A Slayer of Bullshit
If you have ever picked up Vaclav Smil’s 2021 book, Numbers Don’t Lie, you probably did it for the same reason most of us reach for a data-driven book right now: you are tired of the noise.
Not the ordinary kind of noise, like a busy inbox or a loud street. I mean narrative noise. The kind that convinces you the world is collapsing, or magically improving, depending on who is speaking.
The kind that turns every complex issue into one with a simple villain and a simple solution. The kind that makes people certain what the problem is, angry, and ready to share a post before they have even asked, “Certain about what, exactly?”
Smil’s central promise is refreshingly unromantic: numbers, handled honestly, can pull you back to reality. Not because numbers are holy, but because they are stubborn. They refuse to care about your politics, your personal brand, or your favourite theory.
In his book, when Smil moves from population to food to energy to machines, he is not trying to win arguments. Instead, he is trying to restore proportion. He is reminding you that the world is physical and constrained, that trade-offs are real, and that slogans do not power grids or feed cities.
In a world flooded with false narratives, your competitive advantage is not “having an opinion.” It is being able to filter out the signal from the noise.
Let’s take a closer look.
Why False Narratives Spread
Disinformation does not win because people are stupid. It wins because people are busy.
Every day, we are flooded with claims, charts, clips, and outrage. Our brains respond to these by using shortcuts.
To avoid having to deal with complexity, we reach for simple stories: heroes and villains, single causes, dramatic turning points.
Social media algorithms reward those stories with attention, and attention is the fuel of modern persuasion.
So misinformation often arrives wearing a costume of certainty:
A single number without context (“crime up 200%”).
A dramatic graph with missing axes and cherry-picked timeframes.
A comparison that feels intuitive but is mathematically meaningless (“this is bigger than that,” while quietly switching denominators).
A story so emotionally satisfying that you want it to be true.
Smil’s work pushes against this by doing something that feels almost old-fashioned: he slows down and asks, “Compared to what?” “Over what timeframe?” “Measured how?” “At what scale?”
Those questions sound small. They are not small at all. They are the filter.
The Signal-vs-Noise Toolkit
Smil does not preach “trust numbers” in the naive sense. His deeper lesson is: numbers only tell the truth when you force them to answer the right question.
Here are the practical filters that repeatedly matter:
1. The Denominator Is Where The Story Hides
A classic narrative trick is to talk about the numerator and ignore the denominator.
“Our city had 30% more incidents this month.”Okay. More incidents per what? Per population? Per number of calls? Per patrol hours? If the population grew, or reporting changed, or a new category was added, the “increase” can be a paperwork phenomenon rather than a reality shift.
“This hospital has twice the mortality.”Twice compared to what case-mix? Are they taking the most complex cases? Are they a trauma centre? Without adjusting for context, the number tells a moral story that may be false.
Smil’s broader point is that scale and context are everything. A big number can be scary until you realize it is spread across a big system. A small number can be terrifying if it concentrates risk.
Quick Habit: whenever you see a percentage, ask: percentage of what? Whenever you see a count, ask: count per what?
2. “Average” Is Often a Comforting Lie
False narratives love averages because averages feel authoritative. But averages are the perfect hiding place for inequality, skew, and weird distributions.
“Average salary is RX.” If the distribution is skewed, the average can rise while most people feel worse off. Median tells a different story. So does the distribution itself.
“Average temperature increased.” That may be true, but the lived reality is often about extremes: heatwaves, drought cycles, flood frequency. A single average can hide volatility.
One number is rarely the point. The pattern is.
Quick Habit: ask whether the mean, median, or range is the signal.
3. Trendlines Seduce Us, But Timeframes Can Be Rigged
A narrative can be “proven” with a carefully selected start date.
If you start your chart at a peak, everything looks like a decline.
If you start with a crash, everything looks like progress.
If you end at an outlier month, you can manufacture urgency.
Smil’s subject matter often spans decades because the underlying systems move slowly: energy infrastructure, agriculture, demographics, and industrial capacity.
Short windows can be emotionally compelling and analytically dishonest.
Quick Habit: When someone shows you a trend, mentally pull the camera back. What does it look like over 5 years? 10? 30?
4. Correlation Is Not Causation, And Stories Love “Therefore”
Disinformation loves to move from “these two things happened” to “this caused that” because causation feels like control.
“After policy X, problem Y rose. Therefore, X caused Y.” Maybe. Or maybe another variable moved: policing methods, demographic changes, reporting standards, economic cycles, even weather.
“People who do A are happier.” Great headline. But happier people might be more likely to do A in the first place. Or both could be caused by a third factor.
Smil’s approach is not to eliminate interpretation, but to anchor it in mechanisms.
How would causation actually work in the physical world? What constraints apply? What inputs must be present?
Quick Habit: replace “therefore” with “possibly, if…”
5. Compare “Like With Like”
A lot of public debate is built on lazy comparisons: countries, technologies, lifestyles, and industries. The numbers “prove” something, but the comparison is mismatched.
Smil is known for being relentlessly specific about what is being compared and what is being counted, especially on energy, technology, and environmental claims.
That is where so many modern narratives go wrong, because modern arguments are often about moral identity (“good” vs “bad”) rather than system performance.
Quick Habit: Ask whether the units, system boundaries, and context are the same.
Some Practical Examples Of How Easily We Get Misled
Example 1: The Viral Chart With The Missing Axis
If you ever see a graph that looks like a rocket ship, shared with a caption like “This proves everything.”
Then you may notice:
The y-axis starts at 95 instead of 0, exaggerating a small change.
The time window is conveniently short.
The source is unclear, or is a screenshot of a screenshot.
This is not a technical error. It is persuasion by distortion.
Signal Filter: Look for axes, timeframe, source, and whether the chart could be redrawn honestly with a zero baseline or a longer window.
Example 2: “It’s Doubling” Because The Base Rate Is Tiny
“Cases doubled!”
If cases went from 3 to 6, that is technically doubling, but it may not be practically meaningful. The narrative weapon is the percentage. The reality often lives in the absolute number and the rate per population.
Signal Filter: Convert percentages back into absolute counts and per-capita rates.
Example 3: The Energy Transition Story That Ignores Constraints
One of the most common modern narrative battles is about energy: what we should do, what we can do, what will happen quickly, what will happen “soon.”
Disinformation thrives here because most people do not intuitively grasp scale.
They hear “we will replace X with Y” without asking: replace how much, with what materials, built where, connected to which grids, supported by which storage, maintained by which supply chains?
Smil’s broader work, and the themes highlighted in Numbers Don’t Lie, repeatedly revisit this point: modern civilisation runs on vast physical systems. You cannot swap them out on vibes.
Signal Filter: When you hear “rapid transformation,” ask what infrastructure must exist first, what the timeline has been historically, and what the bottlenecks are.
Example 4: “This One Study Proves It” When Replication And Context Are Missing
A headline claims a miracle effect. Everyone shares it. Six months later, you never hear about it again.
The story was never “false” in a strict sense. It was premature, overgeneralized, and unreplicated. Disinformation often lives in that grey zone: not a lie, but a leap.
Signal Filter: Look for a body of evidence, not a single paper. Look for effect size, not just statistical significance. Look for real-world constraints.
Example 5: The Moralised Metric
Some metrics become identity badges. People use them to signal virtue or superiority.
Once a number becomes moralised, it becomes resistant to correction.
Smil’s style is helpful here because he refuses to let numbers be purely symbolic. Numbers are not costumes for ideology. They are tools for understanding the world we actually have.
Signal Filter: Notice when a metric is being used to shame or sanctify rather than to understand.
Becoming “Numerate” In Public
Developing a Smil-style temperament is useful. This is what it may look like:
Curious about reality, not addicted to outrage.
Sceptical of certainty, including your own.
Patient with complexity, because complexity is often the truth.
Respectful of constraints: energy, time, materials, geography, trade-offs.
Willing to say: “I do not know yet. Show me the baseline.”
This is what it means to filter the signal from the noise. And in an age where false narratives spread faster than corrections, this kind of discipline is moral clarity.
A Simple Everyday Checklist For Navigating Disinformation
When a claim hits your screen and tries to recruit your emotions, run it through this quick filter:
What is the claim, in plain language? If you cannot restate it without the rhetoric, it is probably rhetoric.
What is the unit and the denominator? Counts, rates, per-capita, per-hour, per-user, per-dollar.
What is the timeframe? Is it cherry-picked?
What is the comparison group? Are we comparing like with like?
What would change my mind? If the answer is “nothing,” you are in belief territory, not analysis territory.
Smil will not make you omniscient.
But he will make you harder to fool, because he repeatedly returns you to first principles: scale, context, constraints, and honest comparison.
Until next time, remember that this is how you stop being emotionally managed by the feed.
Dion Le Roux
References
Gates Notes. “Numbers Don’t Lie” (book notes and review context).
Goodreads. Numbers Don’t Lie (bibliographic listing and overview).
Penguin UK. Numbers Don’t Lie (publisher description).
Tunstall, S. L. “Review of Numbers Don’t Lie” (Numeracy / USF Digital Commons).
Smil, V. Official book page: Numbers Don’t Lie: 71 Stories to Help Us Understand the Modern World.