Three Questions on Truth, Character, and Who You Decided to Believe

Bernard Williams spent his final years defending a distinction most of us have quietly abandoned — and the bill has come due.

She is sitting in her car, phone propped against the steering wheel, and she is crying — real tears, the kind that catch in the throat. She has just watched a measles outbreak sweep through her community, and she wants her four million followers to know the truth: vitamin A, not vaccination, saved her children. Her voice shakes. She is not performing. You can watch the video a hundred times and never catch her lying, because she isn’t lying — she believes every word.

Public health officials will spend the next six weeks trying to undo the damage. They will cite peer-reviewed studies, outbreak data, decades of epidemiology. They will be accurate. They will also lose, because accuracy was never the metric her audience was using to decide who to trust.

This is not a story about vaccines. It is a story about a distinction almost nobody makes anymore, and a British philosopher who spent the last years of his life trying to explain why losing it would eventually cost us everything.

Two Virtues Nobody Taught You to Tell Apart

Bernard Williams published Truth and Truthfulness: Accuracy and Sincerity in 2002, a year before he died, and it reads like a man racing to finish an argument he considered urgent. His claim was deceptively simple. There are two separate virtues bound up in what we call “telling the truth,” and we have collapsed them into one word so completely that we can no longer diagnose what’s failing when a society starts to rot from the inside.

The first virtue is Accuracy — the discipline of actually finding out what’s true before you speak. It requires effort: checking sources, tolerating uncertainty, revising your view when the evidence embarrasses you. It is unglamorous. It rarely trends.

The second virtue is Sincerity — the practice of telling others what you actually believe, rather than what flatters them, or you, or the algorithm. It is about honesty of transmission, not correctness of content.

Here is the trap Williams saw closing around us two decades early: a person can be perfectly sincere and catastrophically inaccurate at the same time. The mother crying in her car has sincerity in abundance. What she lacks is the far harder, far less rewarded virtue of accuracy — the willingness to have done the boring work of finding out before she felt entitled to an opinion. And increasingly, we have built a culture that rewards the first virtue and punishes the second.

First Question: Did They Do the Work?

The first question Williams’ framework forces onto any claim, any source, any confident voice in your feed is unglamorous but decisive: did this person actually do the work of finding out, or did they arrive at their certainty some other way?

Notice how rarely this question gets asked anymore, and how it’s been replaced by a substitute: does this person seem like they care? Caring is visible instantly — it’s in the cracked voice, the impassioned thread, the willingness to “say what everyone’s afraid to say.” Accuracy is invisible. It lives in footnotes, methodology sections, retracted claims quietly corrected, years of unglamorous labor that never trends because it doesn’t perform conviction — it just is correct, slowly, after the fact.

A financial commentator who spent eighteen months studying a company’s balance sheet before making a prediction will lose every engagement metric to a trader who screams “I’ve never been more sure of anything in my life” into a microphone that afternoon. The screaming is sincere. It is also frequently, provably wrong. And it wins anyway, because we stopped asking the first question and started only asking the second.

Second Question: Do They Believe Their Own Testimony?

The second question is subtler and cuts differently: when this person tells you something, are they telling you what they actually believe — or what it currently serves them to say?

This is where Williams gets uncomfortable for everyone, because sincerity’s opposite isn’t stupidity — it’s calculation. The pundit who privately doubts the position he’s paid to defend on air. The executive who knows the product has a flaw and says nothing in the earnings call. The politician who tests three different explanations in three different rooms before settling on the one that polls best. These people may be perfectly accurate — they may know exactly what’s true — and still fail the test of sincerity completely, because what they’re transmitting to you has been engineered, not believed.

This is the second betrayal, and it’s arguably worse than the first, because it’s harder to detect. The inaccurate-but-sincere person at least gives you the raw material of their real confusion, which you can examine. The accurate-but-insincere person gives you a performance calibrated to move you somewhere they’ve already decided you should go. You are not being informed. You are being managed.

Third Question: Why Did You Choose Them?

But the most uncomfortable question in this whole framework isn’t about the speaker at all. It’s about you.

Why did you decide to believe this particular person, on this particular subject, in the first place?

Sit with that honestly for a moment, before you answer. It was very likely not because you audited their methodology. It was because something about their character, as performed, read to you as trustworthy — their tone, their apparent bravery, their willingness to say the thing you already suspected, the sense that they were “one of us” rather than one of them. You outsourced the accuracy question to a character judgment, because character judgments are fast and accuracy audits are slow, and you are a finite creature with a finite number of hours and an infinite feed.

This is not a personal failing. It is the oldest cognitive shortcut humans have. Long before universities or fact-checkers existed, humans decided who to trust in a village by watching who kept their word, who showed up, who flinched when they lied. Character was always the proxy for truth, because in small groups, character and truthfulness were actually correlated. Williams understood this — his account of truthfulness is explicitly genealogical, tracing how these virtues evolved because cooperative life required them, not because some abstract law demanded it.

The problem is that we still run the ancient software — trust the person whose character reads as sincere — inside an environment engineered by strangers who have studied exactly how to perform that character at scale. The heuristic that kept your ancestors alive in a village of forty people is now being reverse-engineered by media operations who understand it better than you understand yourself.

Why the Algorithm Prefers the Liar Who Believes Himself

Here is the structural force that makes this pattern not just common but inevitable, and it has nothing to do with any particular platform, party, or personality.

Attention economies cannot price accuracy, because accuracy often looks boring, hedged, and slow — “the data is mixed,” “we don’t know yet,” “I was wrong last month.” Attention economies can price sincerity, because sincerity is legible instantly, in a face, a tremor, a raised voice. A system built to capture eyeballs will always select for people who seem certain over people who are correct, because certainty is a better performance and performance is the only currency the system can measure.

This is why the confidently wrong outperform the carefully right in almost every domain now — medicine, markets, geopolitics, parenting, nutrition. It’s not a conspiracy. It’s a selection pressure, as blind and structural as evolution itself. The environment doesn’t reward truth. It rewards the appearance of conviction, and conviction is cheap to fake and expensive to earn.

And once audiences learn — even unconsciously — that conviction is what gets rewarded, they start selecting for speakers who supply it, which trains the next generation of speakers to manufacture it, whether or not the underlying accuracy is there. The loop closes. Nobody designed it. Everybody is now trapped inside it.

What a Society Loses When It Can No Longer Tell the Difference

Williams did not write this book as an academic exercise. He wrote it because he believed something specific was at stake: the possibility of a shared world at all.

A society doesn’t need everyone to agree on values to survive. It needs a working consensus on method — some shared sense of how disputes about fact get resolved. Once that consensus collapses — once “who said it with more conviction” replaces “whose claim survived more scrutiny” as the tiebreaker — you don’t get a society with different opinions anymore. You get several societies occupying the same geography, each with its own trusted narrators, each immune to the others’ evidence, because evidence was never really the currency being exchanged. Character was. And character, unlike a fact, cannot be independently verified by a stranger on the other side of the argument — it can only be believed in, tribally, like a patron saint.

This is the condition Williams feared: not a world full of lies, but a world that has lost the conceptual tools to notice the difference between a lie and an honest mistake, between a liar and a true believer, between a source that did the work and a source that merely felt the feeling. Once that distinction is gone, every argument becomes a referendum on loyalty rather than a search for what actually happened. That is not a healthy disagreement. That is the early architecture of a civil war conducted entirely in the register of vibes.

The Discipline of the Unglamorous Citizen

There is no algorithm that will fix this for you, and no institution left standing that both sides trust enough to adjudicate it on their behalf. Which means the three questions Williams’ distinction hands you are not a policy proposal. They are a private discipline, to be practiced alone, before you share, before you nod, before you decide someone has earned your trust.

Did they do the work? Do they believe what they’re telling me, or are they managing me? And the hardest one — why did I choose to believe them, and would that reason survive being said out loud?

None of this makes you popular at dinner. It will not trend. It offers no tribe, no dopamine hit of righteous certainty, no crying-in-the-car catharsis. It only offers you something quieter and, in the end, rarer: the chance to be someone whose beliefs were actually earned, in a century that has made earning them optional.