Deciding Under Uncertainty
7 Famous Decision Tools and Where They Come From

When you cannot know how things will turn out, a handful of tools get recommended again and again. Here are seven, traced to where they started and what their authors claimed. This page explains the ideas; it gives no money, betting or investment advice, and the examples are about ordinary choices.

1. “Decision theory says: always choose the option with the highest expected value”

Partly true The formal version is expected utility theory. The Stanford Encyclopedia of Philosophy sums it up: “Its basic slogan is: choose the act with the highest expected utility.” Two things are usually lost. First, utility is not money: “Utiles are typically not taken to be units of currency”, and “Bernoulli (1738) argued that money and other goods have diminishing marginal utility”. Second, the theory has well-known puzzles; “one problematic example is the St. Petersburg game, originally published by Bernoulli”. The entry treats it as “a normative theory — that is, a theory of how people should make decisions”, and whether you should always follow it is argued at length there. For choosing between two job offers, the useful part is the question it forces: how much does each outcome matter to you, and how likely is it?

2. “The Kelly criterion is an investing formula”

Partly true It is J. L. Kelly Jr.’s paper “A New Interpretation of Information Rate”, in the Bell System Technical Journal for July 1956, and it is about information. Kelly imagines a channel “used to transmit the results of a chance situation before those results become common knowledge”, and shows that “the maximum exponential rate of growth of the gambler’s capital is equal to the rate of transmission of information over the channel.” Its later use in betting and investing is another matter, and not one this page covers.

3. The pre-mortem

True as Gary Klein’s, described in Harvard Business Review in September 2007. In a premortem, “team members assume that the project they are planning has just failed”, and then explain why. Its purpose, in HBR’s summary, is “making it safe for dissenters who are knowledgeable about the undertaking and worried about its weaknesses to speak up”.

Unverified is the figure often attached to it, that imagining an event has already happened improves people’s ability to find reasons for it by 30%. The study usually cited is Mitchell, Russo and Pennington (1989), on “prospective hindsight”, which “involves generating an explanation for a future event as if it had already happened”. Its abstract does not report that figure. It says that “in the first experiment, temporal perspective showed little influence while outcome uncertainty strongly affected the nature of explanations for events”, and that “explanations for sure events tended to be longer”. A pre-mortem treats failure as certain, so that is the factor that matters here. Longer explanations are not the same as better ones, and we could not read the full paper, so we do not repeat the number.

4. One-way doors and two-way doors

True as Jeff Bezos’s, in his letter to Amazon shareholders published with the 2015 annual report: “Some decisions are consequential and irreversible or nearly irreversible – one-way doors – and these decisions must be made methodically, carefully, slowly”. “We can call these Type 1 decisions.” “But most decisions aren’t like that – they are changeable, reversible – they’re two-way doors.” Those, he wrote, “can and should be made quickly by high judgment individuals or small groups.” In a footnote on the opposite mistake, which he calls “less interesting”, he adds: “there is undoubtedly some survivorship bias. Any companies that habitually use the light-weight Type 2 decision-making process to make Type 1 decisions go extinct before they get large.”

The rule of thumb often quoted with it is from the next year’s letter, which points back (“I wrote about this in more detail in last year’s letter”) and adds that “most decisions should probably be made with somewhere around 70% of the information you wish you had. If you wait for 90%, in most cases, you’re probably being slow.” Note the hedges, and that he was writing about decisions inside a company.

5. Regret minimisation

True as Bezos’s own account of one decision, in an interview with the Academy of Achievement in San Antonio on 4 May 2001. Deciding whether to leave a job for what he calls “this crazy thing”, after being told the idea would suit “somebody who didn’t already have a good job”, he used what he called “a ‘regret minimization framework.’” “I wanted to project myself forward to age 80”. “I knew that if I failed I wouldn’t regret that, but I knew the one thing I might regret is not ever having tried.” It is one person’s account of one choice, though he went on to suggest it to others: “If you can project yourself out to age 80 and sort of think, ‘What will I think at that time?’ it gets you away from some of the daily pieces of confusion.”

6. The inside view

True as Daniel Kahneman and Dan Lovallo’s, in “Timid Choices and Bold Forecasts”, Management Science (January 1993). They wrote that “Decision makers have a strong tendency to consider problems as unique”, and that “overly optimistic forecasts result from the adoption of an inside view of the problem, which anchors predictions on plans and scenarios.” The paper pairs that with a second problem that gets less attention: “Overly cautious attitudes to risk result from a failure to appreciate the effects of statistical aggregation”. Bold forecasts and timid choices, from the same habit of treating each case alone. “Reference-class forecasting” is a later name for the method, and we did not check who coined it.

7. Superforecasting

True that the research behind it was a real tournament with published results. In Mellers and colleagues’ Psychological Science paper (2014), “five university-based research groups competed to recruit forecasters, elicit their predictions, and aggregate those predictions”. The authors’ group “found support for three psychological drivers of accuracy: training, teaming, and tracking”. The training “corrected cognitive biases, encouraged forecasters to use reference classes”, the same idea as in 6, and tracking “placed the highest performers (top 2% from Year 1) in elite teams that worked together.” A 2015 paper by Mellers and colleagues in Perspectives on Psychological Science describes this as “the winning strategy: culling off top performers each year and assigning them into elite teams of superforecasters”. We did not use the claims on the commercial company’s own website.

Our reading

Several of these tools came with hedges from the people who proposed them: a 70% rule full of “probably”, a paper about information rather than advice, a regret framework described as “a decision that I had to make for myself”. The versions that circulate are more certain. The authors’ own hedges are worth keeping.

Sources

• R. A. Briggs, “Normative Theories of Rational Choice: Expected Utility”, Stanford Encyclopedia of Philosophy.
• J. L. Kelly Jr., “A New Interpretation of Information Rate”, Bell System Technical Journal (July 1956).
• Gary Klein, “Performing a Project Premortem”, Harvard Business Review (September 2007), summary page.
• D. J. Mitchell, J. E. Russo and N. Pennington, “Back to the future: Temporal perspective in the explanation of events”, Journal of Behavioral Decision Making 2 (1989).
• Jeffrey P. Bezos, letters to Amazon.com shareholders with the 2015 and 2016 annual reports, SEC Exhibit 99.1.
• Academy of Achievement, interview with Jeffrey P. Bezos, San Antonio, 4 May 2001.
• D. Kahneman and D. Lovallo, “Timid Choices and Bold Forecasts”, Management Science 39 (1993).
• B. Mellers et al., “Psychological strategies for winning a geopolitical forecasting tournament”, Psychological Science 25 (2014); B. Mellers et al., “Identifying and Cultivating Superforecasters as a Method of Improving Probabilistic Predictions”, Perspectives on Psychological Science 10 (2015).

Checked September 2026.

Related: Mental models, checked · Numbers that mislead · Cognitive biases, checked