The future gets a discount, and the rate is not constant

Discounting isn’t the puzzle

Preferring a benefit sooner is entirely sensible. The future is uncertain, arrangements fall through, and a resource in hand can be used in the meantime. Any reasonable account of choice over time includes a discount, and the existence of one isn’t evidence of any failure of judgement.

The puzzle is the shape. If the discount were constant per unit of delay, then the ranking of two future options would never change simply because time passed, since both would be discounted by the same proportion. Ask somebody to choose between a smaller reward at one distant date and a larger one slightly later, and they generally take the larger. Move both dates forward until the smaller one is available immediately, and a substantial share switch.

What the reversal implies

A preference that flips as the near date arrives, with no new information, means the discount applied to the immediate present is steeper than the discount applied between any two later moments. The near term is treated as a special case rather than as one more interval in a smooth series.

This has been demonstrated in many forms and the qualitative pattern is not in much doubt. The models built to describe it are another matter. Various functional forms are used, they fit the data to different degrees, and the parameters people estimate from experiments are notoriously unstable — they move with the size of the reward, the domain, whether the outcome is a gain or a loss, and how the question is asked. Treat the models as descriptions of a pattern rather than as measurements of a personal constant.

Why this makes plans systematically dishonest

The version of you that makes a plan is discounting all the relevant moments at the gentle distant rate, because every moment involved is distant. The version that executes it is standing in the near term, where the steep rate applies. They are not disagreeing about the facts. They are evaluating the same options with different weights, and the planning version has no way to experience the weights the executing version will use.

This explains why plans made for a future week are so often reasonable and so rarely followed, and why the failure feels like weakness rather than what it is. It also explains why the same person can be genuinely sincere at both moments. Sincerity was never the variable.

What follows for how you arrange things

The useful implication is that decisions should be made as far from the moment of action as possible, and then made binding, because distance is the condition under which your preferences are the ones you endorse. Choosing next month’s arrangement now is not a trick; it is choosing while the steep discount is not applied to anything.

The second implication concerns where to put the friction. A steep near-term discount means small immediate costs are weighted heavily, so a trivial obstacle in front of a tempting option is worth far more than its size suggests, and a trivial obstacle in front of a desirable one does correspondingly more damage. This is the same mechanism that makes defaults powerful, viewed from the time dimension rather than the effort one.

The interpretations that are argued about

Whether present bias reflects a genuine inconsistency or a reasonable response to uncertainty is genuinely disputed. If a promise about next year carries real risk of not being honoured, then discounting it steeply is not irrational, and some of what looks like impatience in experiments may be scepticism about whether the delayed reward will actually arrive.

A related dispute concerns whether measured discount rates say anything stable about a person. Attempts to link them to real-world outcomes have produced mixed results, the correlations between different elicitation methods are often weak, and the whole enterprise of estimating an individual parameter has been criticised on those grounds. What survives is the pattern of reversal, which is robust, rather than any particular number attached to anybody.

Using the asymmetry rather than fighting it

You will not persuade the near-term system to apply the distant rate, and attempts to do so by contemplating long-run consequences at the decisive moment are famously unproductive. What works better is arranging for the decisive moment to have already passed: automatic transfers, subscriptions to things you want more of, deliveries scheduled, commitments announced.

It also helps to bring some part of the delayed benefit forward, since the steep discount applies to whatever is immediate rather than to whatever is important. Making the desirable option pleasant now — not as a reward afterwards but as a property of doing it — is one of the few moves that changes the near-term calculation instead of trying to overrule it.