The information was correct.
Then the world changed.
Information half-life is the practical period over which an observation remains sufficiently representative of the world state for a particular decision or task.
This is the first pillar beneath How Information Works. The master owns latency and timeliness broadly. This page owns the deeper freshness question: when should an accurate record stop being treated as current?
Quick Read
Information does not become stale because time passes in the abstract. It becomes stale when the underlying process changes faster than the receiver can safely assume stability. A historical birth date has a near-infinite validity window. A traffic estimate may decay within minutes. A school timetable may remain useful for months but become wrong after one announced change. Freshness therefore depends on the volatility of the world state, the stakes of the decision, the cost of refreshing, and how reversible the action is. Strong information systems store both the value and the observation time, define expected refresh intervals, mark expiry where useful and distinguish “last known” from “currently verified.”
observation → timestamp → expected world volatility → validity window → refresh trigger → current / stale / unknown state → decision
Accuracy Has a Time Coordinate
A statement can be true at 9:00 and false at 9:10.
Examples include:
- traffic conditions;
- stock availability;
- queue length;
- weather;
- service status;
- live schedules.
When time matters, “What is the value?” is incomplete. Ask:
Observed when?
Different Information Classes Decay at Different Speeds
- Near-static: date of birth, historical event date.
- Slow-moving: curriculum structure, building address, long-lived policy architecture.
- Moderate: staff roles, class schedules, published prices.
- Fast: inventory, traffic, outages, live availability.
- Event-driven: information remains valid until a specific change occurs.
One universal refresh rule therefore wastes effort or creates risk.
Staleness Is Receiver- and Decision-Specific
A one-year-old population figure may be adequate for a classroom explanation.
The same figure may be too old for a current resource-allocation decision.
The half-life belongs to the relationship among information, receiver and job—not to the record alone.
High Stakes Shrink the Acceptable Validity Window
Low-stakes plan:
yesterday’s information may be adequate.
High-stakes irreversible action:
refresh closer to the decision point.
This connects to How Risk Works, which owns consequence and exposure broadly.
“Last Updated” Is Better Than False Freshness
If the system cannot verify a current state, say:
Last verified 2 September, 4:30 p.m.
This preserves honesty about the evidence boundary.
Refresh Triggers Can Be Clock-Based or Event-Based
- Clock-based: update every hour/week/term.
- Event-based: update after policy change, outage, new exam cycle, role change.
- Query-based: refresh only when a decision requests current state.
- Anomaly-based: refresh when observed reality conflicts with stored state.
Stale Information Can Be More Dangerous Than Missing Information
Missing information may prompt caution.
Stale information often looks complete and authoritative.
The second sibling, Missing Information, owns what absence can and cannot tell us.
AI Increases the Need for Freshness Labels
An AI answer may combine current and older material without exposing the age of every component.
For current claims, ask for source dates, live verification or authoritative current records where consequence warrants it.
A Practical Freshness Packet
INFORMATION: Tuition timetable OBSERVED/VERIFIED: 3 Sep 2026 14:00 EXPECTED VOLATILITY: Moderate VALIDITY RULE: Current until timetable change notice HIGH-STAKES USE: Confirm again before travel DISPLAY STATE: CURRENT / LAST VERIFIED / STALE / UNKNOWN REFRESH TRIGGER: New term, teacher change, explicit notice
A 20-Lens Information Half-Life Audit
- What world state does the information represent?
- When was it observed?
- How quickly can the world state change?
- Is change continuous or event-driven?
- What decision uses the record?
- What are the stakes?
- How reversible is action?
- What is the refresh cost?
- What is the delay cost?
- Does the record show last verified time?
- Is freshness visible to the receiver?
- What marks expiry?
- Can anomalies trigger refresh?
- Does a newer source supersede it?
- Are multiple versions circulating?
- Does historical value remain even after current validity expires?
- Can stale data be mistaken for live state?
- What information class is this?
- What is the minimum currentness needed for the job?
- Did the eventual decision use information fresh enough for the world it acted on?
For Primary Readers
A bus schedule from last year may have been correct then and wrong now. Good information tells you not only what it says, but when it was checked.
For Secondary Readers
Compare a birth date, timetable, weather report and traffic estimate. Explain why each has a different information half-life.
For Advanced Readers
Model information freshness as a validity interval determined by process volatility, decision horizon and acceptable state-estimation error. Optimal refresh cadence balances observation cost against the expected consequence of acting on a diverged world model.
Final Thought: Correct Once Is Not the Same as Correct Now
Information stays useful only while its timestamp, the world’s rate of change and the receiver’s decision horizon still fit one another.
INFORMATION · FOUR PILLAR LEGS
Return to How Information Works, or continue through Missing Information, Provenance Breakpoints and Decision-Relevance Threshold. Return to the How X Works Hub.