The Reality Gap

Modern organizations generate extraordinary amounts of data about the physical world.

A shipment was received. A temperature was checked. A product was produced. A sanitation procedure was completed. A package was labeled. An employee performed an inspection. A delivery arrived.

Our systems record these events every day.

But there is a deceptively simple question behind every one of those records:

How do we know the physical event actually happened the way the digital system says it did?

That question exposes what we have come to think of as the Reality Gap: the distance between a digital record and the physical-world event that record is intended to represent.

A record is a representation of reality

Consider a temperature check.

A system may contain:

That is useful information.

But the database entry itself does not necessarily tell us whether someone actually looked at a thermometer at 10:42 a.m.; whether the measurement was entered later; whether the employee was physically present; whether the reading applied to the correct cooler; or what evidence existed when the entry was made.

This distinction becomes increasingly important as organizations replace paper processes with digital ones. Digitization can make records easier to create, retrieve, analyze and share. But digitizing a claim about reality does not, by itself, make the claim more reliable.

Sometimes it simply gives an unverified assertion a better user interface.

Traceability has already moved us part of the way

Supply-chain traceability provides a useful example.

The GS1 Global Traceability Standard organizes traceability around Critical Tracking Events (events such as receiving, transforming, packing and shipping) and Key Data Elements describing those events. GS1 describes traceability information through dimensions including who, what, where, when and why.^[1]

FDA’s Food Traceability Rule similarly requires covered organizations handling certain foods to maintain Key Data Elements associated with specified Critical Tracking Events. Transformation, receiving and shipping are among those events.^[2]

This is important progress.

We are becoming much better at describing physical events digitally.

But another question remains:

What evidence supports the assertion that the described event occurred?

That is a different problem.

The next layer is evidence

Imagine two records that look identical:

One might have been reconstructed hours later from memory.

The other might be associated, as it happened, with a supplier record, timestamp, identified user, photograph, temperature observation, receiving location and subsequent transformation history.

The database may display essentially the same receiving event.

The evidentiary quality of the two records is very different.

This suggests that the next generation of operational systems needs to do more than answer:

What does the record say happened?

It should increasingly help us answer:

What evidence exists that it happened?

This matters beyond food

Our own encounter with this problem came through food production, food safety and traceability.

But the underlying problem is much broader.

• A manufacturer says a required inspection occurred.
• A healthcare organization says a physical procedure was completed.
• A logistics provider says custody transferred at a particular location.
• A contractor says specified work was performed.
• A laboratory says a specimen followed a particular process.
• A maintenance organization says equipment was inspected.

In each case, a digital record can document the assertion.

The Reality Gap concerns the relationship between that assertion and the physical event itself.

AI makes the question more important

There is another reason this matters now.

Artificial intelligence is becoming extraordinarily capable at reasoning over digital information.

But AI generally does not encounter the physical world directly. It encounters representations of it: records, images, sensor readings, transactions, forms, databases and other digital evidence.

That creates an upstream question that better reasoning alone cannot solve:

How trustworthy is the evidence from which the machine is reasoning?

If the digital representation of a physical event is unreliable, increasingly sophisticated analysis can still produce conclusions from a weak representation of reality.

Before asking machines to reason about the physical world, we may need better ways to establish the evidentiary connection between physical events and digital records.

Closing the Reality Gap

We don’t believe the answer is simply collecting more data.

The answer is creating better relationships between events, evidence and records.

That means thinking differently about operational software.

Not simply:

Did somebody enter the required field?

But:

1. What happened?
2. What evidence supports it?
3. When was that evidence created?
4. What physical object or process did it concern?
5. Can another person later verify the relationship?

The digital transformation of physical operations has largely focused on turning activities into data.

The next challenge may be harder, and more important:

Turning data into trustworthy evidence of what happened in the physical world.

That is the Reality Gap.

And closing it is the problem Shrink Verify exists to explore.