maxims for Thinking Analytically by Dan Levy

This means that we can vet the strength of our (or someone else’s) arguments by taking it to the extreme and seeing if it still works. Of course, “extreme” refers both the upper and lower extreme.

For example, many people think that EVs are better for the environment – you can test the validity of this argument by thinking through cases where only EVs are allowed to ply on the road and all ICE vehicles are banned.

Such a world will have to deal with newer environmental problems like battery disposal.

And when natural experiments are available, we can take the corollary of this maxim: Learn from outliers. Because if extreme cases and outliers can work, then you can be sure that business-as-usual (BAU) cases will too.

For example, when a barefoot runner from the Tarahumara tribe in Mexico outruns sponsored elites, it tells us something profound: success doesn’t always require polish – it just needs to be possible. Outliers stretch the bounds of plausibility. If it worked in the wild, maybe it will fly in the lab.

We can also use this maxim to understand what the real problem is.

For instance, everyone says they want to live long, but if you do a imaginative exercise where you gain immortality, then you realise that without something meaningful to do or having a healthy body to do it with, immortality is not much use.

The real problem to solve first then, is not to add more years to your life, but to find something useful to do with the years you already have.

Important for me to point out here that while going to the extreme case, one must be careful of the “Slippery Slope” fallacy.

“Going to the extreme” is a short trip you take to an extreme world to see if the problem is made more clear there, the “Slippery Slope” assumes you are going to become a resident of that world.

Taking an argument to the extreme doesn’t always invalidate the moderate position.

“If we legalise marijuana, next we’ll legalise heroin” is a poor extreme-case argument because you’re assuming marijuana equates to heroin in everyone’s eyes. Extreme cases are supposed to be used to get a handle on the problem, not substitute it for something else.

Also important to note that some phenomena behave nonlinearly. Doses matter. Extremes can trigger emergent behaviours not present in everyday situations (e.g., group panic, market crashes).

Finally, “Outlier logic” can flatten real suffering. E.g., “If Elon can work 120 hours a week, anyone can” ignores privilege and survivorship bias.

This maxim is not an opposite of the previous one, but instead is an alternative perspective with which to look at the same problems. Get the problem down to its most basic, simple version; full understand that version and then add complexity slowly to get to the real life problem, stopping at each stage to see if you still have traction.

For instance, to tackle the difficult problem of managing traffic in a major city like Bangalore, start with a simple case of how to best manage traffic on a single street. What essential things would you do to that street to make it more traffic friendly? Separate walking tracks, zebra crossings, dividers, bumps at crossings? Add complexity from there.

And similar to the previous maxim, you need to remember that your goal is still the more complicated problem.

You’re just simplifying the problem to gain traction in a systematic manner.

Beware of the “Substitution Effect” here. Mastery of a simple case can create the illusion that you’ve mastered the full complexity. But adding scale, time, or interdependence can produce entirely new behaviours (see: chaos theory, emergence).

And on a related note, remember that not all problems are practically reducible. Some systems are complex by nature, not by accident. Reducing them loses essential properties—e.g., climate models, ecosystems, macroeconomies.

This maxim reminds me of Occam’s razor. You should add complexity once you have a good handle on a simpler case. Adding complexity to your model without first having a strong grasp of a simpler model will just make your resulting model even more intractable.

This is why XAI (explainable AI) is important. – I remember Narayanan and Kapoor’s “AI Snake Oil” where they advise that the Predictive AI model should be fundamentally explainable by the sales man, otherwise you might be buying vapourware.

Note that a model can still output results that are counterintuitive, but the builder should still be able to explain the unexpected result in plain English, using analogies if needed (which is the next maxim).

Complexity is a necessary and indeed, natural feature of real world systems, and you cannot escape it if you’re trying to deliver real world results, but if you find yourself taking refuge in complexity, I have some bad news for you: Your fundamental understanding needs work.

Even if you have a strong grasp of the fundamentals, you must realise that you audience may not – In such cases as well you need to leave behind the complexity.

Domain experts are generally warned about this as they may forget to simplify a topic enough for their audience. You must remember the goal of communication to is create comprehension not confusion.

Like understanding international negotiations by considering negotiations between parents and children, or between siblings. What happens when one child is rewarded for tantrums? You get North Korea.

Of course, analogue also help communicate complicated concepts to lay people. It’s a fine balance, some people dont like simplifying concepts or giving analogies because they fear it loses its nuance.

Which is true, but that is also the point – you have to make sure “affordable” concessions to be able to get your point across. RBI changing interest rates can be compared to adjusting the thermostat – small tweaks to create stability over time.

It’s not that unprecedented events (like Covid, or DJT’s re-election) are happening more frequently these days but rather that the number of “things” happening in our ultra-connected and fast moving world (now even more so with AI) are so many that even a rounding error, 0.001% chance of a freak event pretty much guarantees that you will experience several of them in your life.

And the more moving parts in an event, the more uncertain it is – traditionally this is where wars and pandemics live.

The author gives an amusing example of a company that decided to end its contingency plans on the date its analysts expected the 1967 “Arab-Israeli Six Day War” would end. It didn’t, and the company was left scrambling.

Life is uncertain, more than you think, accept it and deal with it. This is the first maxim in this section.

In the strictest philosophical sense you can only prove correlation and never causation, certainty is an illusion, the universe was not designed this way.

Very good. Now what?

Well, to start, knowing this should make you a better planner, and you’ll hopefully be less cocksure.

That your plans can never be perfect but only good.

That you should never go all-in on anything unless you’re willing to go down with the ship. Reminding yourself about the “Planning Fallacy” and “Hindsight Bias” is also helpful.

Helpful too, is knowing the difference between risk and uncertainty.

See the table below – a situation will be called “risky” when you know the possible “end-states” of the world (such as you know that a roll of a die can only have six end-states), and the probabilities of each state occurring.

But it will be called “uncertain” when you know the possible states but dont know the probabilities of each state.

BTW, you will be called “ignorant” about a situation when you don’t even know the possible end-states that the world can end up in. And I suppose your ignorance will hit a crescendo when you don’t even know that you dont know the possible end states of the world – meta-ignorance.

Because there is no certainty, force yourself to always think in terms of probabilities. Previously where you used to be certain, give it a 95% or 5% probability score, but never give anything a 100% or a 0%.

Our brains evolved to prefer certainty since eases decision making and is more energy efficient, so you’ll have to fight against this evolutionary conditioning. Society might also consider you as someone who is never sure about anything.

But the reward is worth it in terms of intellectual humility and rapid pivoting due to disconnection of ego from public position (much easier to switch away from something you were 99% sure of versus something you were 100% sure of).

Zeck suggests that you build this like a muscle by exercising it daily in mundane situations, such as when you’re driving a car, you can ask your co-passenger what they think is the possibility is of the upcoming stop light turning red by the time you reach it.

Even while interacting with others, realise the value of asking probabilistic questions versus asking binary ones. Do not ask not whether or not a consequential event will or will not happen, because that will lead to a binary answer you can’t do much with.

Instead, ask what is the probability that one outcome will happen or the other, which will give you a percentage (or if you’re respondent is astute, a range).

Not only does this force your respondent into thinking more critically instead of forcing their thoughts out of a narrow yes/no mould, it also means you can then build on their response to ask follow up questions.

Also very important is to update your priors (Bayesian probability). Also, know that all information is not equally useful, some kinds of information are more useful in certain contexts – for instance, as a parent to a young child I am always concerned of her safety, but I don’t get up from my seat every time I hear her cry (because that is so very often!), instead, whenever I hear her cry, I wait for her mother’s voice – If her mother’s voice appears calm then I continue working, but if she sounds concerned, I bolt.

In “Superforecasting”, Philip Tetlock shows that the best predictors don’t just assign probabilities – they constantly revise them, and express them with ranges not fake precision.

When life is going on, esp. when it is going not very badly – we tend to not want to rock the apple cart. This is the “Status Quo Bias“.

But even uncertain times, we tend to continue behaving in ways we’ve always had.

Being aware of this bias can budge you into action when the time calls for it, it can also help you design such default options that take advantage this bias – such as defaulting to “Yes” the organ donor section in driving license applications.

When it comes to decisions, the first thing Levy advises us is to know the difference between decisions and outcomes.

A bad decision can lead to a good outcome if you’re lucky, and vice versa if you’re not. During good times even your bad decisions may work, while during bad times even your good decisions may fail.

Often people tend to assess the quality of the decision they took basis the outcome (known as “resulting”), intuitive but wrong.

Why is it wrong? Because you should compare apples with apples. And decisions and outcomes are not the same thing.

Levy shares this handy table to explain what being “lucky” and “unlucky” mean in the context of decision making:

Humans tend to carry more regrets (at least in the short term) over errors of commission than they do over errors of omission:

• Errors of commission: Doing a thing that turns out wrong.
• Errors of omission: Not doing something, and the failure to act causes harm/missed upside.

Levy wants us to realise that while we may not weight them similarly, we should try to.

Kicking yourself over a decision you took that turned out to be wrong, while being unaffected when the one you sat out turned into a blockbuster may seem intuitive, but it is wrong.

Why is it wrong? Because you should compare apples with apples. And outcomes and feelings are not the same thing. The outcome is about the objective difference in your cirsumtances before and after, while feeling is your subjective experience.

If an act of commission and an act of omission result in the same objective situation then both outcomes should be treated similarly.

For example, buying shares that gave you a Rs. 10000 loss, and not buying shares that rose by Rs. 10000 in value.

But usually we will overweigh one and under-weigh the other (feeling regret versus not having a reaction) – I’ve tried to explain using the table below.

Information should be treated like a utility. In the more practical domains of life, such as health, careers, finance, it is rare that we are just waiting for the information alone.

For instance you get your annual health check-up done so that you can know which biomarkers are within range, what need to be monitored, and what need immediate intervention.

Or, you wait for the closing price so that you can take a decision on whether you should sell a stock. Or, you read a non-fiction book so that it can help you take better decisions in a particular domain. Even in areas such as stories and novels, we read them, essentially consuming information, so that we can feel a range of emotions.

This does not mean that everything we do is a means to an end, consider for instance, the love a mother showers on her child. But yes, when it comes to most things in life, and definitely when it comes to information – We need to be mindful of what the deal is.

Levy points out in this maxim that if you’re waiting for a piece of information, you must be able to clearly articulate how you are going to use that information for at least one materially different end versus had you not had that information.

If the awaited information does not have the power to change your decision, give you more options, or at least, reaffirm the one you are thinking of making – then you should not wait for it.

Often known as “cost-benefit analysis”, this maxim is another tool for decision making, especially when you have choices. Of course, this requires standardising the input (the cost, time, effort, resources etc.) and the output (the revenue, benefit, savings etc.) to a common metric (”… output per unit of input.”).

For instance, I’ve often made trips from Chandigarh to Delhi, a distance of about 250 KM, and have often thought whether I should take the car or go by train.

Performing a cost benefit analysis gives a clear answer: If I am the only one travelling, I should go by train, but if even one more family member is travelling with me, then the both of us should take the car.

The common metric in this case is “Rupees per KM”.

Taking the train costs an even Rs. 2 per KM per passenger. While taking the car costs a fixed Rs. 4 per KM, no matter the number of passengers.

Ergo, if I’m the only one travelling, the train with its Rs. 2 per KM charge is the right option. But if two of us go by train, the cost for us becomes Rs. 4 per KM (Rs. 2 per KM x 2 passengers), at this stage wed rather have the comfort and convenience of using our own car.

And with 3 passengers or more, the car is even more attractive than the train.

Opportunities for such analysis are everywhere in life, even though the “right answer” may be elusive due to confounding variables, it is a very useful tool in the strategist’s belt.

For instance, if you’re looking to evaluate between social media campaigns in terms of reach, just look at the “shares per impression” metric, this will give you a much better answer than absolute counts of impressions or likes and such ever will.

Figure out “… some metric of “bang per buck” for different options, ordering these, and then choosing the options with the highest bang per buck until the bang becomes worth less than the resources required, or you simply run out of resources.”.

This maxim reminds me of my school days when I was taught the concept of marginal utility.

Basically, the utility of any good (like cookies!) is not fixed, but rather depends on its priors – that is to say, the benefit you will derive from one cookie will depend on how many you’ve had till then. If you’ve not had a cookie in ages, then it’ll be a great experience, and if you’ve just had twenty, it might just be one cookie too many.

For most things in life, marginal utility/cost/benefit falls with every incremental thing (that is why it is called “marginal”, the margin is a single additional unit of the thing).

This means that the first of anything is likely to be really costly/useful/beneficial/enjoyable/painful (depending on the case).

For instance, when an automotive factory produces its first ever first car, that car is several orders of magnitude more costly than the price that the car maker is charging for it, and indeed if the car maker were to only sell a single car from its factory, it will go bankrupt. It is due to diminishing margins than while the cost of producing the factory’s first car may be Rs. 100 Crore, but the cost of producing the first two cars will, counterintuitively (but unsurprisingly), be Rs. 50 Crore and change.

And so over the course of producing thousands of cars the automaker is able to turn a profit, because each incremental car produced does not cost as much as the one before it.

Thinking in terms of marginal cost/benefit also applies to personal life, for instance, if you’ve just clocked a 12 hour work day, spending another hour will hardly be as rewarding to you as spending an hour with the kids. It’s simple marginal utility calculation.

Or in the case of fitness, first hour of movement = high gains. Adding another hour? Diminishing benefit. Meanwhile, the marginal benefit of sleep or recovery might exceed the extra reps.

That is the next maxim that Levy is drawing our attention to. And this one applies to decisions we make in more formal settings, where we are taking decisions for other people, where we are designing products, programs, or policy for a large body of customers (or citizens).

Designing for the average customer or citizen may feel like a theoretically equitable approach but can get practically useless if we’re not careful.

Hence, Levy advises that we break up the heterogeneous block into homogenous sub-blocks. In the business world, this is known as segmentation.

Once this is done, you can then proceed to make the specific product, programs, policies and yes, uniforms, that will actually fit the people in question. Real humans are distributions, not datapoints—and solutions must respect diversity, not collapse it.

A real world example: Cockpit Design. In the 1940s, U.S. Air Force cockpits were built for the “average pilot.” Crash rates were mysteriously high. Turns out no pilot actually matched the average across 10 dimensions (arm length, torso, seat height, etc.).

Solution? Adjustable cockpits.

Important caveats:

• Designing for the average isn’t always useless – when variance is low or the cost of customisation is high, it might be an acceptable proxy.
• But the critical question becomes: Where does variance matter? That’s where the average becomes misleading.

In this maxim Levy is telling us to not shy away from collaboration, especially with those who bring a different but complementary skill set to ours. Such as accountants and marketers, biologists and mathematicians, poets and coders.

It shifts the question from “What am I good at?” to “What can I be even better at when paired with someone different?”

Don’t just seek collaborators who are similar to you – seek those whose strengths fill your blind spots and multiply your output. When distinct skills mesh, the whole becomes greater than the sum.

Most people gravitate toward those who think like them – for comfort, ease, speed. But high-leverage collaboration comes from difference, not sameness.

This is how Spotify, Pixar, Apple, and even the Apollo mission happened: cross-pollination of experts from different worlds.

Levy is channeling a well-known result in production theory, called the optimisation principle: when inputs are complementary, returns increase super-linearly. E.g., 1 unit of skill A and 1 unit of skill B yield more than 2 units of either one.

This maxim is also a call to read and explore widely than just be restricted to your domain. Complementarity applies to bookshelves just as much as it does to boardrooms. It let’s you:

Borrow tools from one domain to solve problems in another. E.g., using Bayesian reasoning (statistics) to navigate personal uncertainty (psychology/ philosophy).

Develop second-order insights. When you read both Nietzsche and behavioural science, you begin to ask questions like: Is the Will to Power just cognitive bias wearing a cape?

Spot hidden analogies. A startup founder reading ecology sees that businesses are not machines – they’re ecosystems. That shift changes how they think about resilience, fragility, competition.

Escape intellectual inbreeding. Over-specialisation breeds echo chambers. Wide reading breaks that. Complementarity thrives on cognitive diversity.

If you’d have told me this a decade ago, when I was just a few years into my job, after having spent time at a business school, I’d have laughed at you.

I’d have said something like, “Are you kidding me old man? It’s a dog eat dog world out there. If I start considering my friend’s success as my own – even helping them achieve it, ensuring they stay on track and so on – then before you know it, they’d have become company vice president and I’ll still be selling soap in the streets. No, I cannot have it that my peers achieve something and I do not.”.

Now that I am a little older, I do not think this way. In fact, I am leaning more towards this maxim.

Besides the obvious benefit of personally knowing someone who just got more successful, your friend’s success can also act as a template, or if nothing else, inspiration for your own success. If you’ve helped them along the way – you can their goodwill and reciprocity. And I feel those who genuinely celebrate other’s success signal to them and the rest of the world a sort of “abundance mindset”.

Everyone likes such a person and wants to work with them, it should not be surprising then if they too win accolades down the line.

It is said that “the envious main pains himself as though he were the enemy” – absolutely true, envy only serves to distract and depress.

And those who feel envious inside but put on a charade of support not only fool no one, they also become lose a little self-respect as to not even have the courage to speak their mind.

Levy is telling us here to learn the lesson from mistakes, both from errors of commission and errors of omission, and once you have done that, move on. Regret is meant to be processed, not prolonged.

The brain uses regret to update models of the world. Once the lesson is extracted, continuing to stew in it is a form of emotional self-harm – not virtue.

Easy to say, very hard to do – we tend to ruminate over a bad decision, nay, bad outcome for days or even months after the fact.

The corollary to this maxim is that one should not take pride when a good outcome results from a bad decision, like when a penny stock gives you an order of magnitude return, investing in penny stocks is still a bad idea for most people.

Taking pride in good luck is just as irrational as taking shame in bad luck. Both obscure the quality of your decisions. Levy also adds that people will go to extreme lengths to avoid regret, which often leads to the sunk-cost fallacy, another thing we should steer clear of. That’s not rational analysis – it’s emotional anchoring. You’re throwing good energy after bad, just to avoid the sting of a prior mistake.

This is a very interesting one. Basically what Levy is saying is that, from a purely utilitarian point of view, you may get a more juice from anticipating an exciting event than you may get from the actual event itself.

For example, imagine that next month you’re going on a week long vacation to an exotic foreign land you’ve never been to, and are planning to try out lots of exciting things.

Levy suggests that simply by day dreaming about the vacation, about all the fun things you will do, and how great the weather will be, and all the breathtaking places you’ll see, and all the people you’ll meet – you can, well, essentially feel a lot better about your life here and now even if the actual vacation turns out to be a dud.

One part of me thinks that this is unfair, but then again, if I think about it – there’s nothing wrong here.

If maximising happiness is your goal, then by all means you should day dream about a vacation for months, juicing all kinds of happiness from it, while the actual vacation will likely not be as great and even if it is, will be over very soon.

This approach can be applied backwards as well by remembering happy events.

Your mind is not bound by time. So why not harvest joy not just in the moment, but before it—and after it? Anticipation is a force multiplier for happiness.

The opposite is also true, overthinking about an upcoming challenging event is not very useful, because you’re taking away from your happiness in the now, while the event will take away from your happiness in the future.

And ruminating over a unhappy memory is also bad (in addition to the points mentioned in the maxim on regret), because the event took away your happiness in the past, and by remembering it, you are taking away from your happiness in the here and now.