I first started thinking explicitly about uncertainty in a conversation with Josh Reich 15 years ago. He asked me why venture capital valuations seemed so haphazard. After some thought I ventured that there was something you just can’t know at the core of every startup. I made the analogy of someone auctioning an unopened box with unknown contents: how much would you bid for it? Josh interpreted this rather poor analogy as Knightian uncertainty: valuations are a prediction and uncertainty–the state of something being not just unknown but unknowable–makes prediction impossible.
How do you value an unopened box? Being problem solvers, we would attempt to find out its provenance, look at its size and markings, ask ourselves why someone boxed something up…it must have value if they did, etc. We would try to place bounds on its value. With size, for instance, we could ask ourselves what the most valuable thing that would fit into the box would be,1 and the most damaging.2 Even just being able to bound the values gives us a huge amount of information. Nothing in the real world is ever completely and entirely uncertain, and this is important when thinking about startups. Knowing the market, the customers, the suppliers, the customers, the regulation, etc. matters.
If the seller, however, knew what was in the box–and why wouldn’t they have looked?–any accepted bid would be an overbid. If you bid less than the worth of the box’s contents the seller would not sell. The seller might have even purposely changed the appearance of the box to make the value bounds misleading. If you have much less information than the seller you should not bid at all.
Most of our economic dealings take on this adversarial cast, even in non-zero-sum games, and so most of strategy is game theory. The correlation of risk and reward is a result of adversarial negotiation: if you wish to sell me risk I need to be compensated, and the more risk you are selling the more I need to be paid.
This is where the box analogy starts to fail: radically uncertain startups are not generally a contest against some adversary; this box has no seller trying to outfox you. Uncertainty about an opportunity means there will be few contenders for it. In startup markets where the opportunity is very large, startups often end up cooperating far more than competing. Starting a company is hard but the source of your difficulties will not usually be competitors, it will be Nature. (Not literally nature, this is a trope economists use to denote an impersonal force.) Nature may throw up obstacles to your success but these obstacles are not about you: Nature is not a strategic opponent. Since you can’t negotiate with Nature, risk and return are no longer correlated.
This may seem like a non sequitur since uncertainty makes risk meaningless, and in fact it is a non sequitur. It’s just a remarkably common non sequitur. We negotiate a higher reward for taking more risk, leading to the ubiquitous correlation. Since it is ubiquitous, when we see a potentially high reward we assume there must be high risk. If not, we ask ourselves “what’s the catch?” because when there is a strategic opponent there’s always a catch. There ain’t no such thing as a free lunch.
But the key to understanding uncertainty in a strategic setting is that much as we like to anthropomorphize her Nature does not set catches. You may only choose to take a high risk offered by Nature if there is a correspondingly high reward but the converse is not true: Nature does not demand you take a large risk to get a large reward. Nature does not care.
This affects how you make decisions. Imagine you have a set of possible decisions where the outcome of each decision is bounded but uncertain. If all of the outcomes have the same upper bound you would make the decision with the highest lower bound, the best worst-case, or maximin.3 This is the same decision game theory would lead you to in a game against strategic opponents. It is also perhaps the most common type of uncertainty we see. Imagine you are a doctor choosing a treatment for a patient. If the best case is always a return to health, you would choose the treatment with the fewest side-effects, absent other data on efficacy.
When deciding which company to start you may be able to roughly bound the possible outcomes while still being uncertain about the actual outcome. The possible startups, however, do not have identical, or even similar, upper bounds. But they do have the same lower bound (for the most part there is no less than zero in business.) This flips the rule on its head: if the lower bound is always the same you should make the decision that maximizes the upper bound. The maximax criterion. In practice this means you should always start the business that has the largest possible outcome. This is both obvious and against our deepest instincts.
Recognize that in an uncertain contest against Nature your instinct is wrong. Bigger potential rewards are not correlated with more risk. If you are pursuing a truly uncertain endeavor, like a startup, there is no way of knowing if the larger or smaller possible outcome is more likely to succeed, so the only rational course is to pursue the biggest possible outcome you can imagine.