Skip to content

Lindy Decision-Horizon Review

Decision review — instantiates Survival-Conditioned Persistence Forecasting

Turns a survival-conditioned forecast into a bounded, reviewable commitment horizon with exits kept open — and a record that longevity, not merit, drove the call.

A remaining-life forecast is not yet a decision, and left to itself it curdles into one of two bad ones: irreversible lock-in, or the belief that because a thing has lasted it must be good. Lindy Decision-Horizon Review is the step that spends the forecast responsibly. It converts "this is likely to persist for a while" into a bounded, reviewable commitment horizon — how much to rely on the incumbent, for how long, and under what conditions the reliance is revisited — sized to a conservative quantile rather than the headline estimate, hedged with kept-open exits, and stamped with an explicit note that the commitment rests on persistence, not on quality, legitimacy, or safety. Its defining discipline is that it never lets a long-life forecast become a permanent one; the horizon is always a checkpoint, and leaving is always still on the table.

Example

A bank runs its core ledger on COBOL systems more than forty years old. One faction wants to migrate immediately ("it's ancient"); another insists it will "outlive us all." The review takes the survival-conditioned forecast — COBOL is genuinely non-aging (a language does not wear out) and has a long, censoring-checked track record, so its expected remaining support horizon is long and heavy-tailed — and turns it into a governed commitment rather than a slogan. It sets a bounded horizon: rely on COBOL for the core ledger over a roughly seven-year reviewable window, chosen from a conservative lower quantile because the tail is uncertain, and fund maintenance and talent to match that window and no further.

Crucially it keeps the exit cheap: an abstraction layer around the ledger and a maintained migration-cost estimate, so the option to move stays live instead of atrophying. And it writes the decision-use trace: this horizon reflects how long the thing is likely to persist, explicitly not a claim that COBOL is technically superior; the contrary evidence that would dominate (a vendor end-of-support date, a hiring cliff for maintainers) is named, and the review is scheduled to fire on those triggers rather than on a calendar out of habit.

How it works

The review governs commitment; it does not compute the forecast. What distinguishes it:

  • Size to a quantile, not a point — the horizon is drawn from a conservative survival quantile or scenario band, scaled to reversibility and downside, so a fat but uncertain tail never justifies an unbounded bet.
  • Fund the exit — a replacement/rollback/migration option is kept genuinely open and paid for, converting an irreversible dependency into a reversible one.
  • Trace the basis — a written record firewalls longevity from merit, names the disconfirming evidence, and binds the review to a real trigger with an owner.

It consumes the remaining-life forecast and the eligibility verdict; it produces the decision and the machinery that keeps that decision honest over time.

Tuning parameters

  • Horizon quantile — which survival quantile the commitment is sized to. Lower quantiles buy safety but risk under-committing to genuinely durable infrastructure.
  • Reversibility spend — how much goes to keeping exits open (adapters, escrow, dual-run) versus sinking cost into the incumbent. More optionality costs now and pays off under regime change.
  • Review trigger — scheduled cadence versus event-driven (a vendor notice, a staffing metric). Event triggers catch regime shifts sooner but demand live monitoring.
  • Downside scaling — how sharply the horizon shortens as blast radius grows; high-dependency systems get shorter, more hedged commitments.
  • Merit-firewall strictness — how explicitly the trace forbids reading longevity as quality, legitimacy, or safety.

When it helps, and when it misleads

Its strength is that it heads off the two decision failures the archetype names outright: converting a persistence forecast into irreversible lock-in, and converting longevity into authority. By committing to a bounded, hedged, reviewable horizon it captures the real value of durable infrastructure while keeping the option to leave — and that flexibility to stop is itself worth paying for under uncertainty.[1]

Its failure modes are the ways governance goes slack. A horizon set from the optimistic tail becomes the lock-in it was meant to prevent; a review cadence that never actually fires is theater; and the whole thing is easily run backwards — the horizon ("keep it forever," or "rip it out now") chosen first and the forecast back-filled to justify it. There is also a subtler trap: a long commitment can be self-fulfilling — exclusive investment deepens switching costs until later survival is no longer independent evidence of durability. The discipline is to size to a conservative quantile, make the review trigger a real event with a named owner, and keep the optionality funded rather than nominal.

How it implements the components

This review fills the decision-and-accountability components — the ones that turn a forecast into a governed commitment:

  • planning_horizon_decision_rule — it sets the bounded, reviewable horizon and the level of commitment, sized to reversibility and downside rather than to a point forecast.
  • decision_use_trace — it records that persistence (not merit, legitimacy, or safety) drove the call, names the disconfirming evidence, and fixes the review trigger.
  • replacement_and_optionality_hedge — it keeps migration, rollback, and replacement options funded so the commitment stays reversible and cannot harden into lock-in.

It does not estimate remaining life or its uncertainty (Lifetime Distribution Comparison, Age-Conditioned Remaining-Life Table), decide whether the subject is eligible (Non-Aging Eligibility Review), or watch for the regime change that fires its trigger (Change-Point Detection) — it consumes those and turns them into an accountable decision.

Notes

Because the commitment can manufacture the very survival it forecasts — pour investment into the incumbent and switching costs rise, which prolongs its life for reasons unrelated to intrinsic durability — the decision-use trace must record that endogeneity. Otherwise a later "it survived, so we were right" is mistaken for independent confirmation, and the review quietly reproduces the survivorship reasoning the archetype exists to discipline.

References

[1] Real-options reasoning — under uncertainty, the flexibility to stop, switch, or defer has value in its own right, so a hedged, staged commitment can be worth more than a cheaper-looking irreversible one. It is why a horizon that looks marginal as a single lump can still be worth entering when the exit is kept open. ↩