Life's Probabilities
Why thinking about "how likely" something is can be misleading in our careers and our private life
What do casino bets, stock market investments, important life choices and strategic business decisions have in common?
Probabilities. Like almost any other area in life.
Our world is intrinsically probabilistic.
Nothing is certain. A lot can happen. Nothing has to.
The problem is, however, that we don’t understand the true nature of probabilities — and we are making ill-informed choices in both our private lives and in our careers.
Here’s why.
🎲 Dice, Stocks & Strategy
Consider the following three types of probability-related questions:
1️⃣ ONE
How likely do you think is it that…
… you will roll a four with a die?
… a coin will show heads?
… you will pull a Queen of Clubs from a deck of cards?
2️⃣ TWO
How likely do you think is it that…
… the US stock market will go up more than 5% this year?
… Bayern Munich will win the German Football Championship?
… a Democrat will win the next US presidential election?
3️⃣ THREE
How likely do you think it is that…
… you will make a nice career at this company?
… you will be happy when you move in with your partner?
… this marketing initiative will succeed?
Now, whenever we assess probabilities, we don’t just shout out a random number (well, some might). We usually do this based on some information.
However, the types of information available to us differ between these categories — so much so that the estimates we can come up with have almost nothing to do with one another.
🎯 Precision, History & Opinions
Let’s look at how we would create probability estimates in each category and we will quickly see the inherent flaws of talking about probability:
Re.: 1️⃣ ONE
You know that a (regular) die has six sides, a coin has two sides and a deck of cards usually has 52 cards in it.
This information reveals that the probability to:
throw a five,
flip heads, and
draw the Queen of Spades
is
1 in 6 (16.67%),
1 in 2 (50%), and
1 in 52 (1.92%),
respectively.
In other words, there is something about the nature of the world that tells you exactly how likely something is, regardless of what you think about it.
There exists precise objective probability outside of our heads, out there in the world.
And whether you know of this objective probability or not — it’s there, waiting for you to understand it.
For example, if you don’t know how many cards there are in a deck (because you’re not a big card player), the objective probability still is 1 out of 52.
In short, the information we use in categorie ONE to make a probability estimate about the future is:
1️⃣ Actual present-day real-world features
Re.: 2️⃣ TWO
There is nothing similarly objective we could say about the stock market, football championships or presidential elections.
There is, however, information about the actual history of stock movements, past championships and previous elections.
There exists a historic frequency we can use to infer the probability of something happening in the future.
If we know that:
in 58 out of 100 years the stock market gained more than 5% that year,
in 25 out of the last 40 years FC Bayern Munich has won the German Bundesliga, and
in 13 out of the last 23 US presidential elections a Democrat won,
we might say that the probability:
of the stock market going up more than 5% next year is 58% (58 out of 100),
of Bayern Munich winning the German championship is 62,5% (25 out of 40), and
of the next US election being won by a Democratic candidate is 56,5% (13 out of 23),
respectively.
There are historical realities we can use to make a probability estimate. Using objective historic frequencies to come up with how likely something is is a very common thing (startup success rates, car accidents, plane crashes, stock movements, sports betting, etc.).
If we now directly compare the types of information we use in categories ONE and TWO to make a probability estimate about the future, it looks like this:
1️⃣ Actual present-day real-world features vs.
2️⃣ Actual historic real-world happenings
Re.: 3️⃣ THREE
If you’ve ever been part of a strategic planning session, you’ve almost certainly been confronted with something like we need to quantify the likelihood of our success here or which measurable outcomes do we expect?
And even if you haven’t, you can easily imagine managers at your company talking like that, can’t you?
Likewise, though a lot more subtly, whenever we contemplate life decisions such as changing jobs or moving in with our partner, we “calculate” how likely it is that we will climb the career ladder, make more money etc. — or that we will be happy.
It’s intriguing to use probabilities and create a sense of plannability. A sense of certainty in an otherwise uncertain world.
But what kind of information is it that we use here to come up with our probability estimates?
Whether it’s an explicit number we attach to our projects at work or an implicit, unspoken estimate in our heads at home — we use neither unshakeable actual real-world features (like we do in ONE), nor actual historic frequencies of events (like we do in TWO).
What is it then?
Let’s stick to the strategic planning example:
When you discuss potential focus projects for the first half of 2021 in your Product, Engineering, Marketing, HR or R&D departments, you might be asked to say how certain you are that your idea will be a commercial success or that it will create some market value or revenue for the company in x years from now.
In other words, you should estimate the probability that your suggested initiative of, say, Building a People Connection Platform, will be a market success.
Now, there are two ways to do this (and a lot of variants in between):
Way A: you think about it…
… okay, 70%.
Way B: you analyse the shit out of it. You look at competitors and what they are doing, you calculate addressable market sizes, you sketch out potential scenarios and implications for sales cycles, you forecast societal trends and how they will change user needs, you even account for possible political changes…
… okay, 75%.
Regardless of the way you take, and regardless of the sophistication of your assumptions, you will not use objective probabilities and you cannot count on objective historic frequencies as your information.
You make observations which you then have to interpret, transfer and extrapolate. Based on this, you create beliefs and assumptions.
In short, you rely on subjective opinions.
If we now directly compare the types of information we use in categories ONE, TWO and THREE to make a probability estimate about the future, it looks like this:
1️⃣ Actual present-day real-world features vs.
2️⃣ Actual historic real-world happenings vs.
3️⃣ Hypothetical scenarios, assumptions & subjective opinions
🧮 Don’t Be A Fool
Now, there’s nothing wrong with making assumptions and forming educated opinions. Quite the opposite; it’s often needed and (when done well) better than just guessing blindly. Likewise, quantification of uncertainty can be very helpful.
That said, there’s a huge mistake we often make with category THREE types of probability estimates:
Assuming that we actually calculate probabilities.
We don’t.
Just because we call the result of our interconnected thoughts and seemingly sophisticated subjective opinions and assumptions a probability, doesn’t make it one.
I believe that we are too sloppy with the concept of probability in everyday language.
What does it mean to say that there is a 75% probability? That’s hard to say without differentiating between the categories sketched out above:
In category 1️⃣ ONE, this means that there exists an objective fact about the world that results in an actual 75% probability.
In category 2️⃣ TWO, this means that there exists an objective series of past events in the world that carries some (not all) information about what the actual probability is, and it happens to point to a 75% probability.
In category 3️⃣ THREE, this means that we have connected our perceptions of the state of the world with subjective forecasts and personal beliefs about implications and label our subjective thoughts a 75% probability.
Not accounting for these differences is a huge problem.
Probability labels can give a very false sense of having understood something about the real world when, in fact, we haven’t.
Good decision making depends on not fooling ourselves.
Stay well and merry Christmas 🎄,
Phil