What is the world’s largest desert?
This question has two answers, broadly speaking, the wrong answer – and many other incorrect answers – the Sahara, the Mojave, Judah (I know no one thought of Judah), and any other number of answers that we might think are right, that we’ve heard their name (for which we may have availability bias), and then, of course, the right answer.
For our purposes, the correct answer isn’t relevant at all (and to be only semi-annoying, we’ll put it at the end), so if you know or have it, cheer – we deserve to have that warm and pleasant feeling of knowing something that others don’t (like the feeling of peeing on yourself, where you enjoy the heat, until you find out you’re alone, you’re cold and the smell isn’t exactly rosy in essence) for another moment, and that’s cool.
Why bother with this question? What’s the point?
My post won’t be entirely about biases and logical fallacies (although I do love me some biases), but about problems that are well-defined.
What is this well-defined yout are talking about?
Let’s start with the concept of ‘well defined’ (without, well, defining it)
The term ‘well-defined’ comes from the magical world of mathematics (as it does from the term ‘Injective function’, which I am known to use) and it describes the way in which the mathematical definition of an expression is constructed – an expression will be well-defined if it has a unique definition or a unique value.
This unique value or definition enables us to clearly and precisely identify the expression and its properties, allowing us to better understand and build upon it.
I agree with the statement that when a term moves between two similar domains of meaning, its meaning is not always an injective function (Oh my! What is it? Was it the gun from the first act? ) – that is, there are nuanced changes that we make to the phrase in order to better fit what we are trying to convey.
Having said that, what exactly is a well-defined process?
I consider a well-defined process to be one in which all the steps (in the planning phase) are clear, so there is a clear understanding of what the goal, the measurement method and the expectations are set out to be from the start.
There is a tendency for us to want to define things clearly – anyone who has ever designed an AB test knows exactly what that means – such a test clearly requires to be well defined – what is the confidence interval, what is the defined change, how long will we run it – all of these actions are necessary to reach a state of a well-defined definition – without all these actions, all the test results do not really change, as they can be interpreted in a number of different ways.
Where are we not well-defining aspects?
In the transition between the demands of those who hold the steaks (you know, stakeholders) and the technical aspect – that’s where we can improve, and to expand a bit on this – the real problem here isn’t parked with the analyst or at least it’s not parked with the analyst in its entirety.
For instance, let’s say that the marketing manager wants to know which campaign performed the best, or if the product manager wants to know if there are more mobile users or more computer users using our app – both of these questions are legitimate and important, but they are not clearly defined.
Let’s take this apart and reassemble
In the case of a marketing manager, asking ‘What is the form of good’ is a very difficult question; thousands of years of philosophy have not been able to come to an injective function for it (oh, wow, the gun just fires and fires and so on).
However, it is also true that there are certain indicators that can be agreed upon by a manager with the marketing department and the analyst for the question “What is the best” – but as soon as it remains at the implicit level – it creates a space for mistakes.
Where can this gap be seen very clearly?
Well, let’s say that the product manager will have to try to figure out why the best campaign (let’s say the one that got the most candidates for the least money) attracted not so good customers (only using the free version).
When we have not defined time constants – campaigns have a tendency to be very general sometimes, which can pose a bit of a problem, since the need to normalize the data to determine ‘days since it went live’ is often not taken into consideration when defining time constants.
There are a few geographical constants that we need to define – Israel is rarely, and contrary to what we Israelis might like to believe, comparable to the United States on a geographical level. Even inside the United States, there is no doubt that comparing California with Texas isn’t the most obvious thing in the world without a preliminary analysis taking place first.
Let’s continue with the product manager – who wants to know the platform of use – the naive approach is to simply create a table, a bar graph or just say that X percent of users are using mobile devices and one minus X percent are using computers – but if the need is not well defined (to determine whether development resources should be invested in the process of onboarding new customers on mobile devices) – it is not only likely that we will not answer the right question, but we may cause harm to the company as a result (I am probably not the only one who has said this – you don’t need to know the answers but you do need to know what to ask)
What could be even worse than this situation?
In terms of questions that get emotionally charged, or in questions that have a political nature, as in global warming, evolution, vaccines or anything else that gets a political touch and starts to leave the field of science in order to gain more votes at the polls, and even hard scientific facts turn into pawns in a poorly managed debate.
Our role as analysts is to investigate, isolate variables, and clean up the noise, i.e. to identify what is real and what is not real. Therefore, the arguments between analysts can be highly emotional at times, because we are not good enough at it.
So, what is the world’s largest desert?
The answer is Antarctica (mmmm, that warm feeling), and the answer can be derived from the very definition of a desert, which is defined as an area receiving less than 200 ml of precipitation per year.
It is clear that a dissonance exists regarding deserts as a hot, sandy place filled with camels and snakes, and as less a place where it does not rain.
to conclude , deserts are not only hot and sandy, but their climate can vary from cool to hot, from dry to wet. Therefore, deserts can be seen as a diverse and unique environment.
Just kidding – the concept of a well-defined process is essential in many fields, including mathematics and data analysis. It enables us to clearly and precisely identify the expression and its properties, allowing us to better understand and build upon it. However, when it comes to the demands of stakeholders and the technical aspects, there is often a lack of well-defined questions, leading to misunderstandings and mistakes. As analysts, it is our responsibility to investigate, isolate variables, and clean up the noise to identify what is real and what is not. By improving our ability to define questions clearly, we can ensure that our analysis is accurate and valuable. And as for the answer to the question, the world’s biggest desert is actually Antarctica – but as the author suggests, that’s not really the point.