Narrative Bias and Recipes for Disaster

How unreasonable expectations for estimation set your people up for failure

Consciously we know that everyone, every single human on this planet, suffers from cognitive biases. If we are interested in improving our selves as professionals and people, we try to be introspective and cultivate metacognition (thinking about thinking). But the fact is we may always fall prey to errors in logic, especially when we are the ones who construct those errors. This article is written for people who are stakeholders in any sort of project but the fact is I hope it can be useful for anyone, especially Product Owners or Scrum Masters to use when working with stake holders to help manage unreasonable expectations when they come. And they always come.

Imagine you are reading an interesting new recipe that you want to try. It’s formatted in the typical way and calls for ingredients like 1 hand full of chopped pecans, a cup of flour, a teaspoon of salt, etc. The serving size it produces is given as 4-6. You think this sounds great. Just under the recipe is the nutritional break out. 1 serving is 150 calories with 20 grams of carbs, 5 grams of fat, etc. Does this sound like something you would serve your family for dinner? It sounds great to me. But there is a logical fallacy at work here. The recipe has inputs (i.e. requirements for ingredients) that are less precise than the final outputs (the nutritional profile). Is this profile based on the 4 or 6 serving batch? How exactly is it established how many grams of fat are in a handful of pecans? Is it my handful or my wife’s that is used as the standard measure or someone else’s entirely? You can’t dismiss the rest of the ingredients either. How many grams of carbs are in a cup of flour? Grams is a measure by weight. A cup is a very imprecise measure by volume. Do you pack your cup of flour and scrape any extra off the top or just stick the cup into the bag and take what comes out? Is a slightly heaping cup passible in your kitchen? For a recipe that you got off the Internet all of this is probably fine. But would you stake your career on the number of grams of fat, or calories, or serving in the final product? How about someone else’s career or job? The fact is the nutritional breakdown is a complete lie and cannot be trusted in any way, not even as a general guide.

It is a fact of reality that if A + B = C, then C cannot be a number with more precision than either A or B. Given A = 2 and B = 4, then C cannot be a non-whole number such as 6.0004.  Put this way it seems very apparent and obvious to us. But humans are easily deceived by narrative. The narrative in the example above is the recipe. We are so used to seeing recipes or ingredients lists followed by nutritional profiles that we don’t think about the logic of what is going on.

Let’s look at the famous study by Tversky and Kahneman regarding Linda the feminist bank-teller.

Linda is 31 years old, single, outspoken, and very bright. She majored in philosophy. As a student, she was deeply concerned with issues of discrimination and social justice, and also participated in antinuclear demonstrations.

Please rank the following statements by their probability, using 1 for the most probable and 8 for the least probable:

A.      Linda is a teacher in elementary school.

B.      Linda works in a bookstore and takes Yoga classes.

C.      Linda is active in the feminist movement.

D.      Linda is a psychiatric social worker.

E.       Linda is a member of the League of Women Voters.

F.       Linda is a bank teller.

G.     Linda is an insurance salesperson.

H.      Linda is a bank teller and is active in the feminist movement.

A large portion of the people who take this test have a tendency to rank H as more probable than F. But this is impossible. The probability that A is true can never be less than the probability that both A and B are true. It is impossible for two things to be more probable than just one of those things by itself. Yet the narrative regarding Linda’s personality, which has nothing to do with the question of probability, fools many people into making an error that is obvious when spelled out mathematically.

And so it is with life. We know that the Cone of Uncertainty is real. That even with the best measurements that the impact of small errors in estimations compound exponentially over time. And yet stakeholders constantly expect their teams to make unreasonable estimations based on the current narrative regarding their project because it all seems so easy to them. They work at such a high level they don’t consider that it’s the details of implementation that are required before any sort of scope or time frame can be estimated. But what shocks me about this is that no one ever stops to consider the moral implications of this habit. As a stakeholder, you occupy a position of authority over the team. It doesn’t matter if you are a member of the executive management, a departmental manager, or a Product Owner the team respect you and want to please you. But in the proverbial breakfast meal that is the project you are just a chicken. All you add is some eggs. The team are the pigs. And it’s their bacon in the fryer if something goes wrong. The only conclusion that you should be able to come to is that it is morally irresponsible for you to expect any individual to provide you with an output that is more precise than the inputs regardless of how convincing the narrative that has been constructed happens to be. Because it is their reputation and career that is on the line, but for you their error would represent little more than a professional embarrassment.