Irreversibility and the Value of Waiting

The idea in plain words

You can always build a factory next year. You cannot un-build one this year. When an investment is irreversible and its future value is uncertain, deciding today kills an option you would otherwise hold: the option to wait, learn more and then decide. That option has value. A standard business case, which compares the cost of the investment with its expected return, ignores it, and therefore tends to commit too early.

The same logic runs in reverse. Some things are irreversible if you do not act: a market window, a talent pool, a standards battle. Then waiting destroys the option. The art lies in seeing which way the irreversibility runs.

Where it comes from

In 1974 two papers, one by Kenneth Arrow and Anthony Fisher and one by Claude Henry, made the point in the context of environmental policy. Developing a wilderness is irreversible; preserving it is not, since it can still be developed later. Under uncertainty about future values, they showed, the irreversibility itself justifies more caution than a naive expected-value comparison would suggest. The extra value of keeping options open came to be called quasi-option value.

In the 1980s and 1990s the idea was absorbed into corporate finance as real options. Avinash Dixit and Robert Pindyck’s Investment under Uncertainty (1994) showed that firms facing irreversible investments should wait until expected returns exceed costs by a substantial margin, not merely break even, and that this helps explain why firms often invest less than simple calculations predict. The value of flexibility grows with uncertainty.

The managerial version that spread most widely is Amazon’s. In the 2015 shareholder letter, Jeff Bezos distinguished one-way doors, consequential and nearly irreversible, from two-way doors, which can be walked back through. The first deserve slow, careful process; the second should be made fast by small groups.

There is a much older, existential version. In The Concept of Anxiety (1844), Kierkegaard described anxiety as the dizziness of freedom: the vertigo a person feels on recognising that they can choose, and that choosing will close off other possibilities. A century later Sartre wrote that we are condemned to be free: there is no way to avoid choosing, since refusing to choose is itself a choice. Irreversibility is what gives choice its weight. A decision that could be undone without cost would hardly be a decision at all.

What it does to the mind

People handle irreversibility badly in both directions. Some, faced with a large irreversible choice, delay indefinitely, unable to bear the closing of options. Experiments in behavioural economics have found that people will pay to keep options open even when doing so lowers their total payoff. Others, especially in organisations that prize decisiveness, commit early to escape the discomfort of uncertainty, and then defend the commitment.

Irreversibility also interacts with confidence. The more certain we feel, the less we value the option to wait. Overconfidence therefore leads directly to premature commitment.

Where it shows up

Iridium committed billions to a satellite network that could not be redesigned once launched, while cheaper terrestrial networks spread quickly. Itanium committed an industry to a new instruction set years before it shipped. Nokia’s 2011 decision to adopt Windows Phone, in the Burning Platform memo, was explicitly framed as a jump that could not be reversed. The Tesla Master Plan is a staged strategy: each step funds and teaches the next, keeping later commitments open until earlier ones succeed. The virtualization essay is in part a history of technologies that made computing decisions more reversible.

What it asks of someone deciding

Classify the decision before you deliberate. Which parts are one-way doors and which are two-way? Spend your caution on the first.

Stage commitments. Pilot, prototype or sign short contracts where you can, so that you buy information before you buy permanence.

Check the window. Waiting is valuable only if the option will still be there. In markets with strong network effects, the cost of delay can be the whole market.