Using Possible Selves Exercise to enhance Superforecasting

What each one is actually for

Possible Selves Exercise

The Possible Selves Exercise asks you to stop having a fuzzy feeling about your future and write two people down on paper: the specific version of you that you're hoping to become, and the specific version you're afraid of becoming, both in enough detail that you could picture a Tuesday in their life. Not 'successful' — the actual scene: what you say in the meeting, what's on your desk, whether you're the one who apologizes first. Not 'stuck' — the other actual scene: still checking the same job posting at midnight three years from now. You write both, in the same sitting, and let the distance between them do the motivational work.

Superforecasting

Calibration training, the discipline behind Philip Tetlock's superforecasters, takes a belief you already hold about some uncertain future event and forces you to say it as a number — not 'probably' or 'unlikely,' but 70%, or 15% — and then, when the event resolves, you go back and check: were the things you called 70% actually true about 70% of the time? Not once. Repeatedly, on a running scorecard, until your gut sense of 'probably' and the actual frequency of being right start to line up. The number is what makes the belief checkable at all; the checking is what makes the number improve.

What you get out of pairing them

Most people's relationship to their own future is a single vivid scene they replay — the dream job, the version of themselves who finally has it together — and they never once check it against anything. It stays a mood. Calibration training's real gift, applied here, is that it turns your hoped-for and feared self from a feeling you have into a claim you can be wrong about on a specific date, which is the only kind of claim that can actually move you, because a feeling doesn't have a deadline and a forecast does. Going the other direction, most people doing calibration training only ever forecast about the external world — elections, markets, whether a project ships on time — and never turn the same rigor on the question that would actually keep them doing the tedious tracking work three months in: who becomes of me if I keep doing this, and who becomes of me if I stop.

What Possible Selves Exercise rests on

• Concrete goals outperform vague aspirations observational studies
  Specifying a concrete hoped-for or feared image of the future is more motivating and actionable than an abstract aspiration.
• Future self continuity motivates present action observational studies
  Vividly imagining one's own future self increases psychological connection to that self and motivates present behavior aligned with it.
• Gap awareness motivates change plausible mechanism
  Awareness of the gap between one's current self and a hoped-for or feared future self motivates action to close that gap.

What Superforecasting rests on

• Decomposition improves estimate accuracy observational studies
  Breaking a complex uncertain question into smaller more tractable sub-questions produces more accurate estimates than judging the whole question at once.
• Explicit probability estimates improve calibration randomised trials
  Forcing a numeric probability rather than a vague verbal hedge makes a prediction falsifiable and trainable.
• Outcome feedback corrects judgment bias randomised trials
  Regularly scoring predictions against actual outcomes and reviewing the misses corrects systematic biases in judgment over time.

Where they actually meet

Both practices are trying to do the same underlying thing to a fuzzy internal representation: haul it out of the realm of vague feeling and into the realm of a specific, checkable claim. The Possible Selves Exercise does this to your identity — 'successful' becomes a scene you could describe to a stranger. Calibration training does it to your beliefs about the world — 'probably' becomes 65%. Neither leaves it at the description stage, either: both build in a return visit. Possible Selves relies on the gap between the two scenes pulling you forward day to day; calibration training relies on the scorecard you check after the event resolves. Different material, identical move: name it specifically enough to be wrong about, then go find out if you were.

The principle underneath both

A vague internal state — a feeling about who you're becoming, a hunch about how something will turn out — is nearly inert. It doesn't move you and it can't be corrected, because there's nothing precise enough in it to check against anything. The moment you force it into a specific, dated, checkable form, two things become possible that weren't before: it can motivate you (because now there's a concrete gap to close, or a concrete date to beat), and it can be wrong (because now there's something for reality to disagree with). Precision is what turns a mood into a lever.

What that makes the real goal

The Possible Selves worksheet and a forecasting log are both just one route to the same underlying goal: build an explicit, checkable model of an uncertain future — whether that future is a fact about the world or a fact about who you'll be — and use the act of checking it against what actually happens to correct the model and, in the identity case, to correct your behavior. Naming a percentage and writing a scene are the same move applied to different unknowns.

Other practices that serve the same goal

Decision journals — writing down a call before you make it, along with your reasoning and a confidence level, then rereading it after the outcome is known — chase the same objective and are common practice among professional traders and poker players, though the evidence for them is mostly anecdotal rather than trial-tested. Cognitive behavioral therapy's behavioral experiments do it with clinical-grade evidence behind them: a person predicts, specifically, what will happen if they do the feared thing ('everyone will notice I'm shaking'), does it, and then compares the prediction to what actually happened, which is calibration training aimed at a fear instead of a market. Premortems — imagining a future failure in detail and working backward to its causes before starting a project — hit the same target from the feared-outcome side, and carry mostly observational support. All three, like the pairing here, make an internal guess specific enough to be checked.

A tradition that already does this

Cognitive behavioral therapy's behavioral experiments already do a version of this pairing, and they predate the term 'superforecaster' by decades: a therapist has a client write down a specific, falsifiable prediction about a feared personal outcome, assign it a rough confidence, take the action, and then compare the prediction to what actually happened. That is calibration training run on the question 'who am I, and what happens to me' rather than on the question 'who wins the election' — the exact move this page is proposing, already standard practice in one clinical lineage before the forecasting research existed to name it.

Using Possible Selves to enhance calibration training

The well-documented failure mode of calibration training is not inaccuracy — it's attrition. People start a forecasting habit with real enthusiasm, keep a spreadsheet of predictions about the news or their industry for two or three weeks, and then quietly stop, because checking whether you called the Fed's next move correctly doesn't feel like it's about you. Here is the move: once a month, alongside your external forecasts, write two Possible Selves entries about your own forecasting practice specifically — the version of you a year from now who has eighteen months of tracked, scored predictions and has visibly gotten better at reading uncertain situations at work, described as a scene, not a trait; and the version of you who has a folder of three abandoned tracking spreadsheets and still argues from gut feeling in meetings, also described as a scene. You are not forecasting the world in this exercise. You are doing to your own identity as a forecaster exactly what you already do to interest rates: making the future specific enough to want, or specific enough to dread, instead of a habit you can quietly let lapse.

Using calibration training to enhance the Possible Selves Exercise

The well-documented failure mode of the Possible Selves Exercise is the opposite one: people write the vivid hoped-for and feared self beautifully, feel the motivational jolt for a week, and then never open the page again — the exercise has no return visit built in, so the gap it names never gets checked against anything real. Borrow the forecaster's discipline directly. After you write your feared self ('still checking that job posting at midnight, three years from now, if nothing changes'), attach an actual number to it as a genuine prediction, not a rhetorical flourish: 'I'd put this at 40% likely if I make no changes to how I spend Thursday nights.' Put a date on your calendar six months out. When it arrives, don't rewrite the exercise from scratch — pull out the old page and score yourself honestly against the number you wrote, the way a forecaster reviews a resolved question. Did the gap close, stay the same, or widen? That single scored check-in does more to keep the feared self from becoming a well-worn but inert worry than any amount of re-reading the original vivid paragraph, because now there's a number that can turn out to have been wrong.

Six months after the worksheet

You wrote the Possible Selves page in January — the hoped-for you running the team meeting without notes, the feared you still cc'ing your manager on emails you should be able to send alone — and it felt true and a little frightening at the time, the good kind of frightening. It's July now. You haven't looked at the page since. If someone asked you tonight whether you're closer to the first version or the second, you'd have a feeling but no actual answer, because nothing about the exercise told you how you'd know. Now imagine you'd done one extra thing back in January: written '55% chance I'm still cc'ing my manager on routine emails by July if I don't change anything' on the bottom of the page, and put a reminder in your calendar for today. You'd pull up the page right now, not to feel something, but to check something — and whatever the answer is, you'd know exactly what to write down for the next six months, because a scored miss tells you more about what to do next than a feeling ever does.

When this is worth reaching for

You will know it is time when a Possible Selves entry has gone stale — you can still picture both versions of yourself vividly but have no idea whether you're actually closer to one or the other, because you never built in a way to check. You will know it from the other side too, when a calibration-training habit has become mechanical: you're still logging forecasts about the news or the market on schedule, but you've lost the sense of why any of it matters, and turning the same rigor briefly on your own trajectory tends to restore the stakes.

Where this does not apply

If your Possible Selves practice is doing real work as a felt, emotional experience — the vividness itself is what moves you, and it's actually changing your Thursday nights — forcing a premature number onto it can flatten exactly the quality that was working, turning a motivating image into a homework assignment you resent. And if you're doing calibration training purely to get better at judging the external world — markets, elections, project timelines — for its own sake, there's no obligation to import identity work into it; a forecaster who has no attrition problem and enjoys the practice as is doesn't need this addition, and adding it could just be a distraction from the discipline of estimating things outside yourself.

What the evidence does and does not support

The pieces here rest on genuinely different levels of evidence. Explicit numeric probability estimates and outcome-based feedback improving calibration are supported by controlled research, most visibly Tetlock's Good Judgment Project. The motivational effect of vividly imagining a concrete future self is observational and mechanistic — real correlational and lab evidence, not trial evidence, and it hasn't been isolated as a stand-alone cause of long-run behavior change. The pairing itself — using a scored forecast to sustain a Possible Selves practice, or using Possible Selves to sustain a forecasting habit — has not been studied at all. What's offered here is a shared underlying pattern, reasoned from what each half is separately known to do, not a tested combination.