Part 1: Fragility, Robustness, and Antifragility

“Wind extinguishes a candle and energizes fire. Likewise, with randomness, uncertainty, chaos; you want to use them not hide from them. You want to be the fire and wish for the wind.”

Nassim Taleb, Antifragile

This material owes an enormous debt to Taleb’s five-volume work called the Incerto. You’ve probably heard of many of the books that belong to this series, most famously The Black Swan (Taleb coined the term as we use it in risk), Fooled by Randomness (required reading for anyone I train in advertising/analytics), and Skin in the Game. I commend them all to you.

Taleb refers to himself as a “probabilist.” He began his career as an options trader before becoming more broadly interested in the impact of probability across disciplines. He specializes in domains that are defined by rare, high-impact, unforeseeable (i.e., “Black Swan”) events.

We’ll be focusing mostly on the book Antifragile: Things that Gain from Disorder, from which the above quotation comes. Taleb’s observations about the nature of fragile and antifragile leave us a trail of clues as to how we might leverage the concept for business.

What Do We Mean By “Antifragile”?

I’ve been throwing around this concept of the “antifragile startup,” but now I will lay out this concept more precisely.

To get at the concept of antifragility, we start by reexamining what it means for something to be fragile, and then ask ourselves, “What are the opposites of these fragile qualities?”

We all have an intuitive sense of what it means to be fragile, but the concept is hard to pin down with precision. We might all agree that a china teacup is fragile. But compared to what? A brick building? Sure. An egg? Even still perhaps. A house of cards…?

And why do we need this special word, “antifragile”? Why isn’t it enough to just say “strong”? Why isn’t it enough to just say “robust”? Aren’t we simply talking about something that survives stuff? Isn’t that the opposite of whatever “fragile” is?

The Opposite of “Fragile” Is Not “Robust”

People really only imagine two possible outcomes from coming into contact with a stressor:

  1. Surviving the stressor (robustness)
  2. Not surviving (fragility)

Generally, they do not consider the possibility of a third outcome:

  1. It grows from overcompensation to a stressor (antifragility)

If you think about it, robustness (the mere survival of a stressor) in not the true opposite of fragility.

A fragile thing takes on harm and damage when stressed; its true opposite would not merely be resistance to damage, but positive improvement from the stressor.

We see daily examples of such phenomena; the easiest example that comes to mind is muscle growth through resistance training. You apply stress to your muscles, and instead of breaking completely as a teacup would, they break a tiny bit and then grow back to a state that is stronger than their initial state.

Normally when we think of stressors, we think of things to be avoided. We think of them as nuisances at best, destructive threats at worst. We try to dodge them or circumvent them.

Stressor-impelled growth is counterintuitive, so it captures the imagination.

We must train ourselves to think of the stressors of the market the way a resistance trainer thinks of weights. We should understand them as information, and opportunities for growth through overcompensation.

Why Are Startups Still So Fragile?

A question for you: Why, in the age of cheap productivity technology, are startups still so fragile?

In 2020, Wells Fargo sponsored a Gallup poll reporting that 73% of startups polled received no funding for their startup and that the average startup used less than $10,000 of launch capital.

Over the last 20 years, cheapening productivity technology has decreased the personal financial risks to founders, and thus made it easier to start a business. So why have startups not outgrown this reputation for being long-odds bets?

When I ask live audiences this question, their answers commonly include:

  • Lack of knowledge
  • Stuck on their vision
  • Ignoring “digital evolution”
  • Inability to pivot
  • Not answering the core problem
  • Challenging the status quo
  • Innovation is too early/too late
  • Lack resources

These are all great answers, all nibbling around a single core truth: The key reason why startups are still so fragile is because of a core disconnect between expectation and reality.

Founders’ Harsh Reality

Startup expectations vs reality
Startup expectations vs reality

We have been surrounded all our lives by institutions that provide well-trodden paths to success.

When we went to school, there was a certain prescribed path towards getting good grades. We didn’t always follow it, but we knew more or less where it was.

We knew it had something to do with studying, doing your homework, preparing for tests, perhaps seeking a tutor. Generally speaking, if you wanted your grades to improve, there was some kind of yellow brick road to which the Munchkins of the educational system could point you.

And if you were able to hold to that straight-and-narrow path, always putting in the work, and putting in extra work when required, you might expect to see your grades improve.

Likewise, most corporate environments boast some “path for advancement” that managers are supposed to lay out for us. It usually involves completing high-quality work, collaborating well, showing mature judgment, bribing your boss (kidding), and all other manner of gaudy displays of loyalty and conscientiousness.

And typically, during the high-stakes poker game of the “year-end review,” you stare down your manager with your most imperturbable game face and say, “I have done A, B, and C as you have laid out for me. I will now take that promotion to which I am entitled because I have met your pre-set goals.”

When Institutions Are No Longer There For You

Along most of our paths through life, we are surrounded by organizations and institutions that shield us from risk.

That’s part of the reason that we build institutions in the first place: to give us some bulwark against the hurricane gusts of reality’s chaos, and the raw forces of the unbridled market.

This is the key: sometimes when founders try to deploy startups, they carry with them the soft, breezy, linear expectations of the institutionalized world.

They can grasp the journey won’t be easy, but they still harbor a covert belief that the path to success will be, like other paths they’ve stepped, both obvious and linear.

Oh, they may say to themselves, “Sure this business might not grow quite as fast as I want it to, but if I do all the right things, if I read the right books on Lean Startup methodology, if I make it through a good incubator program, if I fund successfully, I will be fully entitled to receive the success and money that is due to me.

Then they then step out of their front door and whoosh! They are promptly sucked into the tornado of unfamiliar forces against which they no longer have shielding.

When you found a startup, you step onto a new planet the events of which no longer conform to a nice, predictable bell curve. Things that can happen to you can be much more extreme, for good or for bad. A single press mention or a newly emerged close competitor could instantly change the trajectory of your business.

Becoming “Long” on Volatility

Taleb’s first key insight on fragility came when he realized that concepts like luck, chaos, entropy, providence, randomness, “perfect storms,” prediction error, and disorder are actually all synonyms. They all reduce down to the same single concept: volatility.

Taleb began his career as a trader, working every day with a form of mathematical volatility that dealt with deviations from a certain expected asset price.

But what he realized is that the rules of volatility cross domains. Another way of saying “volatility” is “deviation from expectation.” This is fundamentally what’s screwing up a lot of startups: they rely on everything going to plan and are caught flat-footed when the outcomes are not at all like what the founders thought they’d be.

We always expect the golf ball to go neatly down the fairway. We rely on the ball going down the fairway. We/ve already planned our next shot ahead of time because we anticipate that the ball will land perfectly on the fairway. Then we spend most of our time hacking around in the rough, wondering what the hell happened.

But the domain of startups is much more volatile than the secure, institutional world to which we’ve become accustomed, and single events can have much more extreme impacts. We fail to understand that we’re playing golf in a hurricane.

It All Comes Back To One Thing…

When, when you say you are fragile to something, what you are saying is that it is something that does not like volatility.

In finance terms, it is “short” volatility or short gamma. It has more downside than upside from volatility.

If you have a china teacup sitting in a cabinet, the cup does not like you taking it out of the cabinet and waving it around (introducing, for example, physical volatility). The teacup much prefers being in the cabinet, reposing, having nothing happen to it.

Why? Because it has more to lose than to gain from introducing that form of physical volatility.

That’s an important concept with a ton of implications that we will explore. But for now, think about where your business is situated right now, what your value proposition is, where it sits in terms of liquidity, partnerships, workload, pressure, competition, and market fit. See if you can identify those things to which you might be especially fragile. Think about the things that really might sink you if they don’t go as planned.

Remember always that to be fragile to something is to have more downside than upside if events deviate from expectation.

A Million Little Pebbles

One large boulder vs one million little pebbles

There was once a king who one day grew very angry with the son. At the peak of his rage, he swore that he would have his son dashed to death with a giant boulder.

Once the king calmed down, he realized he was in trouble because a king cannot go back on his word. It’s an indicator that he is unfit to rule.

This made for quite the predicament: How could he keep his son alive, while at the same time not go back on his word and thereby fragilize his rule?

One of his advisers, a wise old man, came up with a solution. The king did in fact hurl the large boulder at his son. But before he did that, he had the boulder smashed into a million little pebbles. And simply flung each little pebble at his son one by one.

The point is important and, if you think about it, rather counterintuitive.

It makes intuitive sense that if you pelt somebody with little pebbles, it will not do the same damage that a large boulder would, but why doesn’t it?

Stress is, after all, cumulative. Smack a china teacup against a table, and it cracks a little. Repeat with the same force, and it cracks a bit more. Repeat again and again, the cup eventually breaks.

So why didn’t the million little pebble pelts somehow add up to the same harm done by one huge boulder?

Fragility is Accelerating Harm

In any domain, from options trading to china teacups, the relationship between any stressor and its impact is non-linear.

Accelerating increase of harm in proportion to stress
Accelerating increase of harm in proportion to stress

For anything that absorbs an impact from something, the harm does not just increase in direct proportion to the stressor; it increases in accelerating proportion.

If I show you a platform three feet off the ground, and I ask you to jump off that platform onto the ground, what’s the harm to your body? Unless you’re 90 years old, the harm to your body is virtually nothing.

Now I show you one that is 30 feet off the ground. What’s the impact on your body going to be? The impact on your body is going to be much more than 10 times the impact of jumping off of that three-foot platform.

How do I know that? Because of the jumping off the three-foot platform was almost zero, and almost-zero times 10 is still very close to zero.

If I tell you right now to double the dose of any medication you’re currently taking, the effect on your body will not be twice that of the current dose, it will be more than twice the effect.

If I give you seven bottles of wine, the harm from drinking all the bottles at one sitting is going to be much, much more than seven times the harm from drinking a bottle a night for a week.

The Distribution Matters More Than The Amount

The distribution of the stressor—whether you take it all at once or spread out—matters more than the average amount of stress you take.

Why? Because anything that absorbs something—physical energy to a teacup, traffic to a highway, water to a sponge—has a physical limit to what it can absorb. We call this saturation.

As you reach the saturation point, you’ve absorbed so much of something that it changes the underlying characteristics of the system. Cars have to accelerate more slowly. Water molecules in a sponge are blocked from entering by other water molecules. The sheer amount of what you’ve already absorbed changes the “rules of the game.”

Put another way, as you saturate a system, it’s not just your costs that go up, but the marginal cost is also increasing.

Here’s an example of what I mean. Anyone who’s ever tried to scale up an advertising campaign within a certain specific ad audience will notice as the audience gets saturated with your ad, the returns get worse.

The marginal cost is going up because that audience is closer to absorbing all the advertising it will absorb. You’ve run out of enthusiastic ad viewers and are now paying more to show your ad to the less enthusiastic people within the same group.

Physicists and engineers have studied analogous phenomena for hundreds of years. The field of metallurgy observes something called the stress-strain curve to graph the impact of stressors on materials. Just as we’ve been discussing, a graph that starts out approximately linear (when the stress is not an issue) curves into a nonlinear acceleration of strain (when the stress is a big issue).

stress-strain curve
Stress-strain curve,

You’re Closer to “Uncle” Than You Think

This accelerating factor of harm is very, very important. There are several ramifications from it. Let me give you one right now. Take the following as a rule:

By the time you realize you were under the effect of a stressor, you are closer to the wheels coming off than you realize.

When you are just coming under the influence of a stressor, almost by definition you will not notice it. Its influence will be too little to be felt.

By the time the stressor exerts enough impact to be felt, you are somewhere on the part of the curve that’s really sloping downwards. This means it will take only a little incremental stress for the wheels to come off the wagon (we will refer to this breaking point as the “uncle” point, the point at which you say “uncle” because you’ve had enough).

Think about all the times when you’ve been exceptionally busy. It’s enough stress that you would notice it, but you don’t feel overwhelmed.  You might feel like your head is on a swivel, but you’re keeping it above water, and you feel somewhat exhilarated.

Ask yourself: From that point, how much incremental stress would it take for the wheels to hit your “uncle” point? The answer is, not very much more; you are already approaching the limit of the stress that you can absorb.

Fragile Propositions

All right. I told you all of that to tell you this:

There are certain propositions that you are going to be offered—certain trade-offs, deals, set-ups, situations, ways of doing business, etc.—you will confront, that we are going to call “fragile” propositions.

A fragile proposition looks like this:

Fragile proposition
Fragile proposition: limited upside, dramatic downside

You recognize our harm curve, right? It’s the same curve we’ve been discussing. Badness starts accelerating as you go further along the stress line.

The reason that we call it a fragile proposition is that the usual outcome is on the positive side (“upside”). Everything goes as it’s expected to go (i.e., no deviation from expectation), and you get to reap a modest reward on a frequent basis.

However, sometimes you can have an unusual outcome, one that shows up on the right side of the graph (“downside”). And because the harm is accelerating so fast, it doesn’t take a lot of additional stress to result in extreme downside.

When Optimization Is Less-Than-Optimal

Let me give you an example that most of us have experienced. In the last 15 or 20 years, corporate America has fetishized over-optimization. Optimization is highly fragilizing because it removes redundancies that may serve to cushion you.

Take supply chain as an example.

Let’s say you belong to a company that wants to optimize its supply chain down to a single vendor. It will likely be your cheapest vendor because you want to save costs.

You also probably forgo stocking up on any inventory, because this single vendor can supply you with just-in-time component delivery. Carrying inventory would be inefficient.

So, you’ve got everything working like clockwork, and day after day you cash in a limited but predictable upside. What is your upside? Cost reduction, and the profitability that comes with it.

But what happens when something goes wrong?

Late delivery? Consistent late deliveries? Maybe a systemic defect? Perhaps the vendor introduces a cyber event? Maybe it suffers reputation harm and goes out of business?

Under this configuration, with a single vendor and no inventory, your production stops.

And what do you do? What does every company do? They go through the most predictable of corporate response strategies: the fire drill.

When I present this material live and I ask, at this point, how many people have been part of a corporate fire drill, every hand goes up. When I follow up by asking whether those fire drills were probably unnecessary, virtually every hand stays raised.

Fragile propositions rely on everything going according to plan all the time.

Being Fragile Means Relying on Predictions

People generally assume that all outcomes will stay within the limited range of what they’ve been seeing up to this point. They don’t have enough imagination to ask themselves what they would do if, one day, something goes not quite right. Instead, they get caught flat-footed and make you come in Saturday and Sunday to try and fix it.

You can recognize if someone’s handing you a fragile proposition when you hear them say things like, “I know exactly what’s going to happen, and that bad thing will never happen.”

Or, conversely, “I know exactly what’s going to happen, and this is sure to happen!”

The 2008 Financial Crisis materialized from exactly this dynamic: “I know what’s going to happen. Home values have never significantly depreciated before. Mortgages have a predictable default rate that never goes above a certain threshold. Massive collapse can never happen.”

And then what did they do? They saturated the market with so many securitized mortgages that lending standards eroded just to keep pace. The sheer scale of the securitized debt, along with all the CDOs and derivative securities with which it was interconnected, collected millions of pebble-sized debt risks and fused them into an enormous boulder.

They changed the underlying assumptions of the system so that an “impossible” implosion could in fact happen.

That’s a fragile proposition.

So how do we now leverage this information to create something that is the opposite of fragile?