Your friends’ mother passes away.

You find out a week after it happens.

How would you respond?

First order thinking says, “Why didn’t you call me earlier!? Am I not your friend anymore?!”

Second order thinking says, “I’m proud of you for being so strong. I’m there for you. Whatever you need, let me know”.

A fat bully is chasing you.

You reach a staircase.

Do you go up or down?

First order thinking says, “Go down! You’re faster when going down-hill.”

Second order thinking says, “Go up! Going up is harder for you - but it’s much harder for the fat dude!”.

You’re about to make an investment.

You’re young and the market is crashing. Do you choose stocks, bonds, cash or a mix?

First order thinking says, “Save! All in Cash! Bonds at the worst. Stay away from stocks.”

Second order thinking says, “People are scared, and these stocks are a bargain now. They’re undervalued, and I’m young. I can absorb the risk. Let’s go all in to stocks.”

The first level is the surface. Everyone sees it. Some observe what happens on the surface. A few of them have models to explain things, based on what they saw happen in the past. An even fewer number know their surface level models are wrong.

Things might look the same on the surface. But if they aren’t the same below, what happened last time will not happen this time.

In the physics world, this is like knowing the position of everything. You know exactly where everything is at the given moment. But how do you figure out where they will be a second from now? A day from now? A year from now? You can’t.

But if you know the velocity of the objects – the speed at which they are travelling, and the direction in which they are travelling – then you can tell precisely where everything will be a second from now. This is the second level.

Here’s another example. You have a photograph of runners in the middle of a 100m dash. That’s Usain Bolt below, the fastest man on Earth. Who will win?

Usain bolt and other runners

Source

Well, in this race, he came in third. What did you decide?

I like to call these decisions snapshot decisions. These are decisions made just by looking at the surface, a snapshot, instead of what’s underneath.

We make snapshot decisions frequently, but we seldom have to make snapshot decisions. We have the opportunity to collect as many snapshots as we need, making an “informed” decision in the end.

Collecting snapshots helps because we can figure out the velocity of an object by observing it over time. Doing it for everything in the universe is unrealistic, but we can do it for a few objects – the ones that catch our interest.

It’s not easy, though.

What if there’s acceleration in play too? That’s the third level. Not only is the position of the object changing, but its velocity is changing as well.

How are you going to figure things out now?

What if there are collisions between objects? Collisions that change their acceleration and velocity?

Thinking about this can make heads explode pretty soon.

In physics land, this is a solved problem. We had a great mind, Newton, give us a model that takes in all the signals and discards all the noise. The model? The famous laws of motion. These give rise to Newtons equations, which explain where every object will be at any given time. That’s how we predict the position of the sun, moon, earth and stars so accurately!

Everything isn’t as easy as physics though. We don’t have models to predict the future of humanity.

Our brains face exactly the same problem1. There’s way too much information going around - much more than our brains can handle. So, it filters.

Over time, the brain learns to pay attention to specific things and discard the rest. We choose these specific things by their ability to explain everything going on around us. It’s the signal vs the noise, and everybody gets to chose their signals. This choice is what determines the quality of your thinking.

The brain is pretty easy to trick though. That’s why we have a million cognitive biases - things that can trick the brain.

If you hear about an idea – like the prices of houses will always go up – and you see the prices going up, you’ll start accepting that as a blind truth. That’s confirmation bias. Of course, it all comes crashing down with the housing crash.

All this time, I’ve been establishing how freaking difficult second order thinking is. Charlie Munger, Warren Buffets’ long time silent partner agrees.

[..] It’s not supposed to be easy. Anyone who finds it easy is stupid. – Charlie Munger

I think there are at least three tools to help with second order thinking.

  • Probability
  • Thinking in systems
  • Psychology

Probability

Probability works when you can exhaust your entire state space.

Have you seen someone play rapid fire chess? I find it pretty amazing how they make their next move – even before their opponent has completed the last move. They anticipated the move!

Except, that’s how it looks like. That’s not how it is.

They’re not predicting the opponents moves. They’re thinking through the list of probable moves the opponent can make and coming up with counter moves. If you’ve thought through all your opponents moves - no matter what they do - you know what to do next!

This appears like predicting the exact move.

That’s how AIs play chess. Instead of looking just one move ahead, it can look 10 moves ahead. No matter what you do, it has thought it through. This is incredibly taxing for a human mind to do. But, easy enough for your mobile phone to do it. The state space of Chess is small.

The catch? Outside of Chess, you can’t always predict all moves.

That’s why we have chess-AI and no human-AI yet. Humans are unpredictable. You can’t look at someone and say, “they are going to do one of these things.”

Thus, probability works when you can exhaust your entire state space. The idea to predict the future is simple: Go through all possible moves you / others can make – and come up with a counter measure for them. Beware, even this isn’t easy to do. Keep in mind first order positive and second order negative decisions.

Computing counter moves for every possible move is taxing. So, just like in the physics example, people choose the most probable moves. This is easy to do now, because you have a probability for all the moves. Over time, this becomes subconscious. You aren’t calculating probabilities every time. You generate intuition instead. Some people like to call it experience.

I like to call it understanding a system.

Thinking in systems & Psychology

Probability doesn’t work with complex systems because we have no idea about the range of possible outcomes.

Since we don’t know all possible outcomes, assigning an exact probability is meaningless

That’s what a “Black swan” is. An unexpected outcome. Something that never happened before, staring us right in the face - and blowing away all probabilistic models and thinkers.

Failure to use higher-order thinking when considering outcomes in complex adaptive systems is a common cause of overconfidence in prediction making. - Shane Parrish, on systems

Systems thinking and Psychology both try to explain how systems behave. That’s what we want - a model to help us explain what happens.

I think Psychology is a specific application of thinking in systems. The human system dynamics are so complex that we had to create this specialized branch just for ourselves.
Or maybe we are very self indulgent.

Our main goal with systems thinking is to build a model to explain the system.

We have our building blocks.

For systems, there are archetypes.
For Psychology, there’s the Psychology of Human Misjudgment.

Here’s one of my favourites - the power of incentives.

Marketing & advertising firms use psychology to get us to buy stuff. I think that’s where the best psychology research occurs, because the incentives line up. The more you can figure out people, and what makes them tick, the more you’ll earn. I’m looking for friends in marketing and advertising - just to get a hand on their psychology models.

There’s another example below

How do you hone these skills? Get better at recognising archetypes? And analysing systems?

The short answer, Practice.

Nope, there’s no magic bullet and I haven’t spent time trying to find one.

In system thinking terms, you need to tighten your feedback loops. If there’s lots of delay between your decisions and their outcomes - you wouldn’t know what really caused the outcome. Look at policy makers. If the policies come back to bite us in 5 years - when the original decision makers are gone - you bet there is no learning happening. On average, you can expect the next policy to be just as bad.

The second answer is facing the same situation again and again. You get to make different decisions again and again - and figure out the best ones. This helps understand the structure better. If you don’t play a repeatable game, you can’t improve your decision making.

Some fall in the trap of intuition. Generating intuition here is similar to the Probability approach. You see what you’ve seen again and again, and over time you come to believe that’s what will always happen. This is flawed – we never knew what all was possible in the first place. If we knew, we’d go back to the probability approach.

When people say you need to have lots of experience to be great at this, or that you need to have intuition – I think what they mean is you need to know how to predict what will happen next if you do X vs Y. This builds up over time as your priors – How often have you made the decisions and what came of this decision. Hence, I think understanding the system is the best kind of intuition.

This is one way2 to improve decision making - playing a repeatable game, where you get to make the same decisions again and again, get feedback quickly, figure out where you went wrong, and self correct.

More on how to understand a system.

Going beyond

Second-order is one step into the future. The principles are the same, if you want to go one level deeper, or eleven levels deeper. The deeper you go, the harder it gets. Anything after the second level in life becomes too uncertain. You’re multiplying small probabilities by smaller probabilities. This leads to even smaller probabilities for the events you’re considering. We get a long list of events that might occur, but we forget how small the probabilities are. This leads to analysis paralysis.

You want to find a middle ground - which is usually second order thinking.

“Failing to consider second- and third-order consequences is the cause of a lot of painfully bad decisions, and it is especially deadly when the first inferior option confirms your own biases. Never seize on the first available option, no matter how good it seems, before you’ve asked questions and explored.”
– Ray Dalio

Second level problems

It’s not just your decisions that need second-order thinking. It’s your problems as well.

“We can not solve our problems with the same level of thinking that created them” ― Albert Einstein

The way I see it, there are three levels.

  1. The first level problem - the symptoms
  2. The second level problem - the incentives
  3. The third level problem - the values

You can’t focus? Trying hard to focus is fixing the symptom.

We focus on fixing the symptoms. But, the patches don’t last.

The problem resurfaces again because the second-level, the incentives are still untouched.
It’s like weeding your plants. But, instead of uprooting them, you’re using a lawn mower to cut them. The roots are still there, they are going to grow back.

A more relatable example - from the Psychology of Human Misjudgment.

FedEx faced a peculiar problem. They wanted to deliver packages on time. For this, they needed to load their airplanes before sunrise. The catch - they were never on time. The night shift workers were paid by the hour, and expected to work the entire 8 hour night shift. The idea was, if they spend enough time on the shift, they’ll get the job done. (As some managers still think - in places with fixed work hours)

Finally, somebody got the idea that it was foolish to pay the night shift by the hour. The employer wanted rapid loading of a plane, not maximized billable hours of employee service. Maybe if they paid the employees per shift and let all night shift employees go home when all the planes were loaded, the system would work better. And, lo and behold, that solution worked.

Focus on solving the second and third order problems.

“Show me the incentives, I’ll show you the outcome” - Charlie Munger

I’ll save writing about the second and third level problems for a separate post.3

  1. Sensory gating ↩

  2. I wouldn’t say the only way, because I don’t know the range of possible ways. ↩

  3. I’m so glad I managed to finish this post without quoting Howard Marks. Every google search on this topic is about “what Howard Marks said.” He did do a good job of introducing the concept in The Most Important Thing - but now it’s up to us to stand on the shoulder of this giant and move forward. ↩