The truth of a person’s capability is bounded by time and context. Neither gives an organization the authority to turn a past state into a permanent ceiling.

Publication Note

This article is part of the OSI-PIA public article series. It describes an active research and development direction. It does not claim that OSI-PIA is a finished hiring product, validated employment-screening system, legal compliance tool, or production decision system.

Article

A record can remain true about the past while becoming wrong as a description of the present. That distinction sounds obvious until we look at how institutions actually use information about people. A job title can follow someone for years after their work has outgrown it. An old performance evaluation can continue to shape assumptions after the conditions that produced it have changed. A credential can remain historically valid while becoming a weak indicator of current skill. A personality assessment can capture a person at one point in time and then quietly harden into an identity claim. A resume can preserve an accurate history while still badly underrepresenting current capability.

The problem is not simply stale data. The problem is authority being carried forward farther in time than the evidence supports. That is the problem of temporal authority.

Historical truth is not current truth

Most information systems are built to preserve records. Fewer are designed to reason carefully about the authority those records should retain as conditions change. A project-status document may accurately state that a system is unfinished. Six months later, the system may be running in production. The old document has not become false. It remains a valid historical record. What has changed is its authority to describe the present state. The same logic applies to human capability. A person may once have required close supervision. They may once have lacked a skill. They may once have been operating within a narrow role. They may once have performed poorly under a particular set of conditions.

Those observations may remain historically valid. They do not automatically remain valid descriptions of the person now. This gives us three distinct questions:

  1. Historical validity — Was the claim reasonably supported when it was recorded?
  2. Temporal transferability — Under what conditions may that claim legitimately be carried forward?
  3. Current validity — Is there sufficient evidence to use that claim as a description of the present state?

Conventional records often collapse all three into one. If the record exists, it is treated as if it still speaks with the same authority. That is a mistake.

Evidence has a jurisdiction in time

A useful way to think about this is that evidence has a kind of temporal jurisdiction. Some facts remain authoritative for a very long time. A degree conferred twenty years ago remains a degree conferred twenty years ago. Other claims decay much more quickly. A five-year-old evaluation of someone’s technical proficiency may no longer say much about what they can do today. A statement about a team’s morale may become obsolete within weeks. A project status can change within hours. A timestamp is therefore necessary but insufficient. Knowing when something was observed does not tell us:

  • how rapidly the underlying state can change;
  • whether newer evidence supersedes it;
  • whether the claim concerns a stable trait, temporary condition, completed event, prediction, or interpretation;
  • how much current authority should remain; or
  • whether the evidence is appropriate for the decision being made now.

Temporal authority is not a property of the date alone. It is a property of the relationship among evidence, claim, context, change, and use.

Do not let synthesis erase chronology

This matters even more in AI-mediated systems. Modern AI systems are exceptionally good at taking many pieces of information and turning them into a smooth, coherent account. That is useful. It is also dangerous when chronology disappears inside the synthesis. A system can ingest several historically accurate observations, combine them into one elegant summary, and accidentally produce a description that appears current even though the underlying evidence comes from different states, different contexts, and different periods of a person’s life. The failure mode looks like this:

Historically accurate observations
            ↓
Loss of chronology and context
            ↓
Coherent synthesis
            ↓
Apparent current description
            ↓
Decision or action

The answer is not to avoid synthesis. The answer is to preserve the temporal structure beneath it.

Observation
    ↓
Observed at time t
    ↓
State at time t
    ↓
Subsequent observations
    ↓
Change / continuity / contradiction
    ↓
Current confidence and authority

A responsible system should preserve not only what was known, but when it was known, what state it described, what has changed since, and how much authority the old evidence still deserves.

A career history is not a capability ceiling

This becomes especially important in employment. Conventional labor systems often treat occupational history as evidence of occupational destiny. If a person has held the same kind of role for ten years, the system frequently treats that history as evidence of what the person is qualified to continue doing. But:

“This person held Role X for ten years”

is evidence of employment history. It is not automatically evidence that:

“Role X represents this person’s current maximum capability.”

Those are different claims. Recent work on mid-career mobility makes the distinction concrete. The Burning Glass Institute and NYU School of Professional Studies report Sidetracked: The Hidden Crisis in Mid-Career Mobility examined large-scale career histories and describes how prolonged career stall can become visible before it hardens, including movement into adjacent roles that reuse existing human capital. That matters because a stalled career can produce a self-reinforcing record. A narrow role produces a narrow title. The narrow title limits what employers infer. The limited inference constrains opportunity.

The constrained opportunity produces less visible evidence of broader capability. The record then appears to confirm the original assumption. The history remains true. The inference becomes increasingly misleading. That is a temporal-authority failure.

Context matters too

Time is only half the problem. A capability observation is always made under conditions. A person may perform differently depending on autonomy, trust, information access, role clarity, supervision, tools, workload, team quality, incentives, psychological safety, developmental opportunity, organizational politics, and fit between role demands and actual capability. A past observation may therefore be bounded by context as well as time. This does not mean poor outcomes should be dismissed whenever the context was difficult. It means the observation should be interpreted at the level the evidence supports. “Performed poorly under these conditions” is a stronger statement than “is a poor performer.”

“Required close supervision in this role at this point in time” is more defensible than “requires close supervision.” “Had not yet demonstrated capability X” is different from “is incapable of X.” These distinctions matter because institutions routinely transform context-bound observations into context-free identities. That is not rigor. It is overreach.

Institutions may record history. They do not own the future.

Organizations need records. Managers need evaluations. Researchers need longitudinal data. Employers need to make decisions under uncertainty. Temporal authority is not an argument against institutional judgment. It is an argument for disciplined judgment. Institutions should be able to say:

We observed this.
We observed it under these conditions.
We observed it at this time.
This is how we interpreted it.
This is how confident we were.

What they should not be able to do without current evidence is quietly turn that observation into:

Therefore, this is what this person is.

Or:

Therefore, this is what this person will remain capable of.

Historical evidence is not an institutional license to impose a permanent ceiling.

Why this matters for PIA

PIA is being built around an explicit separation between observation and inference. That distinction becomes much more powerful when time is included. A capability record should be able to preserve:

Observation
    ↓
Source
    ↓
Artifact / evidence
    ↓
Context
    ↓
Time
    ↓
Inference
    ↓
Confidence
    ↓
Subsequent evidence
    ↓
Revised or sustained inference

The old inference need not be erased. It may remain an important part of the history. But later evidence should be able to narrow, expand, contradict, or supersede it. This is especially important for people whose conventional biographies understate their development. Someone may have learned rapidly. Their work may have expanded beyond their title. Their ability may have become visible only after gaining new tools, autonomy, or opportunity. The architecture should be able to represent that change instead of forcing the present person into the shape of the oldest surviving record.

Why this matters for OSI

Organizations change too. An employee once viewed as junior may now be operating at a much higher level. A team once considered dysfunctional may have changed leadership. A unit once dependent on external expertise may have developed strong internal capability. A manager once highly effective may be operating in a different environment with different results. A skill once scarce may now be abundant. If OSI is meant to understand organizations as dynamic systems, then it cannot treat historical states as permanent properties. It must treat state transitions as first-class evidence. That requires a basic discipline:

Preserve history without mistaking history for the present.

Governance consequence

Temporal authority should therefore become an explicit governance rule. Every consequential capability claim should be able to answer:

  • What is the underlying evidence?
  • When was it observed?
  • Under what conditions?
  • What exactly was inferred?
  • How stable is that inference expected to be?
  • What subsequent evidence exists?
  • Has the claim been reaffirmed, weakened, contradicted, or superseded?
  • Is the claim still appropriate for the decision being made now?

This is more demanding than simply storing timestamps. It is also more fair.

The governing principle

The principle can be stated simply:

Evidence should retain its history without inheriting unlimited authority over the future.

And for capability systems specifically:

The truth of a person’s capability is bounded by time and context. Neither gives an organization the authority to turn a past state into a permanent ceiling.

A person is not a frozen record. An organization is not a frozen state. A responsible architecture should know the difference.

Sources

  • Burning Glass Institute and NYU School of Professional Studies. Sidetracked: The Hidden Crisis in Mid-Career Mobility (2026).
  • PIA/OSI internal governance concepts: observation vs. interpretation, provenance, confidence, state transitions, contestability, and human-centered capability infrastructure.