.” (Universität Klagenfurt)
- Abraham Wald, “A Method of Estimating Plane Vulnerability Based on Damage of Survivors.” (Praxis)
- Stanford Encyclopedia of Philosophy, “Simplicity.” (archive.ph)
Metrics, incentives, and design
- W.E. Hick, “On the Rate of Gain of Information.” (Scilit)
- Ray Hyman, “Stimulus Information as a Determinant of Reaction Time.” (Europe PMC)
- Donald T. Campbell, Assessing the Impact of Planned Social Change. (Human Learning Systems)
- David Manheim and Scott Garrabrant, “Categorizing Variants of Goodhart’s Law.” (arXiv)
- Larry Tesler, “The Law of Conservation of Complexity.” (Larry Tesler)
Institutions and reform
- G.K. Chesterton, The Thing. (Google Books)
- Abraham Kaplan, The Conduct of Inquiry. (Garfield Library)
