Reversibility (6.3) - A primer
How costly would it be to reverse this if it turns out to be wrong?
What It Is · Key Takeaways · Do’s and Don’ts · Pair With · Things to Explore
Hernán Cortés did not burn his ships. In 1519, on the coast at Veracruz, he had the metal and rigging stripped out of them and then had them run aground and sunk. His men were told the ships were no longer seaworthy. The burning is a later invention. It first appears in 1558, painted on a tomb in Mexico. Other commanders really did set theirs on fire. Alexander, for one. Agathocles of Syracuse burned his fleet in front of his army in 310 BC. Agathocles made his point explicit. Cortés positioned it as a maintenance issue.
What It Is
Some decisions can be unmade, some can’t, and the difference usually has less to do with the decision than with the process around it. Reversibility is the finality pattern of Momentum, the sixth category of the Negotiation Pattern Language, where Pathos meets Structure. Load is how much is simultaneously held by an actor; Velocity is the pace of decision making; Reversibility is the cost the process imposes on reverting a decision. Flexibility is the options you still have open. Reversibility is what it would cost to reopen the ones that were closed.
Hard reversibility costs are formal and can be settled. A breakup fee, a formal go/no-go gate, elimination from a supplier shortlist — each has a knowable cost. You can look at it and decide if it’s worth paying. Soft costs are social and can’t be paid off. You can’t put a price on saving face, and there is no way to trade away a reputation for being true to your word. In the long run, the soft costs are often the largest.
Because the cost of commitment is determined by the process and not the substance, reversibility can be designed and modulated. The same number is sticky or provisional depending on whether it was spoken or written, shared face to face or in front of an audience, or labeled as a proposal or an exploratory idea. Soft costs accrue through a sequence of individually innocent events — a verbal figure, a follow-up email, a briefed counterparty principal — until the window has shut with nobody able to pinpoint when it happened. The accumulation can crystallize into Frozen Suboptimality: an end result nobody really wanted, but everybody considers too costly to reopen.
Key Takeaways
I.Local flexibility is no evidence of global flexibility.
An actor can have full range of motion on the individual issues, and still be unable to withdraw from the negotiation itself. There's no walking away from a renewal with a monopoly supplier or a merger already announced and priced by the market. Local and global move independently. The signature is an actor who is combative on substance but anxious at the table. He frequently walks out of the room. And always comes back.
II.Hedging is how a party with room to maneuver paces himself.
Conditions, disclaimers and carve-outs get read as weak resolve. They are the conduct of someone preserving his options, and releasing his commitment in controlled increments: as much as this stage of the process requires, and nothing more. Low reversibility sounds different: flat, unhedged assertions, or no offer at all. A party that can't afford to retract can't go on record in the first place. The party who sounds most certain is often the one with the least room.
III.Irreversibility doesn’t just make people careful. It makes them behave differently.
Jeff Bezos writes about one-way and two-way doors in Amazon's 2015 shareholder letter. A door you can come back through leads to decisions in the room. One-way doors lead to phone calls to HQ, escalations to senior decision makers, or insurance items getting added to the negotiation issue list. The pace slows down and free exploration gets stifled when what you say can and will be used against you. Caution is the least of it. What you get is a slower process, a longer issue list, and an expanded or different counterparty.
IV.Positions only hold because if they are anchored.
Stated positions without a reason behind them are just preferences. When a number has no reason, why not any other number? The same stated position backed by a rationale (“We need to meet our cost of capital,” “We have bigger customers who pay more,” “Our CFO never accepts this clause”) is stickier. The price of a concession would now include the loss of the rationale, and that's the more expensive part. What isn't backed by a reason doesn't survive the first challenge.
A Few Do’s and Don’ts
Don’t Let a tentative position travel further than you meant it to.
What you said, who heard it, and what the other side is free to brief upward together decide the stickiness of a position you may have never intended as agreed or conceded. Call it out as soon as you see them run with it. A request to put “a rough figure” in writing will be packaged as a non-binding formality. It usually comes from someone who needs a number to harden and wants you to help with the hardening.
Do Audit your global reversibility before the process design is set.
Most negotiators are aware of the hard costs and can tell you what they are contractually free to do. Far fewer have a good grip on the accumulating soft costs. A self-audit doesn't take long, and it's worth doing before the process starts making decisions for you.
Don’t Take a stated deadline at face value.
With a real deadline, the counterparty can say what will happen when it passes. (“The committee does not sit again until March, the authority expires, the press release window for the trade show will be missed…”) A manufactured deadline just produces atmosphere (“I can't guarantee where things will be,” “Engineering will not look favorably on your responsiveness”). A compressed decision window is how people get talked into things they can't undo. Ask what happens, then listen for a specific event on a concrete date.
Do Probe whether the barrier is a fact or a price.
Many positions presented as irreversible are social costs dressed up in factual clothing. A budget is a human construct, not a law of nature. Very little in a negotiation is genuinely undoable. Most of it is a trade-off, and trade-offs have prices (in money, time, goodwill or standing). The question is rarely whether you can reverse. It’s what it would cost, and whether you’re willing to pay.
Governments make this error in public. In 2017 the United Kingdom gave formal notice that it was leaving the European Union, and both sides then bargained for two years on the shared understanding that the notice could not be taken back. In December 2018 the European Court of Justice held that it could have been — unilaterally, at any point before departure — on the reasoning that a stated intention is by definition a thing capable of changing. Nobody had tested that wall. (Whether anybody was prepared to pay the political price is another matter.)
Pair With
Velocity (6.2). Velocity sets the rate at which the decision gates close; Reversibility determines what reopening them costs. Fast pace with high reversal cost produces outcomes that don't make complete sense viewed at arm's length, a telltale sign of suboptimal process.
Collateral (1.3). Reputational exposure is a common source of soft reversal cost. It's one of the reasons repeat negotiators will approach an otherwise identical case differently than a one-off.
Identity (3.1). A position articulated as “this is who we are” costs coherence to revise. Once character is on the line, backing down is even costlier than saving face.
Convergence Mechanism (5.3). Irreversibility can be a closing device. In major-league baseball salary arbitration, the arbitrator picks one of the numbers submitted by the club and by the player. There is no middle ground and no appeal. The overwhelming majority of cases settle before the arbitration hearing, and the submitted numbers are usually reasonable.
Things to Explore
The origin text
Thomas Schelling, The Strategy of Conflict (1960). Chapter 2, “An Essay on Bargaining,” is where deliberate destruction of your own freedom to concede entered the literature. Schelling emphasizes two things. The commitment has to be visible to the other side to be credible, and it needs to be backed by a reason why the entrenched position is the logical outcome. His examples are drawn from arms control, but they transfer to commercial cases.
The commercial treatment
David Lax and James Sebenius, 3-D Negotiation (2006), chapter 10. Their term is the insecure contract: an agreement in which one side has already sunk an investment it cannot recover, which hands the other side a standing incentive to reopen the terms. Read it for the four countermeasures they name. Read it also for the uncomfortable part of the diagnosis: the exposure is the result of your own commitment rather than the counterparty's bad faith, meaning it is on you to prevent and not anybody else's to apologize for.
A case worth an evening
The Helsinki Final Act, 1975. Thirty-five states signed a document that was deliberately not a treaty. The Soviet delegation insisted on an architecture without legal force, ratification or enforcement mechanism. Because it was non-binding, Moscow believed it was harmless. And yet, over the next decade monitoring groups turned that unenforceable text into a standard Moscow could not be seen to break. Search on Helsinki Final Act, Basket III and Moscow Helsinki Group.
And one you would not expect
Dr. Strangelove (1964). Thomas Schelling was handed a copy of Peter George's novel Red Alert on a 1958 plane ride. He was asked to write a magazine article on accidental war. Kubrick read that article, tracked down George, and the two of them went to Cambridge to see Schelling. “Of course, the whole point of a Doomsday Machine is lost, if you keep it a secret!” is a Schelling lecture on commitment credibility, compressed in two minutes and delivered in Peter Sellers' voice.
Every wall in a negotiation was built by somebody, out of something, at a price. Cortés’s men thought they were looking at the sea.
Reversibility is pattern 6.3 of twenty-seven. The two axes, the nine categories and the full set are laid out in The Pattern Language.