The Gigantic Turnip: An attribution story

One of the common and confusing mistakes that people make is what is technically known as an "attribution error". It has been a source of harm and friction for centuries in families, organizations and societies. We'll start with a children's story and use it as a case study metaphor.

The Gigantic Turnip

A fable with origins in the Russian folklore, actively promoted in the kindergardens of communist countries of the Soviet Union and the Eastern Europe.

In short, a grandfather plants a turnip that grows extraordinarily large. So large that he cannot pull it out alone. He asks the grandmother for help, to no avail.

Successively their granddaughter is recruited to help, then the dog, then the cat. Still no success.

It is only after they get the help from the little mouse that they are able to finally pull it out.

It is a romantic story of extraordinary cooperation and the achievement of an extraordinary goal. The actors are able to overcome their condition and work together for the common good. It's not hard to see why the commies pushed it for kids education.

Attribution Theory

Psychology

Before being professionals we're humans so we have to start there. As humans, we may have a simplistic understanding of why are certain things happening instead of others. Attribution is one of the earliest cognitive actions we take to explain why something happened. We attribute the cause of an outcome to internal factors ( intentions, abilities, mood) or external factors ( situations, actions of others, environment etc.). Some people keep calibrating their attributions, improving their ability to recognize and integrate more factors. The growth usually happens through some level of failure, pain and suffering.

The vast majority of people make attribution errors. We are not able to identify the relevant factors, we pick the wrong ones or even reduce everything to a single cause and miss the others.

The subject is vast and the pit is comedy level deep. Here are a few tips to recognize and deal with psychological attribution errors.

Fundamental Attribution Error

We make it when we're unable to identify the principal factor. Either because we don't have the experience or we don't have all the data.

The grandpa might have been ill that day, yet we might assume he was all right and just generally weak.

Self-Serving Bias

We often attribute our successes to internal factors (skill, effort) and our failures to external factors (unfair circumstances, bad luck, adversity). We do it because we're motivated and conditioned to keep our high self-esteem and avoid guilt or shame.

The mouse might think he's much stronger just because his contribution has finally pulled the turnip out.

Actor-Observer Bias

Occurs when we attribute our outcome to external factors but we attribute others to internal factors.

The grandpa might be thinking that he's unable to pull out because the turnip is super gigantic and although he's sweating and giving everything he's got the girl and the cat seem to have way too much fun.

Computer Science - AI

In data science, attribution involves tracing the influence or contribution of data points, features, or sources to a model’s output or performance. This is particularly relevant in predictive modeling, analytics, and decision-making systems.

Conventional attribution models are driven by heuristic algorithms that aim to establish an 'optimal' mathematical representation of a situation and its complexity.

In the case of the Gigantic Turnip, a conventional model would distribute the successful outcome to every character in the story weight it by pulling strength - which may be approximated by the mass of each actor, from grandpa to the little mouse.

With Machine Learning and AI, constructing an accurate model requires repeated and diverse recordings and a wide set of factors for all outcomes. Inference accuracy has several thresholds and correlate with the records count and diversity.

We would need a lot of records on the family activity in turnip extraction: most if not all extractions, what size of turnips grandpa pulled out on its own or in various groupings. What grandma was able to pull out on her own or in groupings, and so forth down to the little mouse in case he had such inclinations. Only then we would be able to infer the chances of success and the resources required to pull out a turnip of a given size.

Operational Excelence

Achieving operational excellence with hybrid (human / AI) teams requires a sound application of data science and algorithmic thinking. There is extra high complexity in modeling client behavior, marketing signals, pricing and so on and it has to be coupled with a bias-free type of leadership.

Accurate attribution is key to predictability and decision-making success. On a human level, team alignment is usually affected by attribution errors.

Imagine the mouse claiming a new status in the household because without him the turnip would not have been pulled out. Team harmony and efficiency do rest on a fair outcome attribution.

A wise leader should recognize the potential of each team member and expect them to pull their own weight. He should also spot attribution errors in team members and work diligently to address them through education.

On the other hand, a smart decision-making system should factor in a wide variety of features, based on a model properly reinforced with success/failure attributions.

Such a system should be able to predict that let's say: the grandpa, the granddaughter and the dog should be able to pull out another large turnip, while the granny can keep making pies and while the cat keeps the mouse at bay.