Things I Was Certain About

SSearch your old messages for the word “never.”

Depending on how long you've been texting, the results may contain a surprisingly detailed history of predictions you forgot making. You were never going to speak to someone again. You would never live somewhere. You would never tolerate a particular behavior, change your mind about marriage, take a certain kind of job, or recover from something that felt unbearable at the time.

Search for “definitely” and another version of the future appears. These were the things you were sure would happen.

Most of these statements were never intended as predictions. They were ordinary expressions of confidence made in conversations that continued moving. Unless something later makes the statement memorable, there is little reason to return to it.

This creates an asymmetry in how we understand our own judgment. We remember many outcomes. We preserve far fewer records of what we expected before we knew them.

The Future Looks Different From the Past

Once something has happened, it becomes difficult to reconstruct how uncertain it once was.

Psychologists have documented this tendency extensively through research on hindsight bias. After learning an outcome, people tend to see it as having been more predictable than it appeared beforehand. The clues that support what eventually happened become easier to recognize, while competing possibilities fade from view.

Personal history is particularly vulnerable to this because we are constantly constructing explanations for how we arrived where we are.

A career can look inevitable once it has become successful. A divorce can make years of earlier conflict look like an obvious progression toward separation. A marriage can transform a chance meeting into the beginning of a love story. A business failure can make every warning sign seem glaring in retrospect.

Before the outcome, however, several futures may have looked equally plausible. The problem is that most of us don't preserve a reliable record of those possibilities.

The Predictions We Forget

Suppose you believe today that you always knew a relationship would fail.

How would you test that?

Memory is not particularly helpful because the failure itself has become part of the memory. You might remember doubts accurately while forgetting periods of confidence that no longer fit the story.

Contemporaneous material can complicate the picture. Perhaps old messages show that six months before the breakup you were looking at apartments together. A journal entry may describe the relationship as the healthiest you've ever had. On the other hand, you may find evidence that you expressed concerns much earlier than you now remember.

Neither result means your present account is dishonest. It simply demonstrates how difficult it is to evaluate an old prediction after the answer has become known.

This problem appears everywhere. We remember saying that a company would succeed when it succeeds and forget the dozen other companies we were equally confident about. We remember the person we immediately distrusted who later proved us right and forget the people we distrusted for no reason at all.

The successful prediction acquires a story. The failed ones usually disappear.

Seal the Answer

Circa8 approaches this problem with a deliberately simple mechanism: answer the question before you know what happens, then seal the answer.

The question might concern something concrete: Do you think you'll still be at this company in two years? It could be relational: Which five people currently in your life do you believe will still be close to you in ten years? Or it could concern something harder to measure: What do you think will matter most to you when you're forty?

Once answered, the response is closed rather than repeatedly edited as circumstances change. It returns only after a defined period or when the question can meaningfully be revisited.

The point is not to create a game in which your future self scores your predictive accuracy. The more interesting value is preserving uncertainty before hindsight has a chance to reorganize it.

When the answer eventually returns, you encounter the future as you once imagined it rather than as you now remember imagining it.

Confidence Is Part of the Record

The prediction itself is only part of what matters. Confidence matters too.

There is a meaningful difference between saying you think a relationship will last and saying you are almost certain it will. There is a difference between reluctantly predicting that you'll stay in your job another year and being unable to imagine ever leaving it.

Preserving that confidence makes later comparison much richer. You may discover that you were right about an outcome for completely different reasons than the ones you gave. You may be surprised by how confidently you predicted something that never came close to happening. Or you may find that a possibility you treated as unlikely eventually became the organizing fact of your life.

Over enough time, these comparisons can reveal something more interesting than whether you are good at prediction. They can show what kinds of futures you systematically overestimate or underestimate.

Perhaps you repeatedly assume difficult feelings will last longer than they do. Maybe you underestimate how quickly relationships change, or consistently imagine that your current priorities will remain stable. Those tendencies are difficult to see from memory because each prediction is usually forgotten once life supplies the answer.

Prediction Becomes Diff

A sealed answer becomes particularly useful when combined with a Diff.

Imagine answering the same set of questions every five years without seeing the previous responses first. What does success mean to you? Who are the most important people in your life? What are you afraid of losing? Where do you expect to live? What do you believe you'll regret if you don't do it?

At thirty, the questions capture one model of the future. At thirty-five, they capture another. At forty, another.

Comparing them does more than show whether predictions came true. It reveals changes in what the person considered worth predicting.

At one age, career may dominate every answer. Ten years later, it barely appears. Someone once treated as permanent may disappear from the record while a person who hadn't yet been met becomes central. A fear that seemed foundational may eventually become difficult to understand.

This is where prediction becomes a form of evidence about identity. What we expect from the future tells us something about who we believe ourselves to be in the present.

The Questions We Can't Answer Yet

Some of the most interesting questions require patience precisely because there is no way to accelerate the result.

Who in your life today will still know you in twenty years?

Which belief you currently hold will embarrass you someday?

What are you taking for granted that you will eventually miss?

Which problem consuming your attention right now will turn out not to matter?

What part of your current life will you someday wish you could return to for one ordinary afternoon?

There is no AI model that can provide the interesting answer to these questions. The answer is whatever happens.

That creates an unusual design constraint. Most technology tries to reduce waiting. A system built around sealed questions sometimes has to protect it. The value of the artifact depends on enough life occurring between the question and its return.

A question sealed for ten years cannot be made ten times better by opening it after one.

An Experiment That Takes a Lifetime

This is also why sealed predictions become more interesting at scale.

Imagine thousands of people answering the same question at twenty-five and encountering it again at thirty-five. Not merely what they hoped would happen, but what they genuinely expected. Which relationships did people overestimate? Which fears disappeared? How often did the thing they believed would make them happy actually do so? Which ambitions survived contact with reality?

Handled responsibly and with the right privacy boundaries, the resulting material could become an unusual longitudinal record of how people imagine their futures and how those futures actually unfold.

For an individual, though, the value is simpler. A sealed answer preserves something we normally lose almost immediately: the world as it looked before we knew what happened.

Years later, you don't have to ask yourself what you thought the future would be like.

You can ask the person who was waiting for it.