Data For Better Decisions. Nature or Nurture?
I’ve known that my neurodiversity package (severe dyslexia, mild high functioning autism, ADHD) informs how I interrupt the world as my “biological cortex” and gut-brain axis structures process sensory data and memory uniquely. I cannot modify my mind or brain’s basic structure any more than I could change my fingerprint, core DNA or the colour of my eyes; however, I can play with my microbiome. It’s an essential part of what makes me, me. My chemical and biological structures enable me to make sense of the world in my way. Communication (language, words, music, song, dance, art, sound, movement, gesture) enables us to share the sense we create from patterns and align with others who approximate the same (tribe). How we actually make sense (learn) is intensely debated, with one camp believing that language is our sense maker, assuming that we might observe patterns but cannot understand it without language? Other than that, we make sense and then create a language to communicate the insight we have gained. Irrespective, language allows us to both structure and navigate our space and share the journey.
Is how we structure or frame something nature or nurture?
Why does this question matter? We all read, speak and write differently, we all understand differently, but we use questions to clarify understanding, check meaning and create common interruption. How we individually structure meaning is determined from the perspective we have been given (nature), from what we have been taught (nurture) and what we align to (bias). Our structure is an ontology*. Imagine putting one person from each of our worlds religions or faith groups into a room, but assume no-one can speak the same language. How and what would they agree or disagree about as there is no common structure (ontology)
* An ontology is “the set of things whose existence is acknowledged by a particular theory or system of thought.” (The Oxford Companion to Philosophy)
By way of example, the word “Evil” creates meaning for you as soon as you read it. Without a doubt, the nature of evil is a complex and nuanced area too often glossed over in a rush to present or evaluate the defences and theodicies. Let’s unpack the word using the super book “Making Evil” by Dr Julia Shaw. Evil is an unavoidable part of life that we all encounter as we all suffer in one sense or another, but what makes something evil is a matter of framing/ structure/ ontology. “Natural evil” is the pain and suffering that arises from the natural world’s functioning or malfunctioning. “Moral evil” is the pain and suffering that results from conscious human action or inaction. It is evil where a person or people are to blame for the suffering that occurs; a crucial point is the blameworthiness of the person at fault. Moral evil, at its heart, results from the free choice of a moral agent. If we just look at the consequences, it is not always possible to tell whether moral evil has taken place or not; we have many mitigations. Therefore a level of moral evil can be found in the degree of intention and consequence. However, if we compare death rates for natural evil (suffering) and moral evil at an extreme people killing people, the latter is a rounding error in the form of suffering in the world. The point is that by framing something, I can create a structure for understanding. Critically our structures frame our understanding.
Critically our structures frame our understanding.
Structures are ontologies which are philosphies.
To explore that our structures frame our understanding, what ontology makes us human? When we look at the different view below, we can view humans in many different ways. Note: I have deliberately ignored the classical all living things ontology structure (insects, birds, fish, mammal, reptiles, plant). The point is that your framing or how you structure something at the start leads to a guided conclusion.
Our framing or how we structure something at the start leads to a guided conclusion.
Pick a different structure, and you get a different answer; the ontology creates a natural conclusion. It is likely that if you pick a philosophy/ ontology/ structure, you can only get what that framing will shine a light on or enable.
It is likely that if you pick a philosphies/ ontology/ structure, you can only get what that structure will shine a light on or enable.
This matters because all data has structure!
I explore continually the future of the digital business, which are underpinned by data, privacy, consent and identity. Data is Data (it is not oil or sunshine). What is The Purpose of your Data? Quantum (Data) Risk. Does data create a choice? Data and KPI’s. Wisdom is just more data. Data can create what does not exist. Data is not Memory.
I am asking these questions of directors, boards, senior leadership teams and data/ data managers. Directors are accountable in law for ensuring no discrimination and health and safety, but how can we know what we know if we don’t know the structure or framing of the data that gave us the result. If we assume - that is a risk.
Do you have a data philosophy, and what is it?
What is the structure of your data by silo? Is there a single top-level ontology?
Do you know the structure/ ontologies of data for your ecosystem?
What is the attestation and rights of the data in our data lake? How do we check if we are using data for a different purpose than intended?
How would you detect the consequences in your decision making by the aggregation of data with different ontologies?
The Directors are accountable in law for discrimination, health and safety, and decision making (S.172 companies act), but how can we know what we know if we don’t know or understand the structure/ ontology and its limits. We can now longer assume, as it is a known risk.
For most, this is already too much detail and in the weeds! If you want to go deeper, this is a fantastic paper. A survey of Top-Level Ontologies to inform the ontological choices for a Foundation Data Model
Summary
We want to use data to make better decisions and create new value. However, we need to recognise that our data has a structure (ontology). Our data’s very structure (designed or otherwise) creates bias, prevents certain outcomes from being created, and creates others. The reality is that our structures (ontologies) have already committed to the success or failure of your data strategy and business model.
The reality is that our structures (ontologies) have already committed to the success or failure of your data strategy and business model.
As a leader, have you asked what is the structure (ontology) of our data? Has your team informed you about the limitations of your data structure/ ontology on decision making? The CDO should be tasked with providing a map, matrix or translation table showing data sets linkage to ontologies and the implications. As we now depend on ecosystem data, do you know the ontologies of others in your ecosystem and how that affects your decision making capability? Gaps in data sharing ontologies affect decisions and create Quantum Risk. What assumptions do we make about data without knowing is essential for investment risk, as we are using public ESG data to make capital allocation decisions without knowing where the data came from, what ontology the data has, if the right analysis tools have been used.
----
Implication 1. Management and Leadership
The figure-of-8 diagram below shows two interconnected loops. The connection is the mindset of the leader. Outstanding leadership with an open mindset can choose which loop is best at this time. Poor leadership will stick to the lower closed mindset loop. The lower loop never starts with a different way of asking questions or solving problems. Those in this self-confirming loop stick to the same biases, same decisions and same paradigms. This creates the ideas of one culture and a fixed culture. We have our way of doing it. The approach is consistency; the methods are highly efficient and based on the $1bn profit last year, we know it works, and we should continue to do the same. The reward mechanism, KPI and balanced scorecards are structured to keep the same highly efficient and effective thinking. Is assumes that yesterday, today and tomorrow will create the same outcomes if we do it the same. There is nothing wrong with this, and during times of stability, many have made vast fortunes with this approach.
Great leaders follow this loop when it is right but can also swop to the upper loop. Such leaders sense a change. Such a “paradigm shift”, a concept identified by the American physicist and philosopher Thomas Kuhn, “is a fundamental change in the basic concepts and experimental practices of a scientific discipline”. This shift means there is a new structure to understand (ontology). This paradigm shift has a new structure, which means that there is a need to determine the new culture to create value with a new structure. Together a team will form an approach. At this point, the team will question the shift and the assumptions that have led to change, setting a new mindset for the new order.
Critically - understanding structure and ontology is crucial, and it is why I believe Data Philosophy, Data Ontology and better decisions based on data are current board issues. Still, they require new skills, are highly detailed, and often require a mind shift.
Understanding structure and ontology is crucial for a data-driven digital board.
Implication 2. AI and Automation
The Data Paradox. How are you supposed to know how to create questions about something that you did not know we had to ask a question of?
Every child reads a book differently. A child learns to use questions to check and refine understanding. Every software engineer reads code differently. A software engineer is forced to check their understanding of the code and function by asking questions and by being asked questions. Whilst every AI will create sense from the data differently (ontology and code), right now, an AI cannot check its understanding of data by asking questions! Who could/would/ should the AI ask the clarification question of, and how do we check the person who answered is without bias? (Note I am not speaking about AQA).
Sherlock Holmes in The Great Game says, “people do not like telling you things; they love to contradict you. Therefore if you want smart answers, do not ask a question. Instead, give a wrong answer or ask a question in such a way that it already contains the wrong information. It is highly likely that people will correct you”. Do you do this to your data, or can you do this to your AI?
Today (Feb 2021), we cannot write an “algorithm” that detects if AI is going to create harm (evil). Partly because we cannot agree on “harm”, we cannot determine the unintended consequences, and we cannot bound harm for a person vs society.
There is a drive towards automation for efficiency based on the analysis of data. As a Director, are you capable of asking the right questions to determine bias and prejudice created in the automated processes, the data structures, different ontologies, data attestation or bias in the processes? Given Directors are accountable and responsible - indeed, this is a skill all board needs. Where is the audit and quota for these skills, can you prove it is available to the board?