Prefer watching? Short videos appear throughout the article when another explanation may help. They are optional and open on YouTube.
WATCH THISA simple guide to chaos theoryBBC World ServiceA clear explanation of sensitivity to initial conditions and why chaos does not mean total randomness.Watch on YouTube
You cannot control everything
Chaos theory does not mean that everything is random. It shows that some systems can be very sensitive to small differences.
Two very similar initial states are represented by green and purple trajectories. Their paths diverge as time passes in a Lorenz-inspired depiction. The diagram distinguishes evidence about deterministic dynamical systems from the author's analogy for human possibilities: sensitive dependence does not give a method to choose or guarantee favourable outcomes.
Some systems follow fixed rules. Nothing in them is random. Edward Lorenz showed that in such a system, a tiny difference at the start can grow into a much larger difference later.
That is why long-term prediction can become very difficult. It does not mean that one small action will reliably create a huge result, or a good one. Lorenz described the limits of prediction. He did not offer a recipe for making good things happen.
A human life is not a Lorenz system. The comparison used in this article is an analogy, not a finding. It is there to help us think clearly about uncertainty.
The useful idea is narrower. We cannot control the future. But we can sometimes change some of the conditions from which later options grow. We call those conditions your search landscape: the options you can currently see, reach or create. Changing it changes what is possible, not what is guaranteed.
EvidenceWhat Lorenz actually supports
Lorenz (1963) showed sensitive dependence in a specific nonlinear deterministic system. His work supports limits to prediction in that system. It does not show that human choices reliably create beneficial “butterfly effects”. Lorenz (1972) was a talk, not a journal paper.
AnalogyWhy keep chaos in the title?
The life comparison is an analogy. It is useful because it reminds us to aim for influence without pretending to have control.
WATCH THISFour Ways of Letting Go | Ajahn Brahm | 09-04-2010Buddhist Society of Western AustraliaA simple story about letting go of what you cannot control, used here as philosophy rather than scientific evidence.Watch on YouTube Small choices can change what becomes possible
A choice does not need to control your future to change what becomes possible next.
A current state branches into Choice A and Choice B, with subsequent nodes showing some paths available and others presently inaccessible or unknown. The options reached later depend partly on the initial branch and what happens along it. It is a schematic model, not a literal graph of any person's future.
Some choices change which later options become easier, harder or simply visible. Researchers call this path dependence. It is studied in technology adoption and in politics, where early events can shape what comes later.
A small action can open a new route without guaranteeing where that route ends. Sending one email, joining one group, trying one class or speaking to one new person can make a later option easier to reach. None of them fixes the result.
The diagram shows two otherwise separated social networks and a possible bridging relationship across them. A bridge can expose new information, introductions and environments that might not circulate inside one's immediate network. The value of an additional connection is contextual and nonlinear, not automatically positive.
Meeting someone outside your usual circle can expose you to information or chances you would not otherwise see. Research on weak ties supports this idea. It does not support the simpler rule that weaker ties are always better.
This is the search landscape again: the options you can currently see, reach or create. Small choices can change that landscape without controlling the future. They change what is worth trying next.
Path dependence
Arthur (1989) and Pierson (2000) show how timing, sequence and reinforcement can shape what becomes easier or harder later.
Network bridges
Granovetter (1973) and Rajkumar et al. (2022) show that social structure can affect access to information and opportunity. The effect is not simply ‘weaker is always better’.
Personal observationThe £30 / £150,000 pricing lesson
While pressure washing, I once suggested charging about £30 to clean a piece of equipment worth roughly £150,000. The customer's response was simple: I should charge more.
The lesson was not just about pricing. It was about how our first idea of value can be too small.
If you price too low, think too small, or frame your work badly, you may hide better options from yourself. In that sense, your current judgement can limit what feels reachable next.
This is one example, not a general rule. But it shows how a person's starting assumptions can shape which possibilities they even consider.
In optimisation, a local optimum is a good nearby solution that may not be the best solution overall. A search can settle on a small hill and never reach a higher one beyond the valley.
That does not prove life has a single best outcome waiting somewhere. The image is a computational analogy. It shows how a current assumption can limit what you explore next.
A person is pictured on an apparently good nearby peak while the diagram depicts a possible distant higher peak separated by a costly valley. Computational panels contrast one objective with adding another to change the search landscape. The illustration is hypothetical: it does not establish that a globally better human life exists elsewhere, nor that any known objective captures human value.
WATCH THISSteve Jobs on FailureSilicon Valley Historical AssociationA short example of how one small action can open a chain of opportunities that was not visible beforehand.Watch on YouTube You can get better at spotting and using options
Finding better options is partly about what exists around you, and partly about what you are able to notice and use.
The model separates capability (what a person can do), representation (which options and distinctions they perceive), and judgement (which options they consider worth pursuing). Additional learning can change all three, but judgement also needs values and context. It is conceptual, not a validated human optimiser.
Two people can read the same job advert, or face the same business problem, and see different options. One asks a question the other never thinks to ask. That difference is usually experience rather than luck. This is an everyday example, not research evidence.
Knowledge and practice can help. With experience, people often spot useful patterns faster, ask sharper questions and make better use of what is available. Research on skill learning supports this, with a caveat: practice explains some of the gap between people, not all of it.
Expertise has limits. People are not software. We learn and form habits, but we do not update by a simple rule, and knowing a field well does not automatically make judgement better. The same knowledge that highlights one path can leave others out of view. So it helps to ask how something actually works, to borrow an idea from another field, or to talk to someone who does it differently. These are practical habits, not a proven formula.
This is the search landscape again. Better knowledge, questions and judgement can change what you notice and use, even when the world around you has not changed. It is not a promise of unlimited agency. Money, time, health and responsibilities still set real limits.
Skill acquisition
Macnamara, Hambrick and Oswald (2014) found that deliberate practice explains some performance differences, but far from all of them, and the amount varies a lot between fields.
Phronesis
Aristotle’s practical wisdom asks what good judgement looks like when knowledge, values and context all have to fit together.
Go deeperEvolution, learning and representation
Watson and Szathmáry draw formal links between evolution and learning. Caldwell and colleagues give a computing example where a learned representation changes what a search can reach. These are analogies for human learning, not proof that the mind works like an optimiser.
WATCH THISHow Great Entrepreneurs See What Others Don'tStanford Graduate School of BusinessAmy Wilkinson explains how curiosity and analogy can help people notice opportunities others overlook.Watch on YouTube Test small before betting big
When uncertainty is high, a small test can sometimes teach you more than a big commitment.
The upper sequence shows uncertainty, a low-cost reversible test, evidence, an update to the plan and a possible larger commitment. A lower stylised graph compares declining uncertainty with increasing investment; this is *illustrative*, not a computed optimum or a guaranteed minimum-risk point. A practical example shows checking the uncertain fit of a small component before making a full part.
Some decisions are expensive to undo, such as changing career, moving home or investing heavily. Others can be tested cheaply first.
A small test can buy information before a bigger commitment. You might run a short pilot before buying lots of equipment, or offer a sample before building a full service. These are illustrations, not evidence.
The point of a test is not to prove yourself right. It is to learn something that could change what you do next. Cheap, reversible tests can lower the cost of being wrong.
Tests still cost time, money or trust, and some choices cannot be safely trialled. Keeping every option open forever is not the goal either. Sometimes a test reveals a hidden assumption or limit, and changes the search landscape you can see.
EvidenceValue of information and irreversible choice
Howard (1966) shows that information can have value when it could change a decision. That does not make every piece of information worth collecting. Dixit and Pindyck (1994) explain why hard-to-reverse choices under uncertainty can make waiting or staged steps worthwhile. Neither means “always wait”.
Computational analogyParEGO: search when evaluations are expensive
Knowles (2006) built ParEGO, a computer method for searching when each test is costly. It updates its picture of the problem after every test to choose the next one. In computer benchmark problems it performed well on small test budgets. This is an analogy for using limited tests wisely. It is not evidence that human lives follow the method or that any real-life test will succeed.
Business exampleSolve the actual problem
According to Alan Sugar’s own account, Amstrad questioned the high cost of making satellite dishes. It treated the job as ordinary pressed-metal work rather than accepting specialist prices. This is a business example of reframing a problem, not scientific evidence.
Sometimes the breakthrough is not a better answer to the existing question, but a better representation of the problem.
WATCH THISTina Seelig: Classroom Experiments in EntrepreneurshipStanford eCornerA small-budget classroom challenge showing how cheap experiments can produce useful information before bigger commitments.Watch on YouTube Other people change the game
Your options change when other people are choosing too.
A choice that works well when you are alone can work differently when other people react to it. Imagine two small businesses on the same street. Each one's best move depends partly on what the other does.
Game theory studies exactly this: situations where each person's best move depends on what others do. It is a model of strategic interaction, not a perfect map of how real people behave.
Cooperation can create value that no one could create alone. But it depends on trust, information, incentives and whether people expect to meet again. Today's behaviour can change how others respond tomorrow.
People can cooperate, compete, copy, signal, reward or punish. So your search landscape is partly social. Changing who you deal with, how you communicate, or the incentives around a situation can change which options become reachable.
The Prisoner's Dilemma illustrates how individually tempting defection can undermine a jointly preferable cooperative result. A Stag Hunt contrasts this with an assurance problem: a more valuable joint result may require enough trust to coordinate, while a smaller safe individual result remains available. Neither diagram predicts how particular people will behave.
Research on repeated games (Axelrod and Hamilton, 1981; Nowak, 2006) and reciprocity (Trivers, 1971) shows conditions under which cooperation can last. Trust matters too (Balliet and Van Lange, 2013). Reputation carries the effect further: what others believe about your past behaviour can change later opportunities.
This offers a grounded way to look at part of what people sometimes call “karma”. Work on indirect reciprocity and reputation (Nowak and Sigmund, 2005), and cooperative cascades in laboratory games with strangers (Fowler and Christakis, 2010), shows how behaviour can have delayed social consequences. That does not make karma a scientific law. Good conduct does not guarantee good outcomes, and unethical actors can prosper under some incentives.
A helpful action might lead to a direct return, change one's reputation and thereby affect indirect cooperation, propagate through social contacts, or have no meaningful return at all. The diagram deliberately includes dead-end pathways; it does not establish a universal scientific law of karma.
WATCH THISJOHN NASH GAME THEORYOxford UnionJohn Nash uses traffic choices to show how one person's best move can depend on what everyone else does.Watch on YouTube Decide what 'better' means
Before you optimise anything, you need some idea of what 'better' actually means.
Optimising only makes sense after you have chosen what matters. A decision can look good by one measure and bad by another. A job might pay more but leave less time for family. A cheaper option might create more waste.
Aristotle called good judgement practical wisdom: knowing what fits this situation, not just following a fixed rule. Some values are also hard to trade against money, convenience or speed.
The philosopher Robert Nozick asked a simple question with his Experience Machine: if a machine could make you feel perfectly happy, would you plug in? His point is that truth, real achievement and genuine relationships may matter too, not just pleasant feelings.
Changing the goal changes which options count as good. So the search landscape is shaped not only by what exists, but by what you decide to value.
Five lenses - economic, emotional, ethical, ecological and educational - lie within a boundary of acceptable choices. Some apparently attractive opportunities fall outside that boundary because they involve exploitation, deception, significant harm or unacceptable methods. Practical wisdom (phronesis) is shown as context-sensitive judgement, not an arithmetic middle between every alternative.
One practical way to check a decision from several angles is the five-lens framework I developed for this article: Economic, Emotional, Ethical, Ecological and Educational. It is an author framework - a practical checklist, not a validated scientific scale. It does not claim that every decision reduces to these five dimensions, or that all five matter equally every time.
In plain terms: Economic covers money, resources and long-term viability. Emotional covers wellbeing, stress, meaning and relationships. Ethical covers fairness, duties and who may be harmed or helped. Ecological covers environmental effects and resource use. Educational covers what is learned, developed or made possible later.
Research on protected values (Baron and Spranca, 1997) suggests people sometimes refuse to trade certain values against money or convenience. That is a description of how people can behave, not a rule about what your values should be. Aristotle's practical wisdom adds that good judgement depends on context; his doctrine of the mean is not an arithmetic compromise and must not be used to split the difference between good evidence and bad evidence.
PhilosophyNozick’s Experience Machine
Nozick asks whether we would choose a perfectly pleasurable simulated life if doing so meant giving up contact with reality. The thought experiment does not prove what a good life is, but it exposes a distinction between feeling good and valuing what is real, achieved or authentically lived.
Cultural analogyThe Matrix
The Matrix dramatises a related question: whether a comfortable illusion can be preferable to an uncomfortable reality. It is illustration, not empirical evidence.
WATCH THISAristotle & Virtue Theory: Crash Course Philosophy #38CrashCourseA simple introduction to virtue, the Golden Mean and the idea that good judgement depends on context.Watch on YouTube Everyone has different limits
Two people can make equally sensible choices and still face very different sets of options.
Two fictional people face the same hypothetical opportunity. Person A has more time, financial cushion, mobility, energy and capacity to recover from a setback. Person B faces greater constraints on time, financial resources, health, mobility and caring responsibilities. The bars are illustrative resource profiles, not measurements, diagnoses or individual scores. No person's outcome is guaranteed.
People begin with different amounts of money, time, energy, health, mobility, responsibilities, connections and room for failure. That changes which experiments are realistic. Two equally capable people can face very different costs for the same test.
One way to picture this is a search budget: the room someone has to explore, test ideas and recover from mistakes. It is the author's practical framework, not a scientific score.
Compounding helps explain why small differences can grow. If £100 grows by 10%, it gains £10. If £1,000 grows by 10%, it gains £100. The same percentage return gives the larger starting amount a larger absolute gain. This is a numerical illustration, not a claim about every life outcome.
Advantages can accumulate in the same way, through repeated access to opportunities, information, reputation and resources. Disadvantages can build up too. Either way, people still make real choices within their constraints.
The Pareto principle describes a pattern where outcomes are often unevenly distributed. It describes that unevenness. It does not by itself explain why it happened.
This is the search landscape again. People can improve their options, but they do not all begin from the same landscape, and the same experiment costs them different amounts.
Potential opportunity is broken into occurrence, recognition, conversion and compounding. Chance can influence whether an option arises, while preparation and available resources can influence recognition or the ability to use it. A later opportunity may create further options, but those possibilities remain contingent and cannot be manufactured with certainty.
The pipeline is simple. Occurrence: the opportunity exists. Recognition: you notice it. Conversion: you can act on it and turn it into something useful. Compounding: the result helps create later opportunities. Not every opportunity reaches every stage.
Cumulative advantage
DiPrete and Eirich (2006) review how early differences can be reinforced by later access and returns, so advantages and disadvantages can build over time.
Pareto principle
A descriptive pattern in which outcomes are often unevenly distributed. It does not explain the unevenness on its own.
Search Budget in full
Time · money · mobility · energy · responsibilities · access · social permission · health · downside capacity. Downside capacity simply means how much you can afford to lose, risk or get wrong before the cost becomes serious.
This is a practical checklist, not a validated scientific scale. The parts interact, and there is deliberately no single score.
WATCH THISPareto Principle Explained: How the 80/20 Rule Changes EverythingSproutsA short explanation of the Pareto principle and why outcomes are often distributed unevenly.Watch on YouTube Go deeper: Compound interest introduction Khan Academy · See how proportional growth compounds from different starting amounts.Judge the decision, not just the result
A good decision can still end badly, and a poor decision can sometimes get lucky.
Four panels distinguish good process with favourable outcome, good process with unfavourable outcome, poor process with favourable outcome, and poor process with unfavourable outcome. Luck and unexpected events can separate process quality from outcome quality; a positive result alone does not retrospectively prove a sound choice.
Results contain luck. A careful, well-researched plan can still be undone by an accident or a delay. A careless choice can sometimes work out. So a good result is not proof of a good decision, and a bad result is not proof of a bad one.
People also judge differently once they know how things turned out. In a classic experiment, the same decision was rated more harshly after people learned it had ended badly. Nothing about the decision itself had changed (Baron and Hershey, 1988). This is called outcome bias. It describes how people judge. It does not say that outcomes are unimportant.
So review the decision as well as the result. What did I know at the time? What did I assume? What options did I consider? What would I do differently now? These questions are useful whether the outcome was good or bad.
That is how judgement improves. Celebrating wins and punishing losses teaches very little, because both hide the part luck played. You cannot control every outcome. You can improve how you search, choose, learn and update.
Outcome bias
Baron and Hershey (1988) found that people often rate the same decision differently once they know whether the outcome was good or bad. Outcomes remain useful information. They are not a perfect verdict on the reasoning behind them.
The Recursive Search Loop
A practical synthesis of the ideas in this article: values, observation, learning, testing, cooperation, commitment and review. It is a thinking tool, not a validated algorithm for optimising a life.
An illustrative loop moves through values, observation, learning, exploration, testing, cooperation, updates, commitment, review, teaching and searching again. Luck, changing environments, other people and structural constraints operate around it. The order can be revised, skipped or repeated; no predictive success guarantee or empirically validated life-optimisation algorithm is claimed.
A shorter version of the same loop: Observe - notice what is happening. Frame - decide what problem you are actually solving. Generate options - think of more than one possible move. Test - try the safest useful experiment you can. Learn - compare what happened with what you expected. Update - change your beliefs or your plan. Repeat - run the loop again with better information.
Not every problem needs every step, and not in the same order. The loop is a practical way to keep searching, not a formula that guarantees a result.
Practical conceptA friendly adversary
Ask someone you trust to try to find the weak points in your plan before reality does. That could be a colleague, a friend, a checklist or an AI system. It is a practical technique and an author suggestion, not a validated named intervention.
A traveller faces possible routes associated with knowledge, relationships, values, creativity, meaningful work, cooperation, stability, commitment and resources. Luck and unknown regions can open or close routes. The conclusion is about influencing some conditions of possible future choices while recognising that the eventual path is not fully visible or controllable.
WATCH THISWhen good decisions have bad outcomesCassie KozyrkovA clear explanation of outcome bias: a good decision can still end badly, and a poor decision can get lucky.Watch on YouTube Go deeper: Thinking in Bets Wharton / Annie Duke · A deeper look at separating decision quality from outcome quality.We cannot see every path. We cannot control every event.
But we may be able to change some of the conditions from which the next set of possibilities emerges.
Bias the search landscape. Do not pretend to control it.
References
References supporting the research, theory, philosophy and documented examples discussed in this Insight. Personal reading recommendations appear separately below.
Aristotle (2009) The Nicomachean Ethics. Translated by D. Ross. Revised with an introduction and notes by L. Brown. Oxford: Oxford University Press. doi:10.1093/actrade/9780199213610.book.1.
Arthur, W.B. (1989) ‘Competing technologies, increasing returns, and lock-in by historical events’, The Economic Journal, 99(394), pp. 116–131. doi:10.2307/2234208.
Axelrod, R. and Hamilton, W.D. (1981) ‘The evolution of cooperation’, Science, 211(4489), pp. 1390–1396. doi:10.1126/science.7466396.
Balliet, D. and Van Lange, P.A.M. (2013) ‘Trust, conflict, and cooperation: A meta-analysis’, Psychological Bulletin, 139(5), pp. 1090–1112. doi:10.1037/a0030939.
Baron, J. and Hershey, J.C. (1988) ‘Outcome bias in decision evaluation’, Journal of Personality and Social Psychology, 54(4), pp. 569–579. doi:10.1037/0022-3514.54.4.569.
Baron, J. and Spranca, M. (1997) ‘Protected values’, Organizational Behavior and Human Decision Processes, 70(1), pp. 1–16. doi:10.1006/obhd.1997.2690.
Caldwell, J., Knowles, J., Thies, C., Kubacki, F. and Watson, R. (2022) ‘Deep Optimisation: Transitioning the Scale of Evolutionary Search by Inducing and Searching in Deep Representations’, SN Computer Science, 3, article 253. doi:10.1007/s42979-022-01109-w.
DiPrete, T.A. and Eirich, G.M. (2006) ‘Cumulative advantage as a mechanism for inequality: A review of theoretical and empirical developments’, Annual Review of Sociology, 32, pp. 271–297. doi:10.1146/annurev.soc.32.061604.123127.
Dixit, A.K. and Pindyck, R.S. (1994) Investment under Uncertainty. Princeton, NJ: Princeton University Press.
Fowler, J.H. and Christakis, N.A. (2010) ‘Cooperative behavior cascades in human social networks’, Proceedings of the National Academy of Sciences, 107(12), pp. 5334–5338. doi:10.1073/pnas.0913149107.
Granovetter, M.S. (1973) ‘The strength of weak ties’, American Journal of Sociology, 78(6), pp. 1360–1380. doi:10.1086/225469.
Gunton, R.M., Hejnowicz, A.P., Basden, A., van Asperen, E.N., Christie, I., Hanson, D.R. and Hartley, S.E. (2022) ‘Valuing beyond economics: A pluralistic evaluation framework for participatory policymaking’, Ecological Economics, 196, article 107420. doi:10.1016/j.ecolecon.2022.107420.
Howard, R.A. (1966) ‘Information value theory’, IEEE Transactions on Systems Science and Cybernetics, 2(1), pp. 22–26. doi:10.1109/TSSC.1966.300074.
Knowles, J. (2006) ‘ParEGO: A hybrid algorithm with on-line landscape approximation for expensive multiobjective optimization problems’, IEEE Transactions on Evolutionary Computation, 10(1), pp. 50–66. doi:10.1109/TEVC.2005.851274.
Knowles, J.D., Watson, R.A. and Corne, D.W. (2001) ‘Reducing local optima in single-objective problems by multi-objectivization’, in Zitzler, E. et al. (eds) Evolutionary Multi-Criterion Optimization (EMO 2001). Lecture Notes in Computer Science, vol. 1993. Berlin: Springer, pp. 269–283. doi:10.1007/3-540-44719-9_19.
Lorenz, E.N. (1963) ‘Deterministic nonperiodic flow’, Journal of the Atmospheric Sciences, 20(2), pp. 130–141. doi:10.1175/1520-0469(1963)020<0130:DNF>2.0.CO;2.
Lorenz, E.N. (1972) ‘Predictability: Does the flap of a butterfly’s wings in Brazil set off a tornado in Texas?’, paper presented to the AAAS Section on Environmental Sciences, New Approaches to Global Weather: GARP, 139th Meeting of the American Association for the Advancement of Science, Washington, DC, 29 December. Original presentation (PDF)
Macnamara, B.N., Hambrick, D.Z. and Oswald, F.L. (2014) ‘Deliberate practice and performance in music, games, sports, education, and professions: A meta-analysis’, Psychological Science, 25(8), pp. 1608–1618. doi:10.1177/0956797614535810.
Nozick, R. (1974) Anarchy, State, and Utopia. New York: Basic Books.
Nowak, M.A. (2006) ‘Five rules for the evolution of cooperation’, Science, 314(5805), pp. 1560–1563. doi:10.1126/science.1133755.
Nowak, M.A. and Sigmund, K. (2005) ‘Evolution of indirect reciprocity’, Nature, 437, pp. 1291–1298. doi:10.1038/nature04131.
Pierson, P. (2000) ‘Increasing returns, path dependence, and the study of politics’, American Political Science Review, 94(2), pp. 251–267. doi:10.2307/2586011.
Rajkumar, K., Saint-Jacques, G., Bojinov, I., Brynjolfsson, E. and Aral, S. (2022) ‘A causal test of the strength of weak ties’, Science, 377(6612), pp. 1304–1310. doi:10.1126/science.abl4476. Study abstract
Sugar, A. (2010) What You See Is What You Get: My Autobiography. London: Macmillan. Amstrad’s first-person account
Trivers, R.L. (1971) ‘The evolution of reciprocal altruism’, The Quarterly Review of Biology, 46(1), pp. 35–57. doi:10.1086/406755.
Watson, R.A. and Szathmáry, E. (2016) ‘How can evolution learn?’, Trends in Ecology & Evolution, 31(2), pp. 147–157. doi:10.1016/j.tree.2015.11.009.
Recommended Reading: Books and Papers
These are personal recommendations rather than an evidence hierarchy. Some are academic or philosophical sources used in the article; others are practical, historical or culturally influential works that have shaped how I think about priorities, judgement, business, reputation and the search landscape. Inclusion does not imply endorsement of every claim in a work, and several entries are influential rather than contemporary scientific evidence.