Ensaio

How Much Is Proof Worth If We May Never Need to Use It?

How can we calculate the economic value of evidence before we know whether we will ever need to use it?

**Whats Neyven**

We are surrounded by evidence.

Contracts. Emails. Logs. Legal opinions. Minutes. Photographs. Recordings. Databases. Document versions. System records. Metadata. Artificial intelligence prompts and responses.

Organizations produce these records continuously. They document transactions, decisions, controls, communications, authorizations, processes, and events.

Most of them will never be used in litigation, an audit, an investigation, an inspection, or a dispute.

And yet, some may one day become decisive.

This creates an unusual economic problem:

**How much is it worth preserving today something that we may never need tomorrow?**

Before answering that question, however, an important conceptual distinction is necessary.

The title of this article refers to **proof**, but the economic problem begins before proof exists.

What organizations create and preserve are documents, data, records, logs, metadata, and other elements that may constitute **evidence**. Some will never acquire any evidentiary relevance. Others may later become useful in reconstructing, verifying, or demonstrating a fact and, depending on the context, may ultimately form part of the proof supporting a conclusion.

In other words:

**not every piece of evidence will become proof, but future proof often depends on evidence that had to be created and preserved long before anyone knew it would be needed.**

That uncertainty is precisely where the economic problem begins.

---

The dilemma is real

The intuitive response to information preservation usually oscillates between two extremes.

At one end: **keep everything**.

At the other: **keep only what we already know to be important**.

Neither approach is satisfactory.

Keeping everything creates costs. Storage, classification, cybersecurity, privacy, regulatory compliance, retrieval, interpretation, and information overload all carry economic consequences.

But eliminating what appears irrelevant is also risky.

When we eventually discover that a particular piece of evidence was necessary, it may already be impossible to reconstruct it.

A log may have been overwritten.

An email may have been deleted.

A database may have changed.

A model may have been updated.

A former employee may no longer be available.

The original context may have disappeared.

The question, therefore, is not simply whether to **keep or destroy**.

The real question is:

**How do we decide rationally what deserves to be preserved before we know what the future will require?**

This is where evidence governance meets economics.

And, more specifically, the **theory of options**.

---

Evidence as an option

In finance, an option gives its holder a right that may be exercised in the future if certain circumstances arise.

Its value does not lie only in what it produces today.

Part of its value lies in the **possibility it preserves for tomorrow**.

Evidence can be viewed in a similar way.

Consider a company that retains system logs documenting the approval of a transaction.

Today, those logs may appear insignificant.

Four years later, however, an investigation may ask who authorized a particular modification.

Those records may then allow the organization to reconstruct:

  • who performed the transaction;
  • when it occurred;
  • which system version was active;
  • what permissions existed;
  • which controls were executed;
  • whether there was subsequent human intervention.

The economically relevant point is that **the potential value of those records existed before the investigation began**.

What was unknown was whether the event capable of revealing that value would ever occur.

Preserving evidence can therefore resemble preserving an option.

It maintains the possibility of using that evidence if a relevant future scenario materializes.

Destroying it may extinguish that possibility.

And the decision may be irreversible.

**To preserve is to keep alternatives open.**

---

The cost of evidence is not its value

This distinction is fundamental.

An organization may spend very little to store a particular record for several years.

That does not mean the record is worth very little.

Likewise, an organization may spend considerable resources storing vast amounts of information whose future value is negligible.

Therefore:

**preservation cost is not the same as the value of evidence.**

The economic value depends on what the evidence may enable the organization to do if a particular state of the world materializes.

A simple first approximation might be expressed as:

**EV = p × I**

where:

**EV** = expected value of the evidence; **p** = probability that the evidence will be needed; **I** = economic impact avoided or benefit obtained if the evidence is available.

Suppose a particular record has only a 2% probability of ever being needed.

At first glance, preserving something with a 98% probability of never being used may appear irrational.

But suppose the absence of that record could contribute to a loss of $10 million.

Then:

**EV = 0.02 × $10 million = $200,000.**

If proper preservation costs $5,000, the decision looks very different.

Even evidence with a low probability of future use may therefore justify preservation when the consequences of its absence are sufficiently significant.

But the problem is more complex than this equation suggests.

---

The value of evidence goes beyond money

Evidence does not create value only when it helps win a lawsuit or avoid a financial loss.

It may also:

  • preserve rights;
  • reduce uncertainty;
  • support audits and investigations;
  • enable accountability;
  • preserve institutional knowledge;
  • reconstruct decision-making processes;
  • reduce forensic and remediation costs;
  • improve future controls;
  • support organizational learning;
  • make automated decisions auditable.

This means that evidence has at least three distinct dimensions of value.

Operational and informational value

Evidence may help an organization understand what happened, reconstruct decisions, identify failures, compare versions, preserve institutional memory, and improve future systems.

Evidentiary value

Evidence may later become relevant to demonstrating facts, authorizations, compliance with obligations, execution of controls, consent, ownership, or other circumstances that matter in legal, administrative, regulatory, or audit contexts.

Option value

Even when future use is uncertain, preserving evidence keeps open the possibility of using it if a relevant scenario materializes.

A more realistic economic model could therefore be represented as:

**EPV = Σ(pᵢ × Bᵢ) − PC − RR**

where:

**EPV** = expected preservation value; **pᵢ** = probability of each relevant future scenario; **Bᵢ** = benefit generated by having the evidence available in that scenario; **PC** = preservation cost; **RR** = risks generated by retention itself.

The last variable is essential.

Because **keeping evidence also creates risks**.

---

Keeping information also has a cost

Preservation is not automatically virtuous.

Retaining information can create:

  • exposure of personal data;
  • cybersecurity vulnerabilities;
  • regulatory risk;
  • storage costs;
  • obsolete or redundant information;
  • difficulty in search and retrieval;
  • decontextualized interpretations;
  • unnecessary discovery or investigation burdens.

Good governance therefore does not mean accumulating information indefinitely.

It means preserving information **with purpose**.

The relevant question becomes:

**For how long, at what cost, with what level of integrity, security, granularity, accessibility, and context should each class of evidence be preserved?**

This is a much more sophisticated question than simply deciding whether to delete a file.

---

The economic value of irreversibility

There is another variable that deserves special attention: **irreversibility**.

Some decisions can be corrected later.

Others cannot.

If an organization preserves a document for another year and later concludes that retention is no longer justified, it may eliminate it, subject to applicable legal and regulatory requirements.

But if the organization destroys it today and discovers tomorrow that it was needed, reconstruction may be impossible.

This asymmetry has economic value.

Preservation keeps alternatives open.

Destruction closes one of them.

This is precisely why **real options theory** provides such a useful analogy.

Under uncertainty, waiting can itself have value.

Not because waiting is always preferable, but because **postponing an irreversible decision may allow new information to emerge before a possibility is permanently eliminated**.

The question therefore changes.

Instead of asking:

**How much is this record worth today?**

we ask:

**How much is it worth preserving our ability to decide tomorrow whether this evidence still matters?**

Those are profoundly different questions.

---

Evidence also has informational value

There is another economic lens through which the issue can be examined: the **value of information**.

Imagine an audit involving three competing hypotheses.

Without certain records, none can be adequately confirmed or ruled out.

With those records, one hypothesis can be eliminated.

The evidence did not directly generate revenue.

It did something economically important:

**it reduced uncertainty.**

In decision theory, information has value when it can alter a decision and improve its expected outcome.

This means that the most valuable evidence is not necessarily the document that, by itself, proves everything.

Sometimes the most valuable evidence is the one that allows us to **distinguish among competing explanations**.

A small authentication log may be more valuable than thousands of pages of documents.

A metadata record may matter more than the file itself.

An earlier version of a document may, in a particular context, be more informative than the final version.

The economics of evidence is therefore not an economics of volume.

It is also an economics of **discriminatory capacity**.

---

The role of Law and auditing

Law has always lived with this problem, even when it does not describe it in economic terms.

Documents are created before litigation exists.

Contracts are signed when no one expects their interpretation to be disputed.

Minutes are written before anyone challenges a decision.

Logs exist before fraud is suspected.

Controls are documented before an audit begins.

Something that appears bureaucratic when created may become highly significant years later.

But one distinction is essential.

The mere existence of a document does not guarantee strong evidentiary capacity.

Between **existing** and **being capable of adequately demonstrating something**, several conditions intervene:

  • authenticity;
  • integrity;
  • reliability;
  • contextualization;
  • traceability;
  • accessibility;
  • intelligibility;
  • connection with the fact to be demonstrated.

This is why **evidence and proof are not synonymous**.

Evidence is an element capable of supporting the reconstruction, verification, or demonstration of a fact.

Its **evidentiary capacity** concerns its potential contribution to an inference or demonstration.

Its eventual use and assessment as **proof** depend on the context in which it is presented, the applicable rules, and its relationship with the other available elements.

An organization may therefore possess enormous amounts of information and still have very little ability to demonstrate what actually happened.

Auditors know this problem particularly well.

Sometimes the most important finding is not contained in the document that exists.

It lies in the document that **should exist but does not**.

Who approved the decision?

On what basis?

Which data were available at the time?

Which version was considered?

Who changed the parameter?

When?

Why?

The absence of evidence can transform an operational weakness into legal, financial, reputational, or governance risk.

**Without adequate evidence, even sound decisions may become difficult to defend.**

---

Evidentiary debt

This leads to another useful concept: **evidentiary debt**.

Software systems can accumulate technical debt.

Organizations can accumulate evidentiary debt.

Decisions made without adequate records.

Controls performed without sufficient documentation.

Models used without versioning.

Authorizations without traceability.

Processes that cannot reconstruct their own history.

For some time, this debt may remain invisible.

Nothing happens.

No one asks.

No dispute arises.

No audit reaches the issue.

Then, one day, someone asks:

**How do we know this is what happened?**

At that moment, the debt becomes visible.

And its cost may be very high.

---

Artificial intelligence makes evidence even more important

Artificial intelligence substantially expands this problem.

Consider a business decision supported by an AI system.

Months or years later, someone asks:

**Why did the system produce that recommendation?**

To reconstruct the decision, it may be necessary to know:

  • which model was used;
  • which version;
  • which parameters;
  • which prompt;
  • which context;
  • which documents were retrieved;
  • which input data were supplied;
  • which output was generated;
  • whether a human intervened;
  • which changes were subsequently made;
  • which approval criteria were applied;
  • which policies were in force at the time.

Without this evidentiary infrastructure, the organization may know the final outcome while being unable to reconstruct adequately **the process that produced it**.

That is why AI governance and evidence governance increasingly converge.

Logs are not merely technical telemetry.

Prompts are not necessarily transient commands.

Version histories are not merely technical conveniences.

Depending on the materiality of the decision, these records may become relevant to auditing, accountability, investigation, technical reconstruction, regulatory review, and eventually evidentiary use.

The consequence is significant:

**AI architecture must also be designed as evidentiary architecture.**

Reliable AI depends on reliable evidence.

---

Not all evidence is equal

The conclusion cannot be: **keep everything**.

That would simply replace one governance failure with another.

Different classes of evidence should be assessed according to factors such as:

  • probability of future use;
  • impact of absence;
  • possibility of later reconstruction;
  • preservation cost;
  • retrieval and interpretation cost;
  • obsolescence;
  • sensitivity and security risks;
  • legal and regulatory requirements;
  • audit and accountability value;
  • informational value for future decisions;
  • potential evidentiary capacity.

Reconstructibility is particularly important.

The more difficult something will be to reconstruct later, the greater the potential value of preserving it now.

A report might be reissued.

A database might perhaps be reconstructed.

But the exact state of a system at a specific moment may never be reproducible.

Some evidence can only be created once.

---

From retention schedules to portfolios of options

Organizations traditionally manage records through classification systems and retention schedules.

Those instruments remain essential.

But an additional economic layer can be added.

For each category of evidence, we might ask:

**What future possibility are we preserving by keeping this?**

This changes the logic of information governance.

Instead of treating retention merely as an administrative schedule, organizations can begin to see certain classes of evidence as a **portfolio of future options**.

Some evidence has high potential impact and low preservation cost.

Some is easily reconstructible.

Some is impossible to reproduce.

Some creates substantial privacy or security risk.

Some is legally required to be retained regardless of economic calculation.

This last point is fundamental:

**mandatory legal and regulatory retention requirements are constraints on the model, not variables to be negotiated through cost-benefit analysis.**

Economic reasoning applies primarily where legitimate discretion exists.

---

Perhaps we are measuring the wrong thing

Organizations generally know how much they spend storing information.

But how many know how much they lose because, at the decisive moment, they cannot demonstrate something?

That cost often appears later and elsewhere:

  • in judgments and settlements;
  • in prolonged investigations;
  • in audit and forensic hours;
  • in rework;
  • in reputational crises;
  • in controls that must be reconstructed;
  • in lost institutional knowledge;
  • in missed opportunities;
  • in decisions that cannot be adequately explained;
  • in AI systems whose operation cannot be reconstructed.

Storage costs are visible in the budget.

The cost of evidentiary absence is dispersed.

Perhaps that is why organizations systematically underestimate it.

**The absence of evidence also has a price.**

---

From information to decision infrastructure

The deeper point is that evidence should not be understood merely as archived material.

In many situations, it constitutes **decision infrastructure**.

Evidence allows an organization to reconstruct the past in order to act in the present.

It allows a court to assess a disputed fact.

It allows an auditor to test a control.

It allows management to understand a failure.

It allows an investigator to compare hypotheses.

It allows a regulator to verify compliance.

It allows an AI governance team to reconstruct an automated process.

It allows an organization to explain not only **what** it did, but **why and how** it did it.

That is more than storage.

It is institutional memory with operational, legal, informational, and economic value.

---

The invisible asset

Perhaps we need to rethink the way we perceive archives.

Some records are not bureaucratic residue from past activity.

They are **contingent assets of defense, reconstruction, learning, accountability, and decision-making**.

Their value remains invisible while nothing happens.

That is precisely what makes them difficult to justify economically.

An insurance policy may appear useless during every year in which no loss occurs.

An option may expire without ever being exercised.

Evidence

Whats Neyven

Readers’ Circle

Books continue when they find readers.

Receive selected releases, new articles and projects. Choose your interests and unsubscribe at any time.

I would like news about