The discount rate on the future

Ask anyone in an organization whether decisions should be documented well enough to defend later, and they will say yes. Ask the same people, at the moment a decision is actually being made, to stop and build the record that would make it defensible, and they will — reasonably, calmly, almost every time — decline. The gap between the stated preference and the revealed one is not hypocrisy. It is arithmetic. The cost of keeping the account is paid now, in full, in visible effort. The benefit arrives later, if it arrives at all, in a form no one can quite picture. Put those two on the same balance and the near cost wins, because that is what a discount rate does: it shrinks the future until the present can afford to ignore it.

This is, I think, the real reason accountability is chronically underprovided — not villainy, not laziness, but temporal discounting doing exactly what it always does. We treat the failure to keep records as a lapse of will, and so we reach for the tools you use on lapses of will: exhortation, policy, a memo about the importance of documentation. None of it works for long, because none of it touches the rate. The problem was never that people fail to value defensibility. It is that they value it later, and discount later steeply enough that its price today looks like a bad trade.

The trade every organization gets wrong

Consider the trade in its bare form. To build a real decision record — the evidence actually consulted, the rules in force at the time, enough state to reconstruct how the conclusion was reached — costs something definite and immediate. Time, attention, a little friction in a process everyone wants to be over. That cost is certain: you pay it whether or not the decision is ever questioned. The benefit is the mirror image. Being able to defend the decision, reconstruct it, or contest it is worth a great deal — but only in the branch of the future where someone asks, and most branches are quiet. The benefit is deferred, uncertain, and, in the overwhelming majority of individual cases, never actually collected.

So the decider faces a certain small cost now against a large but improbable and distant benefit. Under any ordinary discount rate — let alone the short-termist rate most institutions actually run — that trade comes out against the record. And here is the part that makes it insidious: each skip is locally rational. No single omission is the reckless one. Every time, the odds that this particular decision is the one that gets litigated, audited, or reversed are genuinely low, and the cost of preparing for it is genuinely real. A careful, self-interested actor, weighing this one choice on its own merits, should skip the record. The trouble is only visible in aggregate: a thousand individually defensible skips compound into an organization with no capacity to defend itself, one that discovers its blindness precisely when it can least afford to.

This structure is not new, and it is not peculiar to records. It is the exact shape of deferred maintenance — the bridge that is cheaper to not inspect every year until the year it isn't. It is the shape of skipped safety spending, where the expected cost of the accident is discounted below the certain cost of preventing it, right up until the accident. It is the shape of technical debt, where the interest is real but unbilled and the principal comes due all at once. In each case the near cost is concrete and the far cost is a probability distribution, and in each case a normal discount rate quietly recommends the neglect. Accountability belongs to this family. It is infrastructure for a future contingency, and infrastructure for contingencies is the thing discounting is worst at pricing.

The rate is the hidden variable

Once you see it as a discounting problem, the interesting quantity is the rate itself — and the rate is doing more work than a textbook would predict. Human and institutional discounting is not smooth; it is hyperbolic. We do not apply a steady annual haircut to the future so much as collapse it: the near term is vivid and the far term is flat, and the drop between them is severe and happens fast. A cost due today and the same cost due next quarter feel worlds apart; a cost due in three years and one due in five feel identical, which is to say both feel like nothing. Against that curve, any benefit that lives in the deferred, uncertain register — which is precisely where defensibility lives — is discounted almost to zero at the instant of choice. Not because it is worth little. Because it is far, and far is where our judgment goes blind.

No one chooses to be indefensible later; they simply discount later so steeply that defensibility is never worth its price today.

The deeper point is that the rate is not a fact of nature. It is a choice — usually an unexamined one. Every institution runs an implicit discount rate on its own accountability, set by its incentives, its horizons, whose bonus depends on this quarter and whose reputation depends on the next decade. That rate determines, more than any policy document, how much of a record actually gets kept. And it is almost never chosen deliberately. It is inherited from the compensation structure, the reporting cadence, the tenure of the people making the calls. Which means the single most consequential governance decision about accountability — how heavily to discount the future in which you must answer for today — is typically made by default, and made in the wrong direction, by nobody in particular.

This is also why the externality and the option persist, if you have read the other entries in this family. The reason the cost of a missing account can be shoved onto the affected party and onto the future is that the future is discounted; a cost you have discounted to nothing is a cost you feel entitled to externalize. And the reason the option value of a record gets sold so cheap is the same: an option that only pays in a deferred, uncertain state is worth almost nothing to a decider running a steep rate, so of course they let it lapse. Discounting is the engine underneath both. It is why the externality is tolerable and why the option looks overpriced at the moment you could still buy it.

Changing the rate, not the sermon

If the diagnosis is discounting, the cure is not a better sermon about responsibility. You cannot lecture a discount rate down. What you can do is move the cost — pull the deferred, uncertain penalty forward until it is felt at the moment of decision, where it can actually enter the arithmetic. Make the far cost near, and the trade comes out right on its own, without anyone having to be more virtuous than they were yesterday.

There are only a few ways to do this, and they are all versions of the same move. Attach liability that fixes now rather than someday, so the expected penalty stops being a distant probability and becomes a present-tense fact on the books. Require the record before the action rather than after the trouble, so keeping the account is not a bet against a quiet future but a precondition of proceeding — the cost paid up front by design instead of discounted away by default. Set a standard that prices the omission, so the absence of an account is itself the expensive thing, felt at the moment of choice rather than years later in a proceeding. Each of these is a way of raising the near cost of skipping until it exceeds the near cost of keeping. None of them asks the decider to care more about the future. They change what the present costs, and let self-interest do the rest.

That is the whole of it, and it is oddly hopeful. The chronic shortfall of accountability has looked, for a long time, like a moral failing that resists every attempt to shame it away — and it has resisted precisely because it was never moral. It was arithmetic run at the wrong rate. You do not fix arithmetic by disapproving of it. You fix it by changing the terms until the sum comes out the way you needed it to all along. The discount rate an institution applies to its own defensibility is a governance choice. The only real question is whether it will keep making that choice by accident, in the direction that leaves it blind, or start making it on purpose.

— Dispatches · Summit Cognitive