Why Good Intentions Fail at the Exact Moment You Need Them

What each one is actually for

Do It Tomorrow

Do It Tomorrow is built around a simple mechanical rule: today's list is closed. Anything that shows up during the day — a new request, an idea, an urgent-seeming email — gets written onto tomorrow's list instead of being acted on now, with rare genuine exceptions. The point isn't to ignore new tasks forever; it's to stop today's committed work from being constantly reshuffled by whatever just arrived, which is what happens by default when every incoming item gets evaluated for immediate action the moment it lands. The list itself is meant to be finishable — a closed, bounded set decided the night before, not an ever-refreshing feed you're always slightly behind on.

Superforecasting

Calibration training asks you to attach a specific number to a belief about the future — not 'this will probably happen' but '70% chance this happens by March' — and then, when March comes, to check whether things you called 70% actually happened about seven times out of ten. Do this repeatedly, across enough predictions, and patterns in your own bias become visible: maybe your 90%s are only right 75% of the time, which tells you something specific and correctable about how you estimate confidence, in a way a vague hedge like 'likely' never could. Tetlock's research on the forecasters who did this consistently found their accuracy improved measurably over time, purely from the discipline of committing to a number and then checking it.

What you get out of pairing them

Both target the same failure point: the moment a general commitment meets a specific situation and the commitment quietly loses. 'I won't let new tasks derail my day' sounds fine until an actual urgent-seeming email arrives at 10 a.m. 'I'll be intellectually honest about uncertainty' sounds fine until you're actually asked a question and 'probably' slides out instead of a number. Both practices work by pre-deciding, before that moment arrives, exactly what happens when it does — so the discipline doesn't depend on remembering to be disciplined right when it's hardest. The payoff shows up less as a single dramatic save and more as a long run of ordinary days where the plan you made in a clear moment actually survived contact with a messier one.

What Do It Tomorrow rests on

• Implementation intention closes intention action gap randomised trials
  Pre-committing to a specific situational cue paired with a specific response lets the behavior fire automatically when the cue occurs, without depending on in-the-moment willpower.
• Closed lists reduce overwhelm practitioner report
  A fixed bounded list of tasks rather than an ever-growing open list reduces the felt overwhelm of an infinite to-do list.
• Deferred capture protects commitment practitioner report
  Writing newly arriving tasks onto tomorrow's list rather than acting on them immediately protects today's already-committed plan from being derailed by reactivity.

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

Do It Tomorrow's rule fires on a recognizable cue — a new task arrives — and the response is fixed in advance: write it down for tomorrow, don't act now. Calibration training's rule fires on a different recognizable cue — you're about to say something vague about likelihood — and the response is also fixed in advance: stop, name an actual number. Different cues, same structure: a specific if-this-then-that rule replacing an in-the-moment judgment call. Neither rule asks you to think harder in the moment it fires; both ask you to have already thought, earlier, about exactly this kind of moment.

The principle underneath both

A plan that specifies the exact trigger and the exact response in advance gets executed more reliably than a general intention left to be applied fresh, under worse conditions, at the moment it's actually needed — because the decision has already been made when judgment was clearest, and the moment of pressure only has to execute it, not make it.

What that makes the real goal

The real target in both cases is making a genuinely hard discipline — protecting your day from constant reactivity, or being honest about uncertainty instead of defaulting to a comfortable hedge — something you actually do under real conditions, not just something you believe in and fail to apply. Neither is really a productivity trick; both are ways of making a value you already hold survive an actual Tuesday.

Other practices that serve the same goal

The broader research finding behind both, called implementation intentions in the psychology literature, shows that a specific 'when X happens, I will do Y' plan measurably outperforms a general goal across dozens of studied domains, from exercise adherence to medication compliance — one of the more robust findings in behavior-change research. Pilots' pre-flight checklists apply the same structure to procedural discipline. Structured decision journals, where you write a prediction and your reasoning before an outcome is known, apply calibration's honesty requirement to business or personal decisions specifically. A simple email rule that auto-files anything not from a fixed list of people into a folder reviewed once daily achieves a version of Do It Tomorrow's deferral without requiring the same in-the-moment discipline at all, at the cost of occasionally deferring something that actually was urgent.

Using Do It Tomorrow's Deferral Rule to Protect a Calibration Habit

Calibration training asks you to pause and assign an actual number before you answer a prediction question, which takes a few extra seconds you don't have if you're mid-conversation and someone's waiting on you. The urge, under time pressure, is to say 'probably' and move on — exactly the vague hedge the whole practice exists to eliminate. Borrow the deferral structure directly: when a prediction question arrives in a moment you can't properly sit with it, don't answer with a hedge and don't force a rushed number either — say 'let me actually think about that and get back to you,' and write the question down for a specific slot later that day or the next, the same way a new task gets deferred to tomorrow's list. The number you give later, with a clear head, is worth more than a hedge given now.

Using Calibration to Decide Which Incoming Tasks Actually Break the Rule

Do It Tomorrow's deferral rule has one weak point: it allows for 'rare genuine exceptions,' and deciding what counts as genuinely urgent, in the moment, under the pull of whatever just landed in your inbox, is exactly the kind of judgment call people get wrong by overestimating urgency. Calibration training's habit of forcing an actual number gives you a fast gut-check: before breaking your own rule, ask yourself for a specific probability — not 'this feels urgent' but 'what's the percentage chance this actually can't wait until tomorrow without real consequences?' Most things that feel urgent, forced into a number, land closer to 20% than 90%, which is usually enough to send them to tomorrow's list after all.

10:14 a.m., an Email Marked 'Urgent,' Today's List Already Full

You sit down with today's list already set — three things you decided last night actually matter. At 10:14, an email lands with 'URGENT' in the subject line from someone two levels up, asking for a document by end of day. Every instinct says drop everything. Instead of deciding in that instinct, you ask yourself the calibration question: what's the actual chance this genuinely can't move to tomorrow morning without a real problem? You sit with it for ten seconds — not zero, but honestly, maybe 15%, since 'end of day' from someone who didn't call or walk over usually has more slack than the subject line implies. You reply that you'll have it to them first thing tomorrow, write the task on tomorrow's list where it belongs, and go back to what you'd already decided mattered most today. The subject line said urgent. The number said otherwise, and the number was closer to true. By lunchtime nobody has followed up, which is itself one more data point for the next time a subject line tries to make the decision for you.

When this is worth reaching for

Your day gets reliably hijacked by things that feel urgent in the moment, and you also find yourself reaching for vague words — 'probably,' 'likely,' 'should be fine' — instead of actually committing to how confident you are, whether in planning conversations, forecasts, or decisions about what to prioritize. In some roles the two problems compound each other: constant interruptions plus a steady stream of judgment calls about what's actually worth reacting to. Someone managing a team through ambiguous, fast-moving conditions is a common example — reactivity and vague hedging both tend to be worse there than in a quieter, more predictable job.

Where this does not apply

If someone's job genuinely involves no forecasting or probability estimation at all — no predictions, no forward-looking judgment calls of any kind — calibration training has little to attach to in that specific work, even though the underlying habit could still be practiced privately elsewhere. And if the tasks arriving during the day are truly non-deferrable — a live customer outage, a medical situation, anything with real same-day stakes — the deferral rule doesn't apply; the method is built for the far more common case of tasks that only feel non-deferrable.

What the evidence does and does not support

Implementation intentions — the mechanism both practices lean on — are supported by one of the larger and more consistent bodies of randomized-trial evidence in behavior-change research, spanning goal domains well beyond either task management or forecasting specifically. The individual components of calibration training, like forcing explicit numeric probabilities and reviewing outcomes against them, have their own trial support from forecasting-tournament research, including the Good Judgment Project, which tracked real forecasters against real outcomes over several years. Do It Tomorrow as a complete, named system is practitioner-reported rather than independently trial-tested — it's a specific application of a well-supported mechanism, not itself the subject of a controlled study. No research has tested whether using the deferral rule to protect calibration practice, or the reverse, outperforms either approach used alone, and anyone claiming a measured multiplier effect from combining them is stating an opinion as if it were a finding.