October 31st 2025

Opinions Are a Last Resort

Having the wrong opinion.

The topic of the essay was “Can people ever really change?” Having aged every day of eleven years up to that point, I approached my response with vigor and focus. I was only a month into my Catholic education. It was sixth grade. Before that, my parents had sent me to public schools. Maybe because my new school was just so new to me, I didn't realize that to the question asked of me, there was in fact a correct answer.

I wrote about motivations, memories, personalities, values. I mustered every line of thought that I could from a little more than a decade of life experience, to write with nuance and care. It seemed like a big topic, and I wanted to give it its due.

The correct answer, it turned out, was this: people can always change, if they get divine help.

So, I was wrong. That's okay. Everyone is wrong once in a while. To my Catholic school, however, I was so wrong that day, the principal called home about me. My dad sat me down that night and explained to me what the principal had said to him. But he didn't stop there.

“This is a really good school,” he told me. “Sometimes you have to get along just to get by. You should still always think for yourself.”

I took my dad's advice to heart. Every piece of it. It was a moment of awakening for me. People – to me at the time, the adults – didn't always have good reasons for why they believed one thing over another. They could be especially sensitive about those things. They didn't have convincing arguments to make, which was upsetting for them.

Get along and think for yourself anyway.

Having the right opinion.

It's been more than twenty years since middle school. I no longer have to charm nuns and theology teachers just to get by. If someone asks me whether or not people ever really change, I can respond honestly. Should they disagree with me, I can argue my points.

So much time has passed, it seems everything is different. Twenty years ago, my little sister was a student at the same Catholic school. She's a doctor now.

Often, I try to think of good questions that I can ask my sister about the human body or about medicine because I never get tired of hearing her answer. It's nothing to do with the answers themselves and everything to do with the way she answers. She speaks in full paragraphs bursting with plain facts, halting herself as she tries to expand on the facts' connections to each other.

It's distinctly unopinionated.

Yet more than anyone I know, my sister will have to form an opinion during her day-to-day because someone else's well-being depends on it. Drawing on years of hard-earned education and experience, her role entails drawing conclusions based on the facts as she understands them. She has built the skill to do that well, as different as it is from the way that she answers questions given the luxury of time and wandering access to information.

Our dad's dad, whom neither my sister nor I ever got to meet because of a lung cancer that took him at age 51, was a firefighter. He would have had to form opinions too. The smoke is moving this way: that hall is probably the safest. A clatter in the kitchen: go through the living room to get to the bedrooms instead. Emergencies would have forced him to draw opinions, like which of two doors to take in a burning building. His opinions, based on a career fighting fires, would have saved lives, just like my sister's opinions do.

Sometimes being right matters a lot.

Having no opinion at all.

One heuristic that may lead to being right, is to avoid being wrong. One way to avoid being wrong, is to never have any opinion at all. But is that possible?

Recently, I've been on a history kick. Don't spoil the surprise for my history-obsessed niece but she's getting a copy of Gordon Wood's Radicalism of the American Revolution for Christmas. Myself I'm building on my understanding of the rivalry between Mohamed Pahlavi and Ruhollah Khomeini, via Scott Anderson's King of Kings, a book about recent Iranian history in which all the US presidents from Johnson to Reagan play a starring role.

In the midst of “here's the slogan I wear on my hat” American politics, it's been comforting to read an explication of such a complex political topic that reminds me so much of the rigorously academic tone my sister uses when she's talking about cardiology. One early chapter of King of Kings ends with this opinion: “Collapse on the magnitude of that which occurred in imperial Iran in the 1970s simply cannot be attributable to the actions of one king or even one royal couple alone. Rather, it required the determined incompetence or cowardice of a great many actors.”

There is something brave about such a loose conclusion: that the reality isn't simple but complex, that a full accounting will involve a lot of uncertainty. You will find no shortage of less-rigorous historians willing to blame the fall of the shah solely on the shah. (This Guardian review of King of Kings actually misinterprets the book and does exactly that.) Or to blame the whole thing on America. Or the French. Or OPEC. And then someone else will put each of these opinions next to the other and think, Here the truth must be, among these pieces.

Yet, often the whole truth just isn't there among the pieces available for you to deduce it out. It turns out not to be knowable. Or if it is knowable, it's complicated and hard to explain; it doesn't make a very clickable headline.

Scott Aaronson, the computer scientist and author of the Shtetl-Optimized blog, had some words about how slippery a thing knowability can be. Last month at the Santa Fe Institute he gave an hour-long talk called How Much Math Is Knowable. In this talk, he explores the limits of the things that we can think, or believe, or know. He refers to the concept of Gödel undecidability, which on a basic level says that any statement we can prove in math relies on a system that has taken something on assumption. You can go back around and prove the assumption, but in doing so you inevitably create a new system with new assumptions.

I didn't learn about Gödel undecidability until a long time after sixth grade. I would have hated it then because it means that for interpreting the world, some amount of opinion is fundamental.

Heartbreaking as it may (and should) be, the minimum amount of opinion seems not to be zero.

Having some opinion.

Let's take this as axiomatic: the minimum amount of opinion needed to interpret the world is greater than zero.

For me, this is an uncomfortable starting point. I like to feel itchy any time I have to say I know something. I like to squirm and stutter when someone puts me on the spot and says “Tell me why that's your opinion.” I don't want to take anything on faith. I don't want to think any of the things that I think because they sound good, or because they feel right, or because that's just what the older, wiser people always told me. I want to go careening through a field of thorny facts and come out the other side scraped and bruised and knowing that if nothing else, I can explain beyond a reasonable doubt why the sun rises in the east and unemployment tends to go down when the Fed cuts the interest rate.

Now the hard part: here's why I think you should think the same way.

The media that is most widely consumed today is like a set of newspapers that someone has cut up and curated lede by lede for personalization. There is no "someone" involved in the actual cutting and curation, though. It's algorithms all the way down. And media that uses algorithms to determine the set of information you see, as both social media and chatbots do, should be innately frightening to you, even before you ever learn anything else about the algorithm owners. Drolly, its purpose in its current form is usually to maintain engagement. To maintain your engagement: it doesn't care which information you get stuck on, as long you get stuck and stay stuck.

So, how are books any different?

Here's an experiment that you can run to compare the substance of a book with the substance of a newsfeed. Spend an hour reading a book, and spend an hour reading a newsfeed. At the end of each hour, write a summary of the points that you've encountered.

The summary of the book will look like a compact version of larger, broader concepts. It may read like this: By the middle of the 1970s, American diplomats had tended to become so used to the status quo under the shah, they rejected any evidence that he might be vulnerable. There seemed to be great social pressure from the ambassador in Tehran to write off disruptions in rural areas as unimportant.

The summary of the newsfeed, on the other hand, will look more like a list of discreet events in roughly the same form that you consumed them. It may read like this: Subway delays are leading to tension among riders. The next Disney movie is supposed to make the young leading actress into a star. Everyone in Los Angeles is upset that hot dogs cost so much at Dodger Stadium.

With the book, I've gotten fewer pieces of information and with them, more uncertainty. Conversely with the newsfeed, I've gotten more pieces of information, presented with almost no uncertainty or ambiguity at all. When algorithm-driven media works toward this end, it's most accurately described as a form of programming. (And you are the Turing machine that is eating the never-ending tape of instructions.)

Conjecture: your default state of knowing anything is zero. The harder it is to move your needle – which is to say, the more empirical evidence you require to approach certainty about anything – the more safely you can navigate algorithm-driven media (if navigate it you must). You should never feel too strongly invested in one conclusion or another. Always insist on running every experiment again. Ask to replace reasonable-sounding statements with falsifiable hypotheses. Go back to your default state of knowing (zero) and from there try to reach the opposite of your original conclusion.

You have the time to look deeper.

It's rare to find yourself standing between two doors in a burning building. You have the luxury of more information to consider, most times. In all situations, opinions are a last resort.

Me, photo used in 2004 middle school yearbook underneath the "most opinionated" superlative.