All Models Are Wrong, Some Are Useful
How to make decisions and thrive in an uncertain world
We live in a world that demands certainty.
We want clear answers, safe paths, and seamless solutions — but here’s the paradox: most of the tools we use to understand the world aren’t entirely true. They’re not accurate. They’re not perfect. And yet, they work.
“All models are wrong.”
An uncomfortable, almost provocative statement. It was said by British statistician George Box in 1976. But the full quote is even more interesting: “All models are wrong; some are useful.”
And that’s the key.
It’s not about having certainty in everything. It’s about usefulness.
We’re not chasing absolute truth — we’re chasing a practical truth, an applicable one.
George Box wasn’t only referring to statistics. He was talking about human thinking, about the mental models we use to build knowledge, make decisions, and navigate a reality too complex to fully comprehend.
Every day, we must decide and act. We can’t afford to stand still.
Let’s explore what models are and how to use them to navigate life without sinking…
What Is a Model?
A model is a simplified version of reality.
Think of an architectural model: a small structure made of cardboard, glass, and wood that represents a building that doesn’t yet exist. You can’t live in it, install elevators, or open its windows, but it lets you see the concept, understand the shape, and imagine the final outcome.
It’s useful, but it’s not the real thing.
The mistake comes when we believe a model is reality itself.
Mental models, scientific theories, economic forecasts, predictions… they’re all attempts to simplify the world. But there will always be things left out. And that’s okay. Because they weren’t designed to be perfect.
They were designed to help us think better — to be useful.
The Key Lies in the Degree of Error
George Box said: “The question is how wrong do they have to be to not be useful.” And here comes another powerful lesson: usefulness comes before perfection.
So, why are we obsessed with perfection?
Because we confuse perfection with usefulness. We think that if something isn’t 100% correct, then it must be completely useless. Nothing could be further from the truth.
As Isaac Asimov said in 1989:
The trouble is that people think that “right” and “wrong” are absolute; that everything that isn’t perfectly and completely right is totally and equally wrong.
, and the least wrong models are often the most useful.
Why Do We Need Models?
Because we have no other choice.
Our brains can’t process the full complexity of reality. We need shortcuts, patterns, simple explanations. A good model lets you do all this:
- Build knowledge that .
- Predict outcomes (even if you don’t always get them right).
- Decide .
Think of a weather forecast: it’s not perfect, but if there’s an 80% chance of rain, you’ll take an umbrella. And if it doesn’t rain, no big deal — the model is usually right, and it remains useful. That’s what models do: they don’t guarantee the future, but they give you context and help you prepare.
They give you an edge in life.
How to Decide in an Uncertain World
If no model is totally reliable… how do we make decisions in life?
Having only one model is like only having a hammer — . You need a toolbox that lets you use the right model for each situation. The more tools you have, the better you can think and decide.
Let’s take an example: you need more rest, your current pace is unsustainable.
- : You might see the problem as poor time distribution. You’re trying to do too much and ending up nowhere. This model suggests focusing on fewer things but making big progress. Less quantity, more quality.
- : Maybe the problem lies in your system. Perhaps you’re using outdated processes. This model encourages you to think about learning new tools, technologies, or skills to do the same things in half the time.
- : Chaos may be overwhelming you due to lack of flexibility. If your schedule is overloaded, with no room for the unexpected, one delay can ruin your whole day. This model urges you to create more flexible systems, with room for disorder.
Are these models perfect? No.
Do they work on their own? Sometimes.
Do they work better together? Absolutely.
That’s what a toolbox of mental models offers: options. Complementary strategies. Diverse perspectives. One tool might leave you stuck. A well-chosen set will get you out of trouble — again and again.
How to Choose Good Models
Not all models are created equal.
Some last for centuries. Others fall apart in just a few years. So, the question is: how do you know if a model is useful? Ask yourself these questions and you’ll find the answer:
- How long has it been in use? The models that have stood the test of time and still work are the most reliable. This is the . A great example is , which has been around since the 14th century.
- Does it apply in multiple contexts? The more versatile, the better. It won’t always be useful, but it will help in more situations. A good example is the . It doesn’t always apply, but it’s worth keeping in mind.
- What is its foundation? In math, logic, and other fields, there are eternal laws. If the model is built on these, it’s more likely to be valid and long-lasting. Great examples are and — concepts from mathematics and physics that we can apply in our daily lives.
A good model doesn’t need to be accurate all the time.
Just focus on using it when it’s useful, and you’ll successfully navigate life’s challenges.
✍️ Your turn: What mental models have helped you out of a tough spot? One that’s helped me the most is the . It completely changed my habits and perspective.
💭 Quote of the Day: “Ponder and deliberate before you make a move.”
— Sun Tzu, The Art of War
🌱 Here I plant ideas. In the newsletter, I make them grow.
Daily insights on self-development, writing, and psychology — straight to your inbox. If you liked this, you’ll love the newsletter. 🌿📩
👉 Join 15,000+ readers:
See you next time! 👋
References 📚
- Asimov, I. (1989). The Relativity of Wrong. Skeptical Inquirer.
- Box, G. E., Hunter, W. G., & Hunter, J. S. (1978). Statistics for experimenters. An introduction to design, data analysis, and model building.
