Beware of Storytelling Abuse!
Decision Errors - Fact-Checking vs Story-Checking
When confronted with an event, our natural tendency is to create a story (create meaning) that explains the phenomenon ie to connect the facts with causal links.
Identifying facts correlating with a phenomenon and establishing a causal link is the nature of living things. Even single-celled organisms have this ability.
Gulls produce vibrations that mimic the sound of raindrops, attracting worms to the surface - a behaviour known as "worm-charming".
Fact 1 = raindrops hit the surface of the water
Fact 2 = worms
Fact 1 => Fact 2
Seagull imitates (Fact 1) by tapping the surface of the water with its webbed feet to attract the worms (Fact 2) = Intelligence.
It is therefore biologically automatic for human beings to systematically create stories (causal links between facts).
But, of course, correlation is not necessarily causation, and this heuristic can lead us astray when making decisions.
The figures speak for themselves. With just 5 verified facts, you can have up to 120 possible stories. So imagine a phenomenon made up of 10, 50 ... 100 facts (or parameters), we arrive at numbers of stories to be verified that are (humanly) impossible.
Example with 10 facts => 10! = 3,628,800 possible stories! The problem for truth-seekers in the human and social sciences is not so much fact-checking as finding the real story.
Difference between fact-checking and story-checking
One of the top salespeople complains to his manager that he has lost a customer because the direct competitor has radically lowered his price and the customer has signed with the competitor.
This sales rep also informs him that 2 sales reps have just resigned to join this competitor because the latter is performing better thanks to its new pricing policy.
Here are the facts:
- the sales rep has lost a customer
- 2 sales reps have resigned and joined the competitorthe competitor has a new pricing policy.
The sales rep's story is that the loss of the customer and the sales reps is due to the direct competitor's pricing policy.
Fact-checking: the competitor has indeed lowered its price and the HR manager confirms that 2 salespeople have indeed resigned. The manager calls another of his very good salespeople, who also confirms that the direct competitor is aggressive.
The story "seems" to make sense, and the manager "believes" in it.
In reality, before launching into a price war, the manager should do some story-checking as well as fact-checking!
Was the sales rep's loss of the customer really due to the price being too high? Or did the competitor offer something else?
Have the 2 salespeople left because of the competitor's new pricing policy, which makes it more efficient? Or is it that staff turnover in the sector is high? What's more, the company has also taken on 3 salespeople from competitors!
Correlation is not causation! Correlating facts can create an authentic but false story, a story told in good faith by the salespeople (they don't realise that they are interpreting what is happening to them in an illogical way)!
Except that in this example, the number of facts to check was only 3, i.e. 3! = 6 possible stories max. And the manager can get away with discovering the real story. But as soon as the number of (verified) facts increases, story-checking becomes impossible.
I've simplified the typology of stories (correlation/causality) in my initial illustration (showing only a series of linear causalities). Because you can make it even more complex (but I don't have the mathematical formula to predict the number of stories!).
Fact-checking means verifying the veracity of a fact, but also making sure that this fact correlates with a phenomenon of interest (the birth of a little panda in China has nothing to do with the loss of the customer by the sales rep in France). And AI will certainly be a great help if a large number of facts need to be checked. AI can also help with story-checking (where the real stakes are) and checking causal links.