Using Self-Concordant Goal Setting to enhance Superforecasting

What each one is actually for

Self-Concordant Goal Setting

Self-concordant goal setting is a screening step you run before committing real effort to a goal: you ask, honestly, whether wanting this thing traces back to your own interests and values, or whether it traces back to guilt, obligation, comparison with someone else, or a reward you think you're supposed to want. The test is not "do I feel excited right now" — excitement is cheap and fades by Tuesday. It's closer to: if no one were going to know whether I did this, and no one would judge me for skipping it, would I still want it? A goal that survives that question tends to keep generating its own effort, because the reason for doing it does not depend on an audience. A goal that fails it usually still gets started — plenty of obligation-driven goals do — but it tends to get abandoned at the first bad week, because the only thing that was pushing you was pressure from outside, and pressure from outside is not there at 6 a.m. when the alarm goes off.

Superforecasting

Calibration training is the practice, built around research on people who turn out to be unusually accurate at predicting world events, of forcing a belief into a specific number before you find out if you're right. Instead of saying "this will probably work out," you say "70 percent," write it down with a date attached, and then, when the date arrives, you check what actually happened and score yourself: were the things you called 70 percent right about seven times out of ten, or did they only pan out half the time, which would mean you're consistently more confident than your track record earns? Doing this repeatedly, across enough predictions, is what makes a hunch trainable instead of just felt — you stop asking "does this feel true" and start asking "how often have hunches that felt this true actually turned out to be true." The skill that gets built is not predicting the future. It's noticing, with evidence, exactly how much your gut can be trusted and under what conditions it can't.

What you get out of pairing them

What this pairing gives you is a way to catch yourself before the six-month version of a mistake, using the cheap, few-minutes version of the same mistake. Right now, when you decide a goal is "really yours," you are trusting a feeling that has no track record attached to it — you've never gone back and checked whether the goals that felt self-concordant at the start actually turned out to be the ones you stuck with, or whether that feeling is just as unreliable as any other unchecked hunch. Borrow the forecaster's habit and you get a track record. You start to learn, specifically about yourself, how much your own sense of "this is genuinely mine" can be trusted, whether it holds up better for creative projects than for fitness goals, say, and where it tends to fool you. That is worth more than getting any single decision right, because it changes how much weight you put on the feeling itself, for every decision after this one.

What Self-Concordant Goal Setting rests on

• Autonomous motivation sustains effort randomised trials
  Goals pursued because they align with a person's own interests and values are sustained with more effort and satisfaction than goals pursued from external pressure or guilt.
• Goal source predicts attainment observational studies
  Whether a goal originates from intrinsic interest versus extrinsic obligation predicts whether a person actually attains it over time.
• Self concordant attainment increases wellbeing observational studies
  Attaining a self-concordant goal increases wellbeing in a way that attaining a non-concordant goal does not.

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

Look at the actual move in each practice, not the goal each one is used for, and they turn out to be doing the same thing to different material. Self-concordant goal setting takes an intention that lives only as a feeling — "I want this" — and forces it to answer a specific question before it gets resources. Calibration training takes a belief that lives only as a feeling — "this will probably happen" — and forces it to answer a specific question, a number checked later, before it gets treated as knowledge. Swap "do I want this for my own reasons" for "70 percent, and I'll check back in March," and you are running the identical test — turn the vague internal thing into something specific enough to be wrong about — on two different kinds of internal thing.

The principle underneath both

An internal signal — a want, a hunch — is worth almost nothing on its own, and only becomes trustworthy once it has been made specific enough to check against what actually happens. Vagueness is what lets self-deception survive. "I really want this" and "I'm pretty sure" are both comfortable precisely because they can never be proven wrong. The moment you convert either one into something falsifiable — a source you can trace, a number you can score — you have taken away its main defense.

What that makes the real goal

The goal underneath both is to know, with some actual evidence, which of your own internal signals deserve to be trusted with your time — and which ones are just noise wearing the costume of conviction. Once that is the goal, it stops belonging to goal-setting or to forecasting specifically. It becomes a general practice: before committing real effort or real certainty to anything, make the internal signal specific enough to be checked, and then actually check it later instead of letting the feeling fade unexamined.

Other practices that serve the same goal

Other practices aim at the same target from different angles. A decision journal — writing down, at the moment you decide something, what you expect to happen and why, then rereading the entry months later — does for everyday choices roughly what calibration training does for world events, and people who study decision-making report that it reliably corrects for the way memory quietly edits a decision to look smarter in hindsight than it was. The "consider the opposite" technique, studied in judgment research, asks you to actively argue against your own first conclusion before locking it in, and has been shown to shrink the overconfidence that a bad goal-choice and a bad forecast share as a common failure mode. A pre-mortem — imagining, before you start, that the project already failed, and writing down why — surfaces the same kind of blind spot a self-concordance check is trying to catch, from the opposite direction: instead of asking whether the motive is real, it asks what would prove the motive was wrong. None of these were built with either lineage in mind. They earn their place here because they attack the identical target: a signal you have not yet made accountable to reality.

Using Self-Concordant Goal Setting to Enhance Superforecasting

Calibration training only works if you keep making predictions long enough to build a track record worth scoring, and the single biggest way people quit is picking questions that never mattered to them — forecasting an election because it's trending, or a stock because someone mentioned it, rather than a question connected to something they actually care about. Before you add a new question to your list, run it through the self-concordance test: would I still want to know how this turns out if nobody were grading my forecasting skill? Will this matter to me — not to my streak, not to my leaderboard rank — in six months? A question about whether a project at your own company will ship on time survives that test in a way a question about a celebrity's divorce usually does not, because you have a reason to keep paying attention that has nothing to do with the practice itself. The concrete move: before each new forecast, write one sentence naming why you, specifically, care how this resolves. If you can't write that sentence, drop the question and pick one you can. This filters out the forecasts you'll quietly stop updating in three weeks, which is where most calibration practices actually die — not from lack of skill, but from tracking things nobody cared about.

Using Superforecasting to Enhance Self-Concordant Goal Setting

The self-concordance question — is this really mine — is usually asked once, answered with a feeling, and never revisited, which means you never learn whether your own sense of "this is genuinely mine" is any good. Borrow calibration training's actual mechanic to fix that: the next time you're deciding whether to commit to something — a course, a move, a relationship pattern you want to change — write down a specific, dated prediction instead of a vague verdict. Not "this feels right," but something like, "80 percent chance I still want this and I'm still doing it, unprompted, eight weeks from today." Put a date on your calendar to check. When the date comes, score yourself honestly: did you keep going because you wanted to, or because you'd already told people, or because quitting felt like admitting a mistake? Do this across ten or fifteen decisions over a year and you get something a single gut-check never gives you — a real answer to how reliable your own sense of "this is mine" actually is, and under what conditions, new relationships, health goals, career pivots, it tends to be overconfident.

Sunday night, and a friend has just talked you into a marathon

It's Sunday night and a friend has just talked you into signing up for a marathon in eight months. Your body is still buzzing a little from the conversation — this is the excitement that self-concordant goal setting warns you not to trust on its own. Instead of registering right now, you sit with two questions. First, the self-concordance one: if your friend moved away tomorrow and nobody would ever ask how training was going, would you still want to do this? Second, the calibration one: you write, in your notes app, "Prediction: 75 percent chance I am still running four times a week, unprompted, in six weeks," and you set a reminder for that date. Six weeks later the reminder goes off. You're running three times a week, not four, and one of those runs was because your friend texted asking about it, not because you wanted to go. That's useful information delivered on time, not a moral failure — you register it, adjust your prediction method for next time, maybe your "excited on a Sunday night" feeling runs about fifteen points too confident, and you decide, with actual evidence now, whether to keep the marathon or scale it down to a distance you can picture wanting on a Tuesday morning with nobody watching.

When this is worth reaching for

A career change, a demanding training block, a decision to end or repair a long relationship pattern — a commitment big enough that getting the motive wrong is expensive, with enough runway before any deadline to actually check a prediction against what happens. It matters most for someone with a history of starting things on real enthusiasm and quietly abandoning them within a season, since that pattern is the exact signature of an unchecked concordance signal. A decision reversible enough to test in miniature first, and score before going all in, is the other place this pays off.

Where this does not apply

This is overkill, and mildly insulting to yourself, for goals that are not actually discretionary. If the choice is take my child to the doctor or file the tax return, running a self-concordance screen on it wastes analysis on something you were never going to opt out of regardless of its source. It also breaks down for anything without a checkable resolution: calibration training needs a moment where you can say, plainly, whether the prediction came true, and a goal like become a kinder person rarely resolves that cleanly on any single date, so forcing a number onto it produces false precision rather than real feedback. And for people who already tend toward over-analysis before acting, adding a scoring layer to every decision can become one more way to delay starting anything at all — in that case, the more useful move is picking something and gathering the evidence in hindsight, not in advance.

What the evidence does and does not support

Each half of this pairing has its own real evidence: the link between a goal's origin and how much effort and satisfaction it produces has shown up across observational and experimental studies for over two decades, and forecasting tournaments built around calibration scoring produced controlled, replicated evidence that ordinary people who track and score their predictions get measurably more accurate over time. Neither fact tells you anything about what happens when you use one practice's discipline to interrogate the other's judgment call, because nobody has run that study. There is no trial where one group screened their goals for self-concordance using calibrated predictions and another group used the ordinary gut-check, with follow-through as the outcome measure. What is offered here is a mechanism-level argument — both practices attack the same failure, an unchecked internal signal — not a demonstrated result for the combination. Treat the specific technique, writing a dated, scored prediction about your own motivation, as a plausible experiment to run on yourself, not a proven method.