Three Urns
An Introduction to Uncertainty: Teaching Note
Overview
Effectuation has been described as a series of heuristics for making decisions under uncertainty. Prior to presenting effectuation, therefore, it can be useful to engage with participants on the definition and examples of uncertainty. There are several approaches we have found effective in the classroom.
This approach uses the evocative metaphor of the three urns from Nobel prize-winning economist Frank Knight, who in 1921 described three different types of uncertainty: (1) known distribution, unknown draw, (2) unknown distribution, unknown draw, and (3) unknowable distribution.
The first urn is transparent (we can call this the predictive urn or the future under perfect or near-perfect information—or any other term tailored to the audience and its prior training of course). Since the contents are visible, even if the draw is random, people can make a good guess at the odds of getting what they want from the urn.
The second urn is opaque (call this the risky urn or the future under imperfect information). But if allowed to draw from it multiple times, people can see regular patterns and over time can formulate a mental model of what the opaque urn contains. So they get better and better at calculating the odds for what they want from it.
The third urn is also opaque but has no pattern whatsoever to it—so even after several draws, people cannot predict what will come next. In fact, they cannot even begin to calculate the odds. All the techniques that we teach them in standard decision making and risk management classes—techniques such as formal analysis and net present value calculations (place a bet), statistical or probabilistic analyses and portfolio diversification (place many bets), or experimental design and [real] options analyses (place staged bets)—all of these become relatively useless. This third urn is designed to capture Knight’s concept of “true” uncertainty—what economists today call “Knightian uncertainty,”2 where the future is not only unknown but also unknowable, where there is effectively no history or prior knowledge to help guide a prediction of what might happen next.
We have had good luck (and good fun) with using three physical “urns” in the classroom. The first clear, the second and third opaque—and the first and second containing objects of a single category—chocolates in different colored wrappings, for example, and the third urn containing a variety of absolutely random stuff—anything and everything that could fit into the container.
Materials
To provide a specific example (and one which will let you build on and substitute items based on your own means and interests), assemble the following into a shopping bag prior to the class:
A clear glass jar (ideally with a cover), filled with an even number of black (darkest chocolate) Lindor wrapped truffles, and whiteLindor wrapped truffles.
An opaque jar (ideally with a cover), filled with an even number of red (milk chocolate) Lindor wrapped truffles, and blue (dark chocolate)Lindor wrapped truffles.
A second opaque jar (ideally with a cover), filled with a collection of really unusual – uncertain – items.
Three pre-printed sheets. A white one with the word “Prediction”. A light blue one with the word “Risk”.A light red one with the word “Uncertainty”. Then when you do your presentation of effectuation, anything causal goes into blue text (as its connected with Risk), and anything effectual goes into red text (as its associated with uncertainty).
Process
As the class if they want to play a game. If they are uncertain, let them know that the alternative is a 60 page reading. And that the game involves Swiss chocolate. Once you have some buy-in to the game, bring out the clear jar, and tell them that the rules are simple:
They have to make a prediction about what they will draw from the urn.
They make the draw with their eyes closed.
If their prediction is correct, they can keep the chocolate they draw plus another.
If their prediction is incorrect, they must return the chocolate.
Select a participant, saying “Black chocolate or White chocolate?” let them make a prediction and take a draw, and repeat until several participants have successfully won two chocolates. Then tell them that this game appears a little below their abilities, but that you have another.
Produce the second (Risk urn – pre-loaded with the red and blue chocolates).Open the cover and select a participant, saying “Black chocolate or White chocolate?” let them make a prediction and take a draw, and repeat until several participants have successfully won two chocolates.Then tell them that this game appears a little below their abilities, but that you have another.
Produce the third (Uncertainty urn – pre-loaded with the all sorts of uncertain materials).Open the cover and select a participant, saying “Black chocolate or White chocolate?” let them make a prediction and take a draw. You’ll get a good laugh when they pull out a yo-yo, a stuffed animal, or whatever you put in there. After three or so draws, people get it, and its time to put down the third urn, and start asking questions.
Questions
What is the difference between these three games?
The point of this question is to extract the salient differences between prediction (where the distribution is known and obvious even before you start playing), risk (where the distribution can be discovered with enough draws and enough attention paid to the information generated by those draws), and uncertainty (where no number of draws will ever provide any information about the next one). As participants figure these things out, respond by taping the label to the respective urn. You’re done with this question when you have all three urns, side by side on a table in front of the room, labelled with Prediction, Risk and Uncertainty.
Advanced: What does the color of the chocolates signify?
Prediction is a black and white game where answers are clear. Risk is a shades game, where answers can be closer or further from right based on your calibration. And of course Uncertainty is any of the colors, in any order and in any shape or size.
Which game did you most enjoy playing?
The answers to this question will surprise you. If you ask people to sequentially raise their hands based on whether they liked Prediction, Risk and Uncertainty best, you will find that there are at least some people in the room for each different game.
Which game do you play in your job?
This one is where the exercise should connect. Most people will say Risk. But then follow up with examples. As you discuss the examples, what participants begin to realize is that most of the things that they face every day (innovation, competition, macro-economy, customers, regulations, hiring, firing, partners, …) are uncertain. And one very common reason is that at the core of many of these elements is human decision-making. Which is not that predictable – or even modellable. Maybe in the macro (The Future That’s Already Happened – Drucker’s accurate predictions about demographics), but when it comes down to one individual, or even an individual firm, outcomes typically fall into the third urn.
Which game do we generally teach you to play in business school?
You may not want to go here, but once you’ve gotten people clear on a) the distinction between Prediction, Risk and Uncertainty, and gotten people warmed up to the fact that what they do most every day is play the Uncertainty game, ask them what we have taught them for playing the Uncertainty game. The answer is not much. But that gives you an opening to let them know that that’s going to change today
Which game does the entrepreneur play?
This is an easy one after the previous questions, and perhaps unnecessary, except that it allows you to talk about why Knight explained Prediction, Risk and Uncertainty in the first place – which is to justify the premium received by the entrepreneur after land, labor and capital had all been covered.
Outcome
A room full of people eating chocolate, who just learned a theory from a Nobel Laureate, and should be ready to take on effectuation.